Reflections on an Essay by Claude Lévi-Strauss

Anthropology is a science of extremes: from the personal indignities of fieldwork to the crystal ball of cultural comparison, from grand theories of human evolution to the fragmentation of meaning itself in postmodernism. Close to the far end of one of the many dichotomies with which our discipline is spiked, we find Claude Lévi-Strauss, an immaculate Frenchman whose influence rivals Durkheim's. Somewhat further removed from the opposite extreme stands a sensibly attired group of British structural functionalists. The balance of power is uneasy. Followers of Radcliffe-Brown have practicality and applicability on their side: Fredrik Barth's ideas, for example, are of use in social planning, and both he and Radcliffe-Brown insist on traditional empirical falsification procedures for sociological hypotheses. Lévi-Strauss, in contrast, works inductively: by constructing a theory to bring out certain aspects of reality and illuminating these sharply, he shows their complex texture. What attracts us in Lévi-Strauss is the fine-honed, exact reasoning of his best work. What repels many anthropologists is embodied in the famous opening quip of Tristes Tropiques (1955): "I hate travelling and explorers". Not to be concerned with fieldwork is one of the mortal sins in anthropology. To many, such as Clifford Geertz (1967), Lévi-Strauss will therefore always remain the "cerebral savage".

A cornerstone of Lévi-Strauss' work is the notion that all social phenomena are reducible to a simple dualistic scheme, reminiscent of dialectics on the one hand, cybernetic "on-off" switches on the other. There is nothing new in dualism, neither in anthropology nor elsewhere. The cosmology of Tsembaga Maring (slash-and-burn horticulturalists in Papua New Guinea) opposes "high" and "low" spirits, which are also "hot" and "cold", "red" and "dark", warlike (male) and fertile (female) (cf. Rappaport 1968). In ancient Chinese cosmology masculine, light, creative forces are opposed to dark, feminine, receptive presence (Wilhelm 1943). The Manichaeans split reality into dark and light, evil and good, a dualism that Christianity adapted in its doctrine of salvation and sin, but with modifications important enough for Augustine to condemn the Manichaeans roundly.

In these cases, however, the dualistically opposed entities embody moral qualities. Such moral attribution is far from Lévi-Strauss' aim. "Binary opposition" is a strictly analytical notion, a means of "taking the world apart and putting it together again", but its amorality is further strengthened by the fact that Lévi-Strauss is concerned with a microcosmos that most of us do not even know is there. Duality is simply the lense through which understanding is made to focus on this level.

An famous example is the so-called "culinary triangle" (see Leach 1970). Food is not simply food for Lévi-Strauss, it is meaningful.⁽¹⁾ And foods acquire meaning by being contrasted with other kinds of food. Cooked food is opposed to raw food - cooked food is "cultural", raw food "natural". The cultural is "masculine"; the natural "feminine", etc. Opposition is grafted onto opposition. But the point is neither the food itself nor that culinary contrasts are based e.g. on sexual contrasts. The idea is rather that all categories, and hence all meanings, are products of opposition. Neither the cooked nor the raw, neither culture nor nature are meaningful in themselves. They seem so because we define them as opposites. Meaning is not immanent in objects or subjects but in their relationships to each other. This has already brought us a step closer to what I consider the essential message in Lévi-Strauss' work: An opposition is not simply an abstract and static contrast, but a relationship - a dynamic, concrete force. It is creative. Out of the distinction between raw and cooked springs the idea of something intermediate, "neither-nor", "betwixt and between", something that is on its way to become either culture or nature but has not yet gotten so far. Fried food is "almost cooked". Rotten food is "almost de-cultured". Thus the binary relationship extends itself into one or more triangles. And from this simple mechanism Lévi-Strauss assumes that we can derive the entire cultural order, in all its existential complexity.

What relevance can such these abstractions have for the imponderables of anthropology? To give a fair answer we must place Lévi-Strauss in some kind of wider context. I shall therefore start out by discussing two theoretical constructs used by other anthropologists, and try to relate them to Lévi-Strauss: (1) The proposal that meaning is a product of context, and (2) Gregory Bateson's idea of dual description.

Context

When we say that a thing (an "element") has no meaning or use out of context, the point is simply that it can have no relevance for other elements in a context (e.g. for people in a society) if it does not stand in some kind of relationship to them. If we think of the context as a kind of map, the element's meaning may be envisioned as its position relative to other points on the map. The context may then be drawn as a network of connections between elements, as in figure 1.

Figure 1

This teaches us several important things:

a) To be part of a context an element must have relationships to other elements in the context (a Norwegian is only Norwegian as long as he does something Norwegian, i.e. relates to some element in the context of "Norwegianness"). This implies some kind of active interaction or exchange between elements. An element's meaning, role, purpose etc. may therefore be thought of as a product of all its exchanges with other elements in the context in which it participates.

b) But if an element is to have a separate identity, it must not only relate to other elements, but mark itself out as different from them. A relationship is simultaneously a confrontation, an opposition; connection is inconceivable without separation. A famous example of this seeming contradiction is the debate on the nature of ethnicity. It has been proposed that a group's consciousness of itself as an ethnic entity results from its interaction with other groups (Barth 1969). Thus the identity which to participants seems uniquely "theirs" is in fact a reflection of their opposition to "others".

Clearly, the simplest form such an "opposition-relation" can take is dyadic - a context with less than two elements is a contradiction in terms. So if we assume that meaning can only arise in a context, it seems clear that binary opposition (which is also binary relationship) must be if not the basis of all meaning then at least its most elementary form - the "atom of significance".

In Mind and Nature (1979) Bateson posits that our perception of reality is based on an oscillation between two elementary principles - "number" and "quantity". Quantities are analogue, continuous entities. Numbers are discontinuous and digital. The two oranges on the table in front of me are not only more concrete than the statement "two oranges". The oranges and their name belong to antithetical worlds. To describe the first of these worlds we start with the various qualities of the oranges (e.g. weight, color) and measure "how much" they have of each ("how heavy"). Since ascertaining this value exactly is impossible, it is also impossible to duplicate real oranges or give a complete description of them. No two identical quantities exist. But there are as many identical numbers as you like. If I put two other oranges on the table, with completely different qualities (bigger, rotten, new colors and varieties), I still have two oranges.

The reason is simply that quantity measures a thing's properties, its "content", while number describes its "form". Put differently, quantity is what the element itself is or has, while number is its position in a context - the "name" of a relationship: When we count "two oranges", we are saying nothing about the oranges themselves, but pointing to a context with a certain number of relationships - between oranges, or anything else for that matter. Thus a number is an opposition and a relationship between quantities. It is a "name" - not the thing itself - and as such it may be regarded as the prototype of all categories and all meaning. We cannot speak of things, be influenced by them or change them without "counting" them - i.e. abstracting from all their unique qualities and viewing them as pure contrast.

But Bateson brings us a step further. We must contrast "quantities" to produce "numbers". Binary oppositions are therefore the "elementary particles" out of which all human meaning is built. It is through contrast that things become things. But such "social particle physics" might turn out to be relevant not only on the micro level but in our dealings with complex aggregates⁽²⁾ as well - i.e. with real people, things, symbols. As Bateson points out, "two descriptions are better than one" no matter what level of complexity we choose. A simple example is binocular vision: With two eyes I see not only more (a different "quantity") than with one. I accomplish a qualitative leap - acquire depth vision (increase the "number" of tangible dimensions). Another example, to which we shall return, is the Rorschach figure of psychologists, produced by spilling ink on a sheet of paper - which is then folded. As any child knows the result is an intricate pattern - order arisen out of Chaos. We have created form - and all we did was fold the paper - make a "binary opposition".

Not only do things arise from contrast. Things themselves can be contrasted to create more complex things. Starting with empty dualities, we can step from level to level, folding and refolding the paper, till we end up with something as complex and intriguing as the world in which we live.

So the idea that meaning is a product of binary opposition is not unique to Lévi-Strauss, nor can its relevance be restricted to the most abstract and elementary dimensions of the human condition. But what are the consequences of thinking about society in this way? Let us take a closer look at one of Lévi-Strauss' own analyses to see what we can learn from it.

"Do Dual Organizations Exist?"

This is the title of one of Lévi-Strauss' more abstruse essays (1958). It starts out with an old debate about so-called "moiety-organized" societies. These are divided into two exogamous halves, which are structurally similar, and (usually) "marry each other". Lévi-Strauss asks if we are not mistaken in calling these societies "dual", since a closer inspection shows that the division into equal halves is only one of many organizational principles coexisting in such groups. He then goes on to delineate three basic forms of division typical of "dual" societies:

a) Diametric duality, which is found when a group is divided into two halves with equal status.

b) Concentric duality, in which society is split into a "central" and a "peripheral" part, and the center has higher status than the periphery (e.g. a central men's house with women's houses in a circle around it).

c) Triads, which are divisions into more than two parts. Lévi-Strauss disposes of all forms with more than three parts by saying that they can either be split into equal halves (8 = 4 + 4) - or one can isolate a single part and split the rest in two, leaving a triad (17 = 1 + 8 + 8). Strange as it may seem, this is not mere sophistry.

Next, he posits that the three types are not simply different, but consecutive stages in a process: Diametric duality is transformed into concentric duality when the line of separation between the halves (the "diameter") is "abstracted", "lifted up on a higher level", where it no longer functions as a dividing line, but as a point of orientation (a "center"). Triads arise when both the center and the two halves are "objectivized" - treated as stable, separate entities with a "life of their own". (See Figure 2.)

Figure 2

Once again this may sound rather too much like an intellectual game - but in fact a very subtle point is being made. Lévi-Strauss is describing how cultural diversity and complexity arise out of "elementary particles" of meaning - binary oppositions. Diametric duality is a simple, binary contrast. Dividing a village into equal, but opposed halves is very similar to folding an ink-smeared sheet of paper to create a Rorschach figure. It is tempting to agree that this is the simplest imaginable form of social organization. Perhaps a society is unable to see itself as a society at all if it cannot establish some kind of elementary opposition-and-relationship within itself first. It is striking that empirical diametric dualities (e.g. moieties) actually seem to have no other function than to create a difference - an opposition to which a "name" - meaning - may be attached. Meaning, as Bateson puts it, is a "difference that makes a difference" (1970:453).

What follows becomes more understandable in this light. When "the diameter becomes the center" this seems like a piece of intellectual slight-of-hand. What it means, however, is that the relationship between two elements has acquired a "name". The diameter itself is opposition and relationship, nothing more. It is a "relationship between two quantities", a mere "declaration of difference". The center, on the other hand, is the name of this relationship - a "number" - and thus potentially a concept to describe reality and an instrument with which to change it. It articulates the opposition, letting it play itself out in a dynamic process of abstraction.

In the final step, this process is brought to its conclusion. Strictly speaking, the end product is not a triad in the literal sense, but any number of units (5, 9 or 576), as long as they are treated as autonomous elements with a "life of their own" - with a "content" that is independent of the dualistic contrasts out of which they sprang. To understand this, one may simply continue hitching triad onto triad indefinitely, while letting parts of some triads "erode away" (this is possible, since the elements are no longer absolutely dependent on each other). The end result will be a diagram identical with Figure 1. By means of the Lévi-Straussian process we have been brought from elementary binary oppositions to a full context. What Lévi-Strauss is describing is thus a generative model of cultural complexity. The transition from Figure 2 to Figure 3 depicts a transformation from an underlying process to the overt results of this process - or with Lévi-Strauss' own words: from deep structure to surface structure.

Figure 3

We have now assigned a place for Lévi-Strauss in the wider landscape of social science. We understand the purpose behind his abstractions: the creation of a model to generate social complexity. This is fair enough, but it is still difficult to discern any kind of practical use for the theory. Is not such a "social particle physics" useless in the real world, where, as we know, quarks and leptons rarely play an independent role? Lévi-Strauss seems like a monomaniac quantum physicist, who insists that architects and engineers should decide whether electrons are particles or waves when designing buildings and dams. Don't we, like these practical physicists, need practical social theories, with direct relevance for the aggregated units we encounter in daily life?

This is a fundamental objection, which in fact applies not only to Lévi-Strauss, but to much of the theory utilized by anthropologists. I think we can state categorically that the Lévi-Straussian model cannot and should not be used directly in the study of social life. The fact that this is still done (by the master himself as well as by his disciples) is a result of the confusion reigning in the social sciences as to the relationship between element and aggregate.

Let us return to an example we have already broached. The analysis of ethnic identity as (nothing but) a result of binary opposition (i.e. boundary maintenance) is clearly an over-simplification. The contrast between "us" and "them" may well be a necessary condition for defining our identity as different from that of an "other", but the substance of our difference, the many concrete and detailed ways in which we perceive ourselves as different from the "other" cannot be reduced to such simple terms. When a Norwegian, for example, perceives him- or herself as different from a Swede, much more than an abstract construction of opposites is involved. According to the popular stereotype, Swedes "are" somewhat stiff, somewhat arrogant, somewhat more sophisticated than Norwegians, who see themselves are more informal, folksy and direct. Swedes are thus opposed to Norwegians in a certain way. Danes on the other hand are opposed in completely different ways: stereotypical Danes "are" more informal, but also more sophisticated, more relaxed, but less direct. Swedes and Danes thus contrast with Norwegians in different ways, and Norwegian ethnic identity is defined by a complex combination of these and other types of contrasts. Sweden and Denmark, moreover, are Norway's traditional neighbors, and are stereotypically portrayed by Norwegians in considerable detail. Frenchmen, Americans, Pakistanis or Nigerians are progressively more distant groups, with which Norwegians have less experience, and the stereotypical "oppositions" by which Norwegians define their relationship to them are less elaborate and more unstable. Nevertheless, these distinctions also contribute to Norwegians' perception of their ethnic identity. Thus, the simple, binary contrast of "us" versus "them" resolves into an aggregate of many, qualitatively different contrasts, coexisting in convoluted and ustable states of contradiction, complementarity and competition with each other. And so, though it remains true on an elementary level that "identity is defined by difference", this statement grossly distorts our understanding of complex social aggregates such as real ethnic identities.

Nevertheless, anthropologists continue to make errors in logical typing of this kind. We refer to whole societies and complex social phenomena (e.g. symbols) as though they were not composites, but "elementary particles". We approach a world which has content (not necessarily because it means anything in itself, but simply because its complexity is too great to comprehend), as if it were "pure" form. All aggregates have their own, internal dynamics, which influence and are influenced by the external opposition (e.g. the ethnic boundary) which the theory of binary contrasts treats as their only valid aspect. Seen from "within" all social phenomena are complex and contradictory webs of meaning. Seen from "without" they participate in larger patterns. We cannot make direct inferences from the "internal" to the "external" pattern or vice versa, nor apply the same model to both. Still (and this is what makes social science such a tangle), the "inner" and the "outer" realities respond to each other - often intensely and in fundamental ways.

When we treat aggregates as if they were elementary particles this complexity disappears, and the analysis degenerates into empty generalizations that say little or nothing about the specificity of the phenomena at issue. It is obvious that there are vast differences between the "ethnic revitalization" taking place in many Muslim countries and among Norwegian Saamis. True, both groups are concerned with establishing identity, and therefore with binary contrasting. But the groups themselves are not even vaguely similar, nor is the world surrounding them the same. The content of the identity they affirm is vastly different.

Understanding the relationship between element and aggregate is not a problem specific to the social sciences. Among other places, it is central to physics. When water boils, one speaks of a "phase transition" having occurred - a transformation from one aggregate condition (fluid) to another (gas). What occurs on a micro-level is well understood. The average velocity of billions of molecules increases until more particles "jump out" of the fluid than are attracted back into it. Still, any physicist knows that the exact boiling point cannot be calculated or predicted by tracing the movements of individual molecules. A map of these movements would be like some monstrous, three-dimensional billiard board, upon which millions of balls were rolling around and knocking against each other. Each movement and each collision follows Newton's laws of motion, and one would therefore think that one would be able to predict exactly the position and velocity of each ball at any point of time in the future. When the water is warmed evenly, a known amount of kinetic energy is transferred to the system - and we thus have all the data needed to predict the exact time when boiling will commence.

But this calculation is not only impractical but impossible. Three important reasons for this should be noted. First, the system consists of very small elements (molecules) - so small that they cannot be observed at all without influencing them (Heissenberg's principle of indeterminacy). One must choose - if one measures the velocity of molecules one cannot know exactly where they are. If one knows their position, one cannot ascertain their exact velocity. Secondly, even a system as small as a single drop of water consists of so many elements that it is impossible - both practically and theoretically - to measure either position or velocity for each of them. Finally, and most important of all, the cumulative interactions between elements very quickly become so complex that even minuscule external influences alter the condition of the whole radically. This is true even of simple, macroscopic systems with few elements. Consider an idealized, frictionless, two-dimensional billiard board with the standard number of balls. The player sets the balls in motion with a single blow from the cue, after which they continue rolling, colliding and bouncing off each other indefinitely. How long will we be able to predict the balls' movements? In principle, as long as the billiard table remains an absolutely closed system, insulated from all external influences, we will be able to extend our predictions into infinity. In practice, the problem is precisely that we are unable to keep the system closed. Thus, it has been demonstrated that after only one minute our predictions will be faulty if we fail to take into account external influences as weak as the gravitational force of a single electron positioned at the other end of our galaxy (Crutchfield 1986, p.41).

We might object that in spite of this, "ideal" conditions will still tend to increase the degree of predictability in the system. It appears, however, that this is often not the case. If, for example, we shoot a round projectile against a window pane, the bullet will make a hole, from which irregular, branching cracks radiate. The pathways of these cracks cannot be predicted, but one would think that if the glass were totally homogeneous, the projectile completely round and the angle of impact absolutely straight, our predictions would turn out to be better than if conditions were less "ideal". But the cracks follow lines of weakness in the glass. So the purer the glass, the greater the influence of even the most minimal impurities. If the glass were "totally pure", even sub-atomic distortions would give visible, macroscopic results. (See Bateson 1979.)

This thought experiment provides a fundamental insight. We tend to think that if everything specific is removed from a situation, the general and universally valid will emerge. But when we purify the glass we rob it of its status as a specific object with more or less predictable traits, a kind of "will of its own" (e.g. to crack right there). The glass, which originally had a clearly defined identity, making it unique among all window panes in the universe, seems to have been transformed into a sort of Platonic Idea of Glass. But this transformation is impossible, in theory as well as in practice, because the glass is not one thing, but a composite - an aggregate. The "purer" we make it, the more it loses its established individuality and autonomy and the clearer does its underlying compositeness emerge - its aggregate character. Thus an "ideal window pane" is merely an unfinished aggregate - an object on the verge of becoming specific and unique but kept from it by artificial means. It is like super-heated water. The cleaner the water, the more it can be heated without starting to boil. But at the same time, the external influence needed to set it boiling becomes less, and once it gets started the transition is explosive and unpredictable. A "pure" window pane is an object with no "will of its own", and so even the smallest external impulse will in an instant transform it into something completely idiosyncratic and indeterminate.

We can predict the movement of one molecule or one billiard ball, and calculate (approximately) the course of a single sequence of interaction. We can also - roughly - predict how specific aggregates will behave under specific conditions. But we cannot know how idealized aggregates will behave under ideal circumstances, because all objects are put together in a specific way and therefore behave characteristically and uniquely.

What Kind of Reality?

At this point we easily end up misusing a theory like that of Lévi-Strauss. We can not simplify the world into "binary oppositions", and pretend that this is how real objects, people and symbols behave. This would be a world of "idealized circumstances" - and, as we have seen, idealized worlds are the most unpredictable of all. What we can do, however, is to formulate a set of rules pertaining to real objects, with non-idealized properties - a set of guide-lines for how such objects may be described as unique, which will later enable us to make tentative predictions based on the fact of their uniqueness.

Lévi-Strauss is challenging because his concepts can teach us how to create such specific descriptions. He lets us understand how aggregates are formed - what an aggregate is. He points out the direction we must go to get from "social particle physics" to the macroscopic world of people, things and meaning. But he says next to nothing about the consequences of this movement. Reality is created out of opposition. But what kind of reality? We go from diametric to concentric duality, and further to "triads" of real objects. But Lévi-Strauss has nothing to say about these "triads" except that they all originated in the same process. This is as if a physicist were content to state e.g. that iron and cotton both are built up out of protons, neutrons and electrons, without attempting to specify the aggregate qualities (specific gravity, color, hardness etc.), of the materials themselves. The "context" generated by the Lévi-Straussian model is as void of qualities as Figure 3. It is an aggregate without identity - like super-heated water or a Platonic window pane. It is true as Lévi-Strauss and Bateson say, that we give names to contrasts, transform "quantity" into "number", difference into identity. But we need to get further, to look for definitions of the various types of names, types of meaning, types of identity that exist.

Nor is it enough simply to assert that we are now on an aggregate level, where the laws of "social particle physics" no longer apply. The examples quoted above demonstrate the opposite. It is in fact possible to create aggregates which are more or less empty of identity - Platonic billiard boards and window panes, superheated teapots. But in such aggregates particle physics reemerge as tangible reality: the ideal window cracked at that precise point, because a single electron vibrated just so, perhaps an electron at the opposite end of the galaxy. The same is true of social analysis. Once in a while one may in fact succeed in predicting the movements of mass humanity. It is indeed the case that the mass follows different laws than the individual. Still, there are places and times when society itself is "super-heated", when a single individual, a single act, changes the course of history.

We therefore need theories that not only distinguish elements from aggregates, but also connect them. Physics has laws of this kind. Newton's and Einstein's laws of motion differ, but are expressible in the same equations, and at times it is important that this is so (e.g. when computing the distance to remote stars). My point is not that the social sciences should seek to express their internal consistency in the same way as physics. But it is worthwhile to formulate certain demands as to what transformations a theory of the Lévi-Straussian type must undergo in order to become useful in the "real" world of aggregates.

I shall list three such demands, all of which pertain to the transition from element to aggregate, to the creation of concrete social identities out of abstract, binary opposition. The list does not pretend to be complete but gives a reasonable idea of the problems to be overcome:

a) The theory must be transformable into a theory of historical change, in particular it must be adapted to analyses of the nature and evolution of power.

b) The theory must address the question of individual freedom. Under what circumstances can one choose in the Lévi-Straussian "structure", and what consequences does free choice entail for the model as a whole?

c) The theory must not only explain different degrees of complexity, but also differing complexity types. How, for example, can the differences between economy, religion, politics etc. be "generated" from binary opposition - which per definition lacks all specific content?

For reasons of space, this article will only address the first of these questions. Let me once again state the purpose of the following: To build a bridge between "micro-theory" and "aggregate theory", from a theory like that of Lévi-Strauss, which says nothing except how the simplest building-blocks of culture are created, to a theory of the almost infinitely complex and composite aggregates which are the object of "normal" social science.

We might perhaps think that a treatment of historical change was implicit in what has already been said. Is not the whole point of Lévi-Strauss' model to show how aggregates are generated - is it not a model of genesis and change? But the kind of genesis described by Lévi-Strauss is a-historical, synchronic. It says nothing about how social phenomena arise and change, merely how they (and their meanings) are reproduced.

To give this model a historical dimension, we must take one of Lévi-Strauss' more puzzling statements at face value - his assertion that he is heavily influened by Karl Marx (see Lévi-Strauss 1962, p.130; 1963, p.332). Part of what he means by this is simply that the generative process we have described above is dialectical: diametric duality corresponds to the opposition of thesis and antithesis; concentric duality marks the transition to synthesis - to an aggregate, an autonomous unit, an autonomous element in a triad. But if dialectics are to describe change, the point of departure must be a contradiction - a specific kind of conflict.⁽³⁾ This demands that the term "binary opposition" acquire a radically different meaning. On the aggregate level an opposition is no longer neutral and empty of meaning. It is far more like a struggle, where different interests or principles are opposed. The units in opposition are now no longer "elementary particles", but independent aggregates, with their own inner structure and a specific relationship to the external world.

This also has consequences for the result of the process. The "name" created by the process of dialectics cannot be adequately described simply as a "number" formed by the relationship between "quantities". The "number" we are now dealing with is a result of conflict between other, pre-existent "numbers". What emerges from their conflict is not complexity per se, but increasing degrees of complexity. The process no longer starts from point zero - from a simple, binary opposition - but from an already existing reality. The process changes this reality to another, which differs qualitatively from the point of departure.

We must now distinguish between two main kinds of change which may result from the dialectical process:

a) The original conflict may be neutralized, its force contained. The result is then a system of checks and balances which increase the complexity of the whole. The system does not change in the linear, irreversible sense. Instead it is gradually stabilized, surrounded by a dense web of meaning which strengthens its autonomy and increases its equilibrium. The system as a whole becomes more durable and flexible. Bateson (1970b) defines flexibility as "unused potential for change". This, from another point of view, implies that the system contains a multitude of more or less identical and overlapping meanings - it is highly redundant. This is the very opposite of a Platonic window pane - a strongly individualized entity, more or less immune to outside influences, which is therefore highly predictable (i.e. stable).

b) If conflict is not neutralized it may be coordinated. This is achieved by establishing a factor which governs, rather than balances the interaction between the hostile parties. The governor has the right and/or capacity to siphon off a certain amount of the energy flowing between the contestants, which it pools and utilizes to limit and direct the conflict into non-destructive channels. Thus a hierarchy of power is established. This kind of change is clearly linear and cumulative - since power, once established, does not voluntarily abdicate. The truth goes deeper than this, however. A hierarchy of power implies an increase in abstraction, since the governor in a sense represents a more general and over-arching reality than the governed. Reversing this process is therefore equivalent to erasing acquired knowledge, returning to innocence, and - as the story of Eve and the apple shows - this is not easy.

We have now laid the groundwork for transforming the micro-theory of Lévi-Strauss to a theory of historical aggregates and their transformations. We shall go on to apply these insights in an extended discussion of Lévi-Strauss' analysis of dual organizations.

One of the societies he presents in the article is an African village of the Bororo people. The village is built up in a circle around a focal point - the men's house and the dancing ground. The women live with their children in separate houses along the rim of the circle. Further out are women's gardens, and beyond, the forest. The circle is divided into moieties - but there are two such moiety-divisions, one running North-South, the other East-West. Thus four quadrants are formed: North-West, South-West, South-East and North-East.

Figure 4

As the figure shows, this is by no means the end of the matter. In each quadrant live two clans, each moiety thus consists of four clans. Each clan is composed of three classes: U(pper), M(iddle) and L(ower). The clans and classes are distributed along the rim of the village as shown in Figure 4. The resulting pattern thus includes both diametric and concentric duality, as well as triads. All stages in the abstract process we have been discussing are present in concrete form.

But this is a real village not an illustration of a theory. What we see is therefore not the process itself but a kind of "precipitate" of each of its abstract stages into the world of real things. In fact, the "precipitates" we have mentioned so far are only a few examples out of a much larger list - the real society is far more complex than Figure 4. Much of this additional complexity is discussed by Lévi-Strauss, but I shall mention just one additional factor here: Four of the clans are distinguished from the rest by their association with culture heros, and are accorded higher prestige. These are the clans numbered 1 and 4, 6 and 7 in the Figure. Each of these two couples are situated "on" one axis of diametrical division, which they are therefore said to "govern". Clans 1 and 4 govern the East-West axis, clans 6 and 7 the North-South axis. Now it would seem that the moieties containing governing clans would naturally attempt to make a bid for some kind of dominant position in the society as a whole. But the two couples are distributed in a very strange way: Clans 1 and 4 lie on the same side of the horizontal axis. To assert the superiority of their moiety they would have to argue the predominance of the East-West axis at the expense of the axis North-South. By doing this however, they would cut their own moiety off from the moiety they sought to control, thus splitting society more deeply instead of coordinating it. Clans 6 and 7 are defeated by the opposite dilemma: Situated on opposite sides of the North-South axis, they would isolate themselves from each other by emphasizing it.

Neither couple is able to assert its dominance, for the simple reason that they could only do so by abstracting themselves "out" of both moieties and "becoming" the axis they dominate. Only in this way can an axis become a center, and diametric duality be transformed into concentric - a necessary first step towards establishment of power relations. It is interesting to note the underlying reason why this movement does not succeed. Each of the two diametrical divisions is in fact on the verge of transforming itself into the center of the circle. But the dominant clans attached to each axis base their domination on opposite principles. Clans 1 and 4 seek to dominate by separating themselves from the dominated; clans 6 and 7 by embracing them. The two principles working together might in fact succeed in establishing hierarchic relations, but since they are attached to different axes such cooperation is precluded. Instead, they end up neutralizing each other.

What we have seen in other words, is a concrete result of the first type of dialectics that I discussed above - which neutralizes conflict and produces balance and flexibility rather than power. External impulses tend to be absorbed by this system, by virtue of its intricate mesh of overlapping claims, its "checks and balances". It is a system with firm identity, high degree of predictability and strong resistance to change.

How then is power produced? To develop a model which will generate systematic inequality we must start by reducing the equilibrium of this society, so conflicts no longer neutralize each other. The simplest way of achieving this is by reducing the redundancy in the system, removing organizational principles that, from a purely instrumental point of view, are "unnecessary". We start out, in other words, by "purifying the glass".

This however, is not enough. A "purified Bororo village" would be liable to change, but it would be too unpredictable to guarantee systematic, cumulative change. I shall not here enter into a discussion of the preconditions for such change, but merely note that in a historical perspective such change is extremely rare. The dialectics of flexibility and equilibrium are far more common than the dialectics of power. Still, we can illustrate some of the preconditions for such change by contrasting two very similar historical situations - the first evading an increase in power, the second achieving it.

The first example is an anthropological classic - Bronislaw Malinowski's (1922) description of the Trobriand islanders and their "big-men". Marshall Sahlins⁽⁴⁾ (1963), defines a big-man as a political leader who acquires prestige by distributing goods and tying people to him by "the power in the gift" (Mauss 1925). The Trobriand big-man gains access to the goods he gives away by manipulating with marriages. This is a matrilineal society and inheritance follows the female line from male to male (from father to daughter's husband, rather than from father to son). Nevertheless, a man lives with his wife. To reconcile the contradictions inherent in this situation a man sends food-stuffs - often in very large quantities - to his sister, or rather, to her children - since her children, rather than his, are his rightful descendants. Since this is still a male-dominated society however, the person controlling these gifts is not the sister, but her husband.

A big-man must first of all have many wives. This makes him the receptor of great quantities of food, which he can distribute to the rest of the population - thus increasing his prestige. The problem with the system is that the more a big-man builds up his position among the population at large, the more he must press his wives' families for contributions. Naturally there is a limit. The families get fed up, gang up against him and throw him out of office.

The example shows a type of "equilibrium" reminiscent of the Bororo. The big-man tries to give a "name" of power to a relationship (of marriage and kin), but the name is unable to force the relationship to continue upholding it for long. The Trobrianders refuse to be "counted". Or, to put it simply - the big-man's power base is too weak. His only source of resources is a group (his kin), which is smaller than the group to which he distributes (the people). (See Figure 5.a.)

To sustain power relations this situation must be reversed, as was the case in the early Germanic retinue system (cf. Anderson 1974a). The wandering chiefs of ancient Europe enlisted a small, mobile gang of armed men with whose help they collected tribute from the population. Like the big-man, the chief based his position on redistributing "gifts". But the recipient was in this case a much smaller group (the retinue) than the group from whom goods were collected. (See Figure 5.b.) For this reason the chief's "name" was strong enough to keep its hold on the relationships it "counted", and stable relations of power resulted.

Figure 5 summarizes the basic dilemma of power relations succinctly. However, it brings us no closer to the question of how a "loss of balance" comes about and is focused into cumulative change. We may assume that some kind of imbalance lay behind the evolution of the retinue system but we have gained no insight into what exactly it was. This should not surprise us, since anthropologists and historians generally disagree strongly on this point, and my feeling is that no really satisfying general answer has ever been given. As the big-men showed us, dominance is by its very nature undesirable for the dominated. They are bound to resist it, and since they are in the majority they are bound to succeed, other things being equal. Perhaps there exists no general reason why other things should "not be equal", and we must seek the origins of power in the specific historical conditions at those few times and places when it in fact prevailed. I shall therefore briefly recapitulate one instance, which is better documented than most, the evolution of the kingdom of Swat in Northern Pakistan through the last century or so (Barth 1959, 1980).

Frederik Barth's analysis of Swat society comprises a monograph and several articles, and little of this material is relevant to the present discussion. However, a few central facts must be recapitulated: Until the 1930s Swat society was dominated by a large number of petty chiefs who mobilized unstable support-groups, consisting mainly of lower-caste peasants, artisans etc. According to Barth, the lack of stability in these mini-hierarchies arose from the fact that no single organizational principle (kinship, locality, caste etc.) was strong enough underpin a stable political leadership. Although exploitation of the peasants was a fact of life, they thus had a certain freedom to chose which chief they should let themselves be exploited by. The chiefs on the other hand, were engaged in ceaseless political manipulation to retain their position, and to this end would shift in and out of alliances with each other. In spite of the fluidity of this situation, the end result was a stable political division of the whole area into two opposed power blocks. But since chiefs were constantly changing sides, the system was unable to sustain a stable centralized authority embracing both blocks. Or to keep to the terminology we have used so far: There was no social relation in the area strong enough to carry a "name" with such great power.

Figure 5

Aside from the chiefs there was also another group of landowners, the "saints". These had a slightly different type of power base from the chiefs', and also a contrasting code of ethics and deportment. The chiefs were expected to be boastful, strong, generous and highly sensitive about their honor - and were therefore constantly embroiled in feuds. The saints were to be humble, restrained and holy. In the traditional system they acted as arbitrators between chiefs.

This is another example of a "balanced" relation - between chief a saint. The power of each neutralized that of the other, and thus precluded centralization. However the political position of the saints was perhaps marginally superior to that of the chiefs, because of their role as arbitrators. Thus, in the course of the 19th century centralized states arose several times in the area, surviving no more than a few years - but each time under the leadership of a saint. The crucial factor acting in favor of the saints may have been their access to information, since their land-holdings were spread throughout the Swat valley - thus enabling them to collect information and exert influence over a wide area. Around the turn of the century another saintly state arose by manipulating relationships between the political blocks, and this time its position held. Clearly, a change in the total state of equilibrium must have occurred for this to be possible - probably a result of some external influence. But it is also clear that this influence must have been very weak in its initial stages - the area was not colonized and the British presence was hardly felt. It is possible, however, that a marginal improvement in communications may have occurred. This might have tipped the scales yet another fraction of an inch in favor of the saints - which could have been enough to create a social relation strong enough to uphold the new "name" of power.

At the outset two "binary oppositions" (block vs. block and saint vs. chief) neutralized each other, none of them being strong enough to carry the new "name" - a situation formally similar to that of the Bororo. Then followed an external influence - slight, but enough to tip the scales in favor of the second relationship, which became strong enough to be "counted" - to be given a new "name", the saintly state of Swat.

I have attempted to demonstrate how the Lévi-Straussian model can be transformed into an instrument for analysis of power and historical change. The precondition was a re-definition of the concept of opposition to a relationship of conflict, which could serve as a basis for dialectical processes leading either to increased flexibility and stability or to increased power. The transformation of abstract, binary contrast to dialectical conflict corresponds to the transition from element to aggregate. Contrast concerns "social particle physics". Dialectics involves the tangible realities of everyday life.

I should stress that I do not imagine my reading of Lévi-Strauss to be even remotely like his own. True, here and there he does advocate a historical theory of sorts, e.g. in "The Savage Mind", which gives an extensive discussion of the relationship between primitive and modern thought, based on the contrast between two ideal types: The bricoleur and the engineer. He repeatedly stresses that there is a distinction in principle, not merely in degree, between these two - an unbridgeable gap it may seem at times: The bricoleur constructs "structures out of events". He collects "bits and pieces" of pre-existent meaning and rounds them off roughly so they "fit together". The result is a structure rather like a collage, assembled out of elements which already have significance and do not lose it in their new context. The purpose of the bricoleur's activity is not to create new "events" - but to superimpose a new structure, a new pattern on the world as it is. We recognize the Bororo world in this - the more pattern, the more "solid" the identity, and the more flexible and durable the system. The engineer on the other hand creates "events out of structures". Starting with an abstract model he forces the world to change, like a real architect or engineer.

There is something very pleasing and (I think) true about these thoughts, but at the same time the underlying contrast - the opposite relationships between structure and event - is faulty. What the engineer and scientist are in fact doing is to go from event (experience) to structure, to new event, to new structure, and so on. But the engineer's activity is cumulative, linear, it is directed towards a goal. The creativity of the bricoleur on the other hand, is cyclic. He alternates between structure and event like the engineer, but does not constantly look for new events - or rather, does not seek to create new events by means of his structures. This may seem rather superfluous criticism, but on the background of what has been said above I think we may say that it is not. The essential point which is lost in Lévi-Strauss' reasoning is that the activity of the bricoleur tends towards equilibrium, while the very root of engineering is the systematic and cumulative utilization of imbalance as a means to sustain power and increase its range. In fact this very point is made by Lévi-Strauss himself in his description of artistic creativity, which seeks harmonization of structure and event. What he seems not to see is that both the bricoleur and the engineer can be artists. In his description the primitive emerges as the artist, modern man as artless. It is indeed tempting to see it in this way since it all too often corresponds to reality. Still, primitive societies can also be thrown out of balance. And one can seek - and find - balance in the modern world.

Binary opposition is a sharply honed instrument of analysis, but cannot be used uncritically as an instrument of description. This is what is done when ethnic identity is reduced to a matter of contrast pure and simple, or when Lévi-Strauss absolutizes the opposition of primitive and modern man. It is an attractive feature, however, that he lets this contrast benefit the former. But what he's really doing, we suspect, is to honor the artist - disguised as the primitive - perhaps because he himself is an artist at heart. As he puts it, in words reminiscent of the quote from Geertz that we introduced him with: "My mind is neolithic..."

Notes

1. "Meaning" as used in the following is not merely a cognitive category. A hoe, for example, has a "meaning" that derives from the fact that it can be used in a specific way. We must have knowledge about it - understand it - to utilize it. "Meaning" may therefore often be read as synonymous with "utility", "efficacy", "purpose".

2. I use the word "aggregate" to denote composite units. A central assumption in the following is that all the objects and phenomena we recognise as units in our daily lives are in fact composites - e.g. concepts, acts, roles, institutions, symbols. Some semiotically inclined thinkers, e.g. Roland Barthes (1975), seem to assert that a word has an underlying culturally defined "true", "pure", "elementary" meaning - a standpoint I strongly disagree with. Roger M. Keesing (1970) has in a similar vein argued that social roles are not unitary, but aggregates which may be decomposed into micro-elements or "building-blocks". The concept of aggregates is discussed in greater detail below.

3. This word is used for the sake of simplicity. Marx' concept of contradiction is complex, and has been the subject of a number of important theoretical debates, which I cannot enter into here.

4. Substantial objections have been raised against Sahlins' account. Briefly, it has been pointed out (a) that the regional contrast between Polynesian and Melanesian political leadership is far more variable than Sahlins seems to assume, and (b) that the evolutionist bias of his text seems to imply that Polynesian chiefdoms arose directly from more egalitarian political forms of the Melanesian type. In the present context these considerations need not concern us. I do not wish to imply that Polynesian society arose on a Melanesian basis, but rather that all hierarchical political forms at some previous time in their history superceded more egalitarian political forms. My concern in the following is to speculate on the reasons why this change did not take place e.g. in Melanesia, while it did, e.g. in Europe.