The Rise of Hierarchy

Some primal termite knocked on wood
And tasted it, and found it good!
And that is why your Cousin May
Fell through the parlor floor today.
— Ogden Nash

Most anthropologists consider themselves politically left of center, and to a large extent contemporary anthropological theory could be said to reflect a left-wing, or at the very least broadly populist, sensibility. So much is this true that we often act as if conservative currents within the discipline do not even exist, or, at best, are limited to marginal and vaguely comical figures like Napoleon Chagnon.

True, courses in the history of anthropology will acknowledge that this was not always so. But the usual way to deal with that is to represent all early anthropology, up until perhaps around 1965 or 1968, as inherently imperialist and racist, whereafter it appears to morph overnight into the uniformly progressive discipline it is presumed to be today. Needless to say, in reality things were never this simple. As with any other discipline, anthropology was marked from the start by a number of often clashing political allegiances and perspectives, and continues so today. An honest history of the discipline, for instance, would show that conservative anthropology has taken very different forms in different national traditions. In the United States, for example, the 'anthropological right' has tended at least since the Cold War to be entangled to one degree or another in the national security state, the paradigmatic and oft-cited example being Clifford Geertz's CIA-funded research in Java and Bali in the 1950s, and the broader effort at Harvard and Yale to find anthropology a place in State Department efforts to create a Weberian social science to counterpose Marx. To be honest, though, it is not clear that it is entirely fair to call such authors 'right wing'; they might better be thought of as Cold War liberals. Here in the United Kingdom, things were very different. There was an overtly conservative movement within anthropology, such as the Catholic traditionalism of Mary Douglas or the latter-day Evans-Pritchard. Intellectual historians usually attribute this to a retreat into hierarchy as a bastion of traditional stability in the wake of the chaos and destruction of World War II and the welfare state governments that followed.[1] It is not clear how influential this current really turned out to be. Certainly, Mary Douglas's work was hugely successful inside and outside the academy, but most of her more explicitly political interventions (in favor of consumerism, against the ecological movement) were not.

France, again, is quite another story. Much though Anglophones have come to think of Paris as a continual font of radical ideas, the anthropological tradition in that country has been predominantly right of center, and far and away the most influential conservative ideas in the discipline have come from France. I speak here not primarily of Claude Levi-Strauss, who though in no sense a man of the Left remained intentionally coy about the political implications of his own work, but above all to the work of Louis Dumont. Dumont's project, about which he was in no sense coy, was, superficially, to introduce the notion of 'hierarchy' as a central tool of anthropological analysis. In a deeper sense, however, I think it was to change the basic mythic structure of the discipline, and through it all academic disciplines, and through that, ultimately, popular common sense. While this might seem an astoundingly ambitious and quixotic intellectual project, the remarkable thing is that it was largely successful. What I would like to talk about in this essay, then, is what this project really was, as well as its long-term political and theoretical effects.

First of all, the project. Where once theorists all saw themselves, after their own fashion, as grappling with Rousseau's problem of understanding the nature and origins of social inequality, Dumont sought to substitute an entirely different problematic—one that assumed society is, by its nature, necessarily hierarchical (since, he argued, societies are structures of meaning and meaning is always organized on hierarchical terms). Therefore, the thing to be explained was how modern, egalitarian ideologies could ever have emerged to begin with. For Dumont, equality itself cannot be a value. This would be a contradiction in terms, since value is by definition the placing of one thing above another. What would appear to be an egalitarian ideology, then, cannot really be an embrace of equality, which is simply the lack of evaluation, but must be the side effect of something else—in the case of modem societies, he argued, individualism.

This is an extremely conservative argument, and Dumont made his case by staking out a series of equally extreme, and in some ways obviously absurd, positions: that all societies other than modern, individualistic ones can be considered 'holistic'; that all holistic societies are hierarchical; that all hierarchies are based on an interlocking set of binary oppositions; and that all binary oppositions take the form of marked and unmarked terms in which the superior value encompasses the lower. It is not clear whether he really believed all this to be true (it obviously is not), but such positions can easily be defended on circular grounds: any 'society' that does not conceive itself as a hierarchical totality is not really a society, any binary opposition that does not involve 'encompassment of the contrary' is not really a binary opposition, and so forth.[2] Taking this kind of maximalist position is often a very effective political stratagem, since it allows one to define the field of debate. Before long, everyone is arguing over whether hierarchy is itself a value, or whether it is really so all-encompassing, but no one is arguing about the relevance of the term itself.

This is precisely what happened in Dumont's case.

Dumont himself was known to complain that his efforts had come to nothing. "I have been trying in recent years to sell the profession the idea of hierarchy, with little success" he wrote in Essays on Individualism (1986: 235). In fact, the project was almost unimaginably successful. True, the word 'hierarchy', barely used at all in the early years of the discipline, was already becoming more popular at the time he was writing. But after the publication of Dumont's ([1966] 1980) Homo Hierarchicus, its rise became spectacular.

Below is the total number of articles in JSTOR's collection of anthropology journals in English in which the word 'hierarchy' appears, followed by the percent of the total number of all articles:

Even allowing for the slight leveling off in recent years, this is startling. Over the course of a century, the percentage of works using the word has gone up by a factor of 50.

It would appear, then, that 'hierarchy' has gradually become a term of choice for describing social arrangements that would previously have been described in other terms, for example, 'rank', 'dominance', 'social stratification', or simply 'inequality'. To give a sense of the change, here is a list of a series of terms for unequal social relations, with a breakdown of how many times each appears in two books about the same Nilotic people: Evans-Pritchard's (1940) classic study, The Nuer, and Sharon Hutchinson's (1996) Nuer Dilemmas:

In Evans-Pritchard's account, neither the word 'hierarchy' nor any of its cognates appears even once. In Hutchinson, the term appears a total of 45 times, roughly once every eight pages. No other word on this list appears more often, although some terms appear a roughly approximate number of times. At the same time, purely descriptive words like 'rank' and 'dominance' decline sharply, and the word 'stratification' disappears altogether.

This is not simply a reflection of authorial taste, nor can it be attributed to the difference between British social and American cultural anthropology. In the definitive collection of Franz Boas's (1940) essays, the word 'hierarchy' never appears, and it is either entirely or nearly absent from classics of the Boasian school, such as Benedict's (1934) Patterns of Culture (two appearances) or Kroeber's (1947) Configurations of Culture Growth (zero appearances), just as it is in the collected essays of Radcliffe-Brown (1952, zero) and Malinowski (1944, two). The change in vocabulary would appear, then, to reflect a much broader transformation in the habits of both ethnographic description and comparative analysis.

Where once anthropologists had tended to present simple, often quite cold-blooded descriptions of relations of rank and power between groups (to speak of 'dominant lineages', in the Nuer case, 'social stratification', and so on), and only then to consider how those relationships came to be ideologically legitimated, there has been a broad drift toward terms—'status', 'authority', and especially 'hierarchy'—that imply those relationships are always already legitimated, or even that they do not need to be legitimated since they are fundamentally constitutive of social reality itself. Now, obviously, all this cannot be an effect of the work of Louis Dumont. The change seems to find its roots in a much more general tendency toward the intellectualization of social life that had already begun well before he took up his pen. To adopt a somewhat ungainly metaphor, the discipline was already beginning to roll in a certain direction, and Dumont just gave it a very sharp kick—speeding things up immeasurably by insisting that we think of myths, rituals, and patterns of marriage alliance not just as mental structures, first and foremost, but as power relations as well.

Such was Dumont's success that in anthropology as it is written today, the question that this volume addresses—how is it that hierarchy has come to be seen as legitimate?—often seems close to redundant. The moment one labels power relations 'hierarchy', one is already claiming that they are considered legitimate. This is because hierarchical arrangements are themselves viewed as the criteria for legitimacy. In fact, I think one could go so far as to say that, given the way we have come to organize our theoretical terms, it is well-nigh impossible nowadays to write an anthropological work that is genuinely critical of relations of what used to be called 'social stratification', because imagining a world without them would be very close to inconceivable.

The intellectualization of social life began, of course, with Claude Levi-Strauss. He himself largely avoided the subject of power and domination, except when it came to gender, where he argued that male power over women was definitive, universal, and (it would appear) unobjectionable. Especially after The Elementary Structures of Kinship, Levi-Strauss (1969) tended to focus on societies where stratification was limited to age and gender, or else, as when he looked at ranked societies such as those of the American northwest coast (Levi-Strauss 1966), on anything other than the inequality itself. In a way, it is all very Rousseauian. There is a sense in much of Levi-Strauss's work that relations of power and exploitation ultimately spoiled the ecologically attuned science of what he terms 'neolithic civilization'—a science that operated by analogy and bricolage in a fundamentally non-hierarchical fashion. The paradigm for this science is totemism. Totemic relations rank neither animal species nor the human groups that are mapped on to them. In fact, Levi-Strauss argues that when rank is introduced, totemism degrades into caste, and the whole system effectively falls apart. Similarly, sacrificial ritual, of which Levi-Strauss strongly disapproves, seems to emerge along with gods and kingdoms, as a false form of engineering that sweeps away the more contemplative science of neolithic thought (ibid.). Rather than thinking through animals, we come to simply massacre them as a way of winning favor with non-existent gods. Once power structures appear, everything goes wrong.

Levi-Strauss's world-weary, apolitical structuralism was conservative, to be sure. But at least one could make an honest case that it was conservative in the best sense of, the term, expressing a desire to conserve social and ecological arrangements that its proponents saw as values in themselves.[3] Dumont had quite a different project in mind. By moving the date of the Fall from Grace from the end of the Paleolithic to the birth of modern individualism, he placed his argument squarely in that tradition of French conservative thought that, in the wake of the Revolution, saw the Terror as a direct consequence of the dissolution of the coherent hierarchical universe of the Middle Ages, in which everyone knew their place.[4] Even the origin of the term 'hierarchy' is theological. It originally meant 'sacred or divine rule' and was first used by PseudoDionysius in the sixth century to designate the orders of celestial intelligences (angels and archangels, thrones, dominions, and powers) that governed the cosmos. It was only in the High Middle Ages that it came to be extended to the ecclesiastical hierarchy modeled after it, and in the Renaissance, to the whole of creation. Dumont's main departure from that tradition is his argument that even medieval Catholicism, in its insistence that all believers were ultimately equal owing to their possession of a unique and incommensurable soul, bore within it seeds of the very individualism that would ultimately destroy it.

According to Dumont, as I have noted, equality cannot be a value because value is hierarchy. Egalitarianism is not and cannot be a value in itself, in Dumontian terms; it can only be a side effect of individualism. However, other traditions—not just the Indian caste system that was his particular area of specialty, but all 'normal' societies, as Dumont put it, from Levi-Strauss's beloved Amazonians to African sultanates—are intrinsically hierarchical in exactly the way that the old reactionary thinkers had imagined the Medieval Church to be. They were each founded on a total ('holistic') view of the cosmos continuous with human society, in which everyone did indeed know his/her place.

If you think about it, this was quite an ingenious inversion of the tradition of Rousseau. For Dumontians, hierarchy, not equality, plays the same role as Rousseau's primordial innocence—one might even say that it is a kind of primordial innocence. The formulation was also a rather clever inversion of contemporary Marxist theories of ideology, as Dumont was occasionally willing to admit. Dumontians agreed with, say, Roland Barthes that ideology was a matter of taking arbitrary power relations and making them seem as if they were inscribed in the very order of the natural universe. But Dumontians saw this as a good thing, since they insisted it was the only way to create a moral order based on stable values of any sort at all. Finally, liberating the conservative tradition from any possible charge of ethnocentrism saved it from the disrepute into which it had largely fallen in the years after World War II. As Nicolas Verdier (2005: 34) has noted, the word 'hierarchy' itself had largely disappeared from the social sciences in the 1950s and 1960s, when Dumont was writing his principal work, tainted by its association with Nazism,[5] Dumont had an answer to that, too. He argued that once modernity is born, and the genie of individualism has been released from its bottle, there is simply no putting it back. Any political project aimed at restoring holism will inevitably lead to totalitarianism, whether of the fascist or communist variety. Stalin and Hitler alike were the products of the impossibility of any return to genuine conservatism.

The Marxist anthropology of the 1970s and 1980s made a brief splash and then largely disappeared. Levi-Straussian structuralism peaked around the same time and is now considered slightly ridiculous. Yet, for some reason, Dumontianism lives on. In fact, the stripped-down, almost cartoonish version of structuralism that Dumont promulgated is really the only sort of structuralism that most anthropologists nowadays take at all seriously. Meanwhile, its key term, hierarchy, has become so utterly pervasive that anthropologists—and I must include myself among them—find it difficult to even think about how one might write about unequal social relations without employing it.

What I would like to suggest, then, is that we might do well to start considering how to think outside the hierarchical box because the effects of the term's ubiquity have in many ways become quite insidious. This is true, I would say, for two reasons above all. The first is ideological. The adoption of the word 'hierarchy' has been essentially to naturalize inequality—not just to treat human systems of domination as forms of meaning (even, in the extreme Dumontian formulation, the only possible form of meaning), but to see them as always already justified. The second is subtler. It is not just that Dumontians claim that holistic societies tend to ground systems of social stratification in the very order of nature; they are doing the same thing themselves when they present their ideas of hierarchy as somehow inherent in the very nature of human language and thought. In other words, they are not just presenting us with a sloppy and ill-thought-out version of structuralism: the sloppiness is inherent in the nature of the program. In order to naturalize hierarchy, they have to persist in logical errors that would otherwise be readily apparent.

This, too, I have written about before (Graeber 1997: 703-709). Here 1 can offer only a brief summary. Essentially, the concept of hierarchy, as currently employed, is based on a kind of conceptual sleight of hand involving the conflation of two different forms of logical operation—ranking and the creation of taxonomies-which, while sometimes overlapping in practice, are in fact entirely distinct.

Ranking involves arranging elements along a single, unilinear chain where any one element is either superior or inferior to any other. The result might be considered a linear hierarchy, since it exists in a single dimension. A classic example of such a linear hierarchy is the notion of 'the chain of being', first proposed (as far as we know) in Plato's Timaeus, developed by Augustine, and popular in Renaissance times. As described by Arthur Lovejoy (1936), this 'Great Chain of Being' ranked every aspect of creation—from angels to animals, plants, and geological formations—on a single scale of proximity to God. Ranking, of course, is key to any system of value, since it enables one to say that one element in the system is superior (more valuable in some way) than another.[6] As Lovejoy was quick to point out, such systems can work only if there is a single criterion of ranking. The Great Chain measured the value, and hence position, of every creature according to the degree to which it was endowed with the faculty of reason (and hence to which it partook of divinity, which was identified with reason). But the moment that any other criteria were introduced, the whole system tended to fall apart.

The birth of modern biology, of course, corresponds to the point where the Great Chain of Being was replaced by Carl Linnaeus's system of taxonomic hierarchy. Taxonomic hierarchies are organized on a completely different principle. They are not linear but operate by creating ranked levels of inclusion: sparrows are included in the larger category of birds, birds are included in the larger class of vertebrates, and so on. While in both cases we can speak of rank orders of a sort, taxonomies involve a very different sort of ranking, since they rank not the elements being classified, but a series of increasingly higher levels of abstraction: sparrow, bird, vertebrate, and so forth. There is no sense in which one sparrow is superior (more sparrow-like, more bird-like, more vertebrate-like) than any other; rather, it is the categories that are ranked, by degree of inclusiveness, not the individuals that make them up.[7] Obviously, this is entirely different from a linear hierarchy, where by definition certain species of bird are superior to others, any given mammal is superior to any given bird, and so on.

Dumont's trick is to treat these two meanings of hierarchy as if they were ultimately the same. He is effectively arguing (I say 'effectively' because he does not make an explicit case for why this should be true, but just writes as if it is self-evident that it is true) that in social hierarchies, one from every category is always selected to represent the more inclusive category. This makes sense if you go back to the first application of the term 'hierarchy' to social relations, where it referred to the ranked organization of prelates in the Catholic Church. Here believers were indeed organized into parishes, each with its priest (one member of the parish who represented the whole parish before God), just as parishes were organized into bishoprics, in which one cleric represented everyone, including subordinate priests, and so forth. So there is both a rank hierarchy of offices—priest, bishop, archbishop, cardinal, Pope—and each is one of a collective meant to represent the whole on a more inclusive level of organization. The paradox here is that the higher up in the hierarchy one is, the more exclusive a group one belongs to, and therefore the more 'sacred' or 'set apart' from ordinary mortals one is taken to be. Yet at the same time, the more inclusive is one's purview, since one represents a larger and more inclusive group. Priests represent their parishioners, who are in a sense 'included' in them, but as a class of people, priests are also an exclusive group of sacred people set apart from the laity. Bishops represent a more inclusive group of parishioners and parish priests, and as a class, they form an even more exclusive and more sacred group set apart from all others. In such systems, rank order and taxonomic hierarchy form a perfect, integrated order.

Nowadays this is indeed a common form of organization. Examples of similar structures, balancing (taxonomic) hierarchies of inclusion with (linear) hierarchies of exclusion, can be found in many contexts in many parts of the world. Armies, for instance, almost always tend to be organized this way. There is a rank order of officers (corporal, sergeant, lieutenant, captain, etc.), and the higher one's rank, the larger and more inclusive a unit of soldiers one commands. Still, it is clearly absurd to argue that all social arrangements, or even all hierarchical ones, are organized by synthesizing linear and taxonomic principles. To take one familiar anthropological example, while ranked Polynesian-style conical clan systems could be said to synthesize the two, most African segmentary forms of organization do not. The Nuer segmentary lineage system is marked by an elaborate taxonomic hierarchy of evermore inclusive sub-lineages, lineages, clans, and tribes, but units on the same level (whether clans, lineages, or for that matter individual lineage members) are not ranked against one another, so no linear hierarchy of exclusions results. On the contrary, Nuer are notoriously egalitarian.

At the same time, within each Nuer tribe, all males are organized into a series of age sets. These age sets are indeed ranked against each other in the linear sense: any individual is either 'senior' or 'junior' to any other. One addresses all people of a higher age set as 'fathers' and of a lower one as `sons'. But this has nothing to do with lineage or clan affiliation. Nor is there any principle of marked and unmarked terms. Members of senior age sets do not 'include' or 'encompass' their juniors: The two principles—taxonomic and linear—co-exist, but they remain separate, or as separate as anyone can possibly allow them to be.

Since ranking and taxonomizing are both very basic logical operations, there is every reason to believe that they will be present, in some form or another, in any human group. People will always be ranking things on scales, saying that this is better or higher or purer or faster or more beautiful than that. They will also always be classifying things into more and more inclusive kinds. It is probably inevitable that they will apply both of these logics to people as well—at least in certain ways in certain contexts. But there is simply no reason to believe that when they do one, they are always necessarily doing the other. Humans are not more mammalian than any other mammal. Five-star restaurants do not 'include' or 'encompass' four-star ones. An 'A' paper does not necessarily incorporate all the good points made in a 'B' paper. True, the two principles can and often do overlap and occasionally fuse together to produce hierarchies of the ecclesiastical, military, or conical clan variety. But there is absolutely no reason to believe that most hierarchies will take this sort of hybrid form, and it is self-evidently absurd to insist that all of them do.

How does Dumont get around this problem? Basically, he begs the question, insisting that since all holistic societies are hierarchical, and all hierarchies are hierarchies of inclusion, then there must be some sense in which the two really are, necessarily, aspects of the same thing. In order to demonstrate this, he often stretches logic to what can only be called a conceptual breaking point. Take, for example, his formal analysis of the Indian varna system (Dumont [1966] 1980: 67). This system looks to all the world like a simple linear hierarchy, a series of groups ranked by ascending order of purity, and this is how he initially describes it. Brahmans (priests) are purer than Kshatriyas (warriors), Kshatriyas are purer than Vaishyas (farmers), and Vaishyas are purer than Shudras (servants).[8] But he then goes on to argue that although this looks like a linear order, it really is not (ibid.):

Thanks to Hocart and, more precisely, to Dumezil, the hierarchy of the varnas can be seen not as a linear order, but as a series of successive dichotomies or inclusions. The set of the four varnas divides into two: the last category, that of the Shudras, is opposed to the block of the first three, whose members are 'twice-born' in the sense that they participate in initiation, second birth, and in the religious life in general. These twice-born in turn divide into two: the Vaishyas are opposed to the block formed by the Kshatriyas and the Brahmans, which in turn divides into two.

So at any point along the ladder, members of a given varna can see themselves as united with those above them in representing humanity as a whole before the gods, and opposed to those below them, who are lumped together as part of the generic, undifferentiated humanity they are representing.

There are two obvious objections to be raised here. One is that what is being described as a series of inclusions is, logically speaking, much easier to describe as a series of exclusions. The Brahmans, who are at the top, see themselves as set off from all others, as particularly pure and holy. From their perspective, everyone else can be seen as a kind of undifferentiated mass, shading into each other and even into non-human creatures, insofar as they all lack the purity of Brahmans. However, from the point of view of the next highest group, the Kshatriyas, the more relevant opposition is that which sets both them and the Brahmans apart from another residual category, which is again relatively impure. Then comes the opposition between the twice-born and the others, which would include both Shudras and presumably whoever else falls in this residual category—Dalits, Adivasis, and so on. This would certainly seem to be how the ritual logic plays itself out in everyday practice.

Now, one can still argue, as a Dumontian no doubt would, that the entire arrangement is 'really' a variation of the logic of a segmentary system, where the more inclusive ranks are also set off as purer, higher, or otherwise superior to those that they encompass. But to make the case that Hinduism is a structure of inclusions requires something very close to special pleading—emphasizing the way things look from the perspective of certain obscure ancient texts, and ignoring almost everything we know about how these categories play themselves out in ordinary life. To go on to argue that all systems of rank order must always necessarily work by such a logic of encompassment is, as I have already pointed out, simply untrue.[9] And to argue that even those that do contain an element of encompassment are not, therefore, 'really' structures of exclusion is not just absurd, but about as politically reactionary as it is possible to be.

Insofar as there is something of lasting importance in all of this—and I would not want to leave the reader with the impression that there is not—it surely turns on the question of value. When Dumontians talk about linear hierarchies as being hierarchies of encompassment, what they mainly seem to be saying is that these hierarchies often involve a kind of ranking of value spheres. This is what Dumont (1982: 230) is getting at when he speaks of "value reversals" on different levels of a hierarchy: in dealings between merchants, wealth is the paramount value, since even power and purity might best be considered valuable as ways of obtaining wealth; between warriors, wealth and purity are subordinated to the interests of power; for priests, wealth and power are only really important as ways of maintaining the ritual purity of people like themselves. Dumont's point is that the varna system ranks not only people but also those value spheres themselves, and that the sphere in which purity is the consummate value is the highest sphere. This in itself is a useful concept, not unlike Weber's notion of status groups (stand) or even Bourdieu's social fields. But the key difference is that for Weber or Bourdieu, the ranking is not fixed—it is always under contestation. In fact, one might even say that the ultimate stakes of politics for either is the ability to assert what one's own group holds dearest as the crowning value of the system. What Dumont refers to as holistic societies, then, are those in which he believes such matters to be definitively, and permanently, settled. Hence, his famous remark that the structure of the Indian. caste system, for instance, cannot, by definition, be changed. It must either stay in place, unaffected by history, or collapse and be replaced entirely:

A form of organization does not change, it is replaced by another; a structure is either present or absent, it does not change. If we are entitled to say that so far the changes that have occurred have not visibly altered what we have taken to be the heart, the living nucleus of the society, who can say but that these changes have not built up their corrosive action in the dark, and that the caste order will not one day collapse like a piece of furniture gnawed from within by termites? (Dumont [1966] 1980: 219; emphasis in original)

While this makes sense according to a certain classic conception of structuralism, it is obviously not the case for contemporary societies. So are we to conclude that structuralism applies to non-Western societies and that only the contemporary West is post-structuralist? It is hard to avoid the impression that this is exactly what Dumont' thinks is going on here. But surely, to insist that pre-modern societies simply do not have politics in this sense, since their overall structure is by definition fixed, and that only in modem societies is value up for grabs is clearly just another variation on the old romantic trope that contrasts timeless societies innocent of history (whether Levi-Strauss's 'cold' societies or Eliade's primordial ones trapped in circular history) with modem ones condemned to live in cumulative, historical time. That this is the one aspect of classical structuralism that is still taken seriously in contemporary anthropology would be very strange indeed.

How does all this bear on the essays assembled in this collection? The unifying premise is that a focus on power, domination, and resistance has made it difficult for anthropologists to talk about the fact that hierarchy (which, the editors note, can in no sense be reduced to power or inequality) is often considered a good thing, even a value in itself, by those we study. Some resist power structures not in the name of opposition to power itself, but in order to restore more familiar forms of hierarchy or utopian hierarchical visions (whether set in the future or in the past).

I would not contest that this is often the case. In fact, the language of 'resistance' took on its present popularity around the same time as the increased appearance of the word 'hierarchy'. This is an interesting phenomenon in itself because one almost never sees an ethnographer speak of resistance to hierarchy, but always instead of resistance to power and domination. This is true despite the fact that the word 'hierarchy' has come to be applied to more and more forms of unequal social relations over time: one now has age hierarchies, gender hierarchies, patronage hierarchies, and so on.

The great political advantage of the term 'resistance', of course, is that it defines political action purely in terms of what it is against, and not what it is for. To draw an analogy from more obvious propaganda, anyone reading mainstream US newspapers in the 1980s would learn that there were basically two types of guerrillas in the world: 'communist' and 'anti-communist'. It was obvious which they were being asked to favor. Any way the anti-communist guerrillas could have been described in terms of what they were actually for would surely have made them sound decidedly less appealing. In a similar way, anthropological discourse that identified popular forces as resisting power, hegemony, global capitalism, and neo-liberal governmentality, or some similar, largely abstract force, managed to avoid grappling with what those engaged in such resistance actually considered a good life worth defending, or what they ultimately aspired to achieve. In other words, this approach allows the analyst to sidestep the question of the values that ultimately motivate such resistance. The self-consciously political framing rather implies that resistance is in the name of some sort of egalitarian ideal or instinct, or at least some principle of justice with egalitarian implications, but this is never stated outright, and often the Foucauldian language used makes it unlikely that this could really be the case. The widespread adoption of the word 'hierarchy' for those forms of inequality that are not being challenged makes a certain complementary sense.

In fact, one might go even further. All this makes perfect sense to describe a world where global power structures—whether NATO or Credit Suisse—increasingly dress themselves up as the voice of human freedom, and the most overtly political forms of opposition to those power structures are increasingly likely to take traditionalist, nationalist, fundamentalist, or authoritarian forms. A Dumontian framework is perfectly tailored to theorize such a world. Dumont himself was no friend of capitalism. Much of his theoretical work (see, e.g., Dumont 1977, 1986, 1994) was dedicated to developing the critique of 'possessive individualism', economism, and utilitarian rationality proposed by left-wing thinkers such as Karl Polanyi (1944), C. B. Macpherson (1962), and Marshall Sahlins (1972; see also Sahlins, et al. 1996)—if only to turn it against the Left as well. In this sense, Louis Dumont might almost be seen as a kind of prophet of conservative anti-capitalist movements to come—movements that barely existed in his lifetime. By 'conservative anti-capitalist movements', I mean those that reject the values of bourgeois modernity, not in the name of egalitarian universalism or of some fantasy of lost European social harmony, but instead in the name of what are taken to be more genuinely holistic (but equally authoritarian) forms of social order preserved on Europe's fringes or within its former imperial dominions. In a world where opposition to the American empire is now being spearheaded above all by Putin's patriarchal authoritarianism, a Chinese government increasingly shedding Marxism for Confucianism, and a ragtag collection of would-be Islamist theocracies, all this seems genuinely prescient.

I have described Dumont's formulations as fundamentally incoherent, based on a false conflation of two kinds of logical operation: linear ranking and the arrangement of taxonomies. I have further argued that adopting the resulting language, and thereby reducing relations of what used to be called power, domination, stratification, or inequality to a single uniform category of hierarchy (as anthropologists of all political stripes have increasingly come to do under Dumont's influence), means presenting those relations never as the contingent result of a play of forces (like Evans-Pritchard's 'dominant lineages'), but always as inherently meaningful arrangements that should be treated as already fully justified in the minds of those we study—indeed, as the very foundation of their sense of what is good and right and beautiful. But surely this is not the case. While some relations of power or inequality might be like this, others are most decidedly not. None go unchallenged. There are tensions within even the most deeply internalized systems—tensions that could, under the right circumstances, allow them to be transformed into something else. Finally, people argue about such matters all the time. It might well be that the only time people are strongly unified around the legitimacy of such arrangements in the way Dumont implies is precisely when they come to seem values in themselves opposed to those of the larger capitalist world-system.

Clearly, we need to rethink our terms. It would be impossible at this juncture to simply jettison the term 'hierarchy' entirely, nor am I suggesting it would be wise to do so. But we definitely would do well to seriously rethink the way we are deploying such terms, and to think more deeply about the tacit assumptions that lie behind their use. This volume, it seems to me, might be considered a first step in such a project. Almost every author sets out to speak of hierarchy in a broadly Dumontian sense and ends up discovering some way that this standard approach is inadequate. Dumont's distinction between hierarchical and individualistic/egalitarian societies is incoherent (Feuchtwang). Rank is not necessarily about inclusion (Smedal, Khan). The structure of values is not necessarily holistic (Haynes and Hickel). Equality can indeed be a value (Howell). The relation of power and legitimacy is not a given (Malaya and Boylston). Hierarchical relations are not prior but continually created through acts of destruction (Damon). Combine all these insights together and very little of the Dumontian edifice remains. Perhaps it, too, should collapse like a piece of furniture gnawed by termites. And on its ruins, we can begin, like the authors in this volume, to think more seriously about what the wellsprings of the deeply felt appeal of unequal forms of social relations actually are.

References

[1] I speak here of outright conservatism: Tories in England, Gaullists in France. There has also been a strain of what I would consider neo-liberal anthropology since the 1980s, of which authors like Arjun Appadurai and Daniel Miller are probably the best-known avatars, but this might be considered a separate phenomenon.

[2] I have critiqued Dumont's formal definition of hierarchy quite sharply in two different places (Graeber 1997, 2001). I have never seen any of these arguments taken up by anyone else or even cited in lists of critiques of Dumont, and I have never understood precisely why.

[3] A case could also be made that Levi-Strauss's structuralism was not conservative. He appears to have been a dogged anti-feminist, and he expelled Pierre Clastres from his laboratoire when Clastres tried to make a case that the anti-authoritarianism of the neolithic societies Levi-Strauss favored might be relevant to the politics of his day.

[4] I would not wish to discount the intellectual power of this tradition, first exemplified by authors like Louis de Bonald, Joseph de Maistre, and of course Auguste Comte. As Nisbett (1966) has pointed out, almost all of the basic problems of the sociological tradition—not just hierarchy, but community, solidarity, authority, alienation—emerge from it.

[5] Verdier (2005: 34) makes this point as follows: "Finally the most recent forms of usage of the word hierarchy can be mentioned, In this respect, one of the most noteworthy aspects is its spread at the end of the 1930s, first in relation to questions of society, in some cases resulting in deviations such as those already noted in the work by Franz Joseph Gall, whose successors are to be found in Germany in the ruins of the Prussian military state and elsewhere, whether among geographers like Christaller ... or among Nazi theoreticians. This probably also explains the disgrace of the word in the years directly after the 1939-45 war."

[6] But not quite any element. There are three ways a value system can operate. It can (1) operate in a binary system, value versus not-value, as with Dumont's individuals who are each unique, incomparable values; (2) form a simple ordinal rank series; or (3) form a more complex cardinal rank system where each item can be seen as a proportion of any other, as with money.

[7] I suppose I should dutifully point out here that cognitive science would qualify this statement somewhat in practice. Whereas in formal terms, taxonomies do operate in this way, in fact we usually have a paradigmatic example of 'bird' in our heads—with English-speakers, often a robin—around which others are measured. However, I am speaking here in terms of the formal logic of the system.

[8] In later times, Vaishyas were more likely to be merchants and Shudras to be farmers. But Dumont is here speaking of the earliest period.

[9] Indeed, such a claim can be maintained only by purely circular arguments, that is, insisting that any way a ruling elite represents its concerns as universal proves that its members see themselves as 'including' those over whom they rule, and any way that the same ruling elite does the opposite, representing its concerns as peculiar to itself, is not an example of exclusiveness, but of the making-sacred of the more inclusive category. By this kind of reasoning, one can obviously prove anything at all. There is absolutely no reason one could not make the exact opposite argument and say that wherever there are structures of exclusion, they have the effect of allowing those in the superior, exclusive group to think of themselves as generic humans simply because they do not have to think very much about those they have excluded.