Skip to navigation | Skip to content

Lévi-Strauss’s ‘double twist’ and controlled comparison: transformational relations between neighbouring societies


“They create a desolation and call it peace.”

Calgacus, Caledonian chief, speaking of the Romans


Here I want to address two aspects of Lévi-Strauss’s work which, I think, point outwards to new possibilities for anthropology. The first is the concept of the ensemble, “larger entity”, from the last chapter of The Naked Man, which I will not attempt to define here, but simply try to empirically prove. The second is the canonical formula or double twist. First proposed by Lévi-Strauss in 1955 in the article on the structural study of myth [1963], and used sporadically by its author and others ever since, the canonical formula looks like this:

fx(a) : fy(b) ≃ fx(b) : fa-1(y)

or this,

fx(a) : fy(b) :: fx(b) : fa-1(y)

As someone who basically sleep-walked through higher mathematics at school, this formula means almost nothing to me as such. What I do know, however, is that the canonical formula concerns a specific kind of transformation, indeed a transformation of a transformation, the double twist. In what follows, I am going to make no further use of the mathematical expression of the canonical formula, and concentrate on empirical examples of it.


In its initial formulation, the canonical formula is used by its author to study transformational variation within given bodies of myth variants, along with a brief glance at aspects of generalized exchange in his theory of kinship and another at the genesis of the neurosis in Freudian theory. However, in the Mythologiques and its companion volumes, Lévi-Strauss increasingly uses the canonical formula or double twist to explore a very specific type of mythic transformation, that caused by the presence of thresholds, whether cultural or linguistic. In particular, when a myth crosses such a linguistic or cultural threshold, it will be subject to transformations of the type explored by the canonical formula.

As far as I can see, Lévi-Strauss’s canonical formula or double twist is a social anthropological response to a problem that originates in classical cultural anthropology, and in particular American and Brazilian cultural anthropologies. For cultural anthropologists, any given ethnographic study of a local community, whether the Bella Coola or the Tapirapé, is never an end in itself, for each local community is seen to merge into larger units, such as Northwest Coast cultures or Tupian cultures, in the cases cited. Each local community is both a part of, and an expressive variant of, such larger units. As such, cultural anthropologists were not particularly bothered about thresholds, apart from the sense in which such thresholds were the boundaries across which cultural traits did or did not diffuse. For cultural anthropologists, the ongoing accumulation of ethnographic data posed no problem, since each new study simply added more and more data to the knowledge of both local communities and the larger units.

The situation for social anthropologists was very different. Malinowski’s functionalism provided a very powerful technique for studying local communities and their life ways as tightly integrated functional systems. In The Argonauts, the reader can trace a discernable retreat from an interest in larger cultural units, such as Rivers’s “the history of Melanesian society”, associated with shallow ‘survey’ ethnography, towards the monumental form of “the Trobrianders”, a vastly intricate object revealed by in-depth ethnography via participant observation. Here linguistic and cultural thresholds almost cease to be of any real interest, since all they do is act as ‘boundary conditions’, as the physicists say, to the real object, the Trobrianders.

Malinowski’s method, however, posed a problem: what exactly is the point of new ethnographic studies? Why study Malekula or Tikopia if the Trobriand islanders were already revealed in their full glory as the savage society? What was the point of ever more studies of primitive societies as vastly intricate objects, as functioning systems? Malinowski had already shown us that. This problem was solved by Radcliffe-Brown, who proposed that the point of studying many different local communities, or societies as he called them, was controlled comparison. By multiplying the number of societies which had been studied in depth, the number of cases was increased, and thereby the potential of comparisons among them proliferated. This solution was, I think, brilliant, and testament to Radcliffe-Brown’s genius: new ethnographies became intellectually exciting, and the new comparative possibilities endless.

But it left social anthropology with a problem with thresholds, whether linguistic or cultural. Because such thresholds were the ‘boundary conditions’ of the objects studied by social anthropology, they were not amenable to analysis by that science’s methods.[1] A pertinent example here would be the Harvard Central Brazil Project, and especially the volume Dialectical Societies, edited by David Maybury-Lewis [1979]. There the Gê-Bororo peoples of Central Brazil, the Xavante, Xerente, Apinajé, and so on are studied as discrete objects amenable to various competing modes of controlled comparison. What is not addressed in that volume is the obvious fact that these various Gê-Bororo peoples are also neighbours, and engaged in complex social relations with each other, such as those between the Bororo and western Xavante peoples. These social relations across linguistic and cultural thresholds were largely ignored by the members of the Harvard Central Brazil Project because they did not know what to do with them, even though the whole point of the project was premised on the Gê-Bororo peoples forming a higher-order cultural and historical unity. It was only later ethnographers who were able to deal with such relations across thresholds, such as Caiuby on the Bororo and western Xavante [1997], or Ewart on the Panará and their relations with their enemies [2003, 2013], especially the Kayapó. Indeed, Ewart notes in her study that the members the Harvard Central Brazil Project habitually ignored the remarkable isomorphism between local and global dualism by treating the former as social structure, the latter as social change.

Given that the members of the Harvard Central Brazil Project were supposedly in dialogue with the work of Lévi-Strauss, it seems remarkable that none of them ever addressed this problem, especially when we consider that the Mythologiques opens with the mythology of precisely these same Gê-Bororo peoples treated as a higher level unit, linguistically, culturally and historically. What we find in Dialectical Societies is a remarkably impoverished version of what Lévi-Strauss had achieved in the Mythologiques. Of course, it could be argued that Dialectical Societies is primarily concerned with social structure and especially kinship, the meat-and-potatoes of social anthropology, while the Mythologiques is concerned with the arcane topic of mythology, the stuff of an alien cultural anthropology. But Lévi-Strauss always considered himself to be a social anthropologist for whom kinship and myth are parallel problems. This is an issue to which I will return, after considering the implications for an understanding of Lévi-Strauss’s work on a Boasian problem.

A Boasian problem

If you look at Lévi-Strauss’s work on myth, it looks Boasian. This look is almost certainly intentional. In particular, the Mythologiques looks like those immense collections of North American myths published by Boas, collections such as his Tsimshian Mythology. This monumental work is probably familiar to social anthropologists only because it is the source of the myth of Asdiwal, famously analysed by Lévi-Strauss. In his conclusion to the volume, Boas wrote,

The distribution of plots and incidents of North American folklore presents a strong contrast when compared to that found in Europe. European folk-tales, while differing in diction and local coloring, exhibit remarkable uniformity of contents. Incidents, plots, and arrangement are very much alike over a wide territory. The incidents of American lore are hardly less widely distributed; but the make-up of the stories exhibits much wider divergence, corresponding to the greater diversification of cultural types. [1916:877-8]

Anyone with even a passing familiarity with the two mythological systems would have to recognize the great justice of this observation.

Boas goes on to contrast the dramatic cultural heterogeneity of North American cultures to the homogeneity of European ones, and continues:

It is evident that the integration of European cultural types has progressed much further during the last two or three thousand years than that of the American types.

For this reason European folk-lore creates the impression that the whole stories are units and that their cohesion is strong, the whole complex very old. The analysis of American material, on the other hand, demonstrates complex stories are new, that there is little cohesion between the component elements, and that the really old parts of tales are the incidents and a few simple plots. [1916:878]

What historical evidence does Boas have that complex stories are old in Europe and new in North America? With few exceptions, the interest in collecting folkloric materials in the two continents is roughly contemporaneous, and derives from very similar social conditions. Indeed, my impressions from recent work on Highland Scottish folk traditions suggests that we actually have a slightly better historical record for American folklore than for European folklore, despite how odd that fact may seem (Gow [2008]). Certainly, we have no genuine historical evidence that the complex story is older in Europe than in the Americas other than that such stories are more homogenous in the former and more heterogeneous in the latter. All we really have is Boas’s assumption that greater homogeneity reflects greater time-depth to diffusional processes. That assumption is quite probably false. Contemporary international airports are much more homogenous than contemporary schools of Buddhist thought, but Buddhism is thousands of years older than international airports. If Boas’s assumption is false, then his historical conclusion is false too.

But the contrast between American and European folklores still stands, and cries out for some kind of explanation. I suggest an alternative hypothesis to explain the contrast that Boas draws. We simply do not know whether complex stories are any older in the Americas than in Europe, so historical explanations are out of the window, or are at least of very limited scientific value. We do know, as Boas brilliantly shows, that complex stories are much more heterogeneous in the Americas than in Europe. As such, it is methodologically more useful to view this fact in terms of what we know about the observable differences between indigenous American and European societies. How is the homogeneity or heterogeneity of the indigenous complex story related to the nature of the kind of society within which it is told?

Take the example of Highland Scotland. Since the Scottish Enlightenment in the latter half of the eighteenth century, it has been commonplace to draw a basic analogy between the traditional social forms of Highland Scotland and Native North America as examples of tribal societies. The analogy was not entirely spurious, for there were important points of commonality. But there was also a key difference. Every Highland community, even the most remote such as the islands of St Kilda some thirty five miles off the west coast of the Outer Hebrides, was linked by direct ties of land ownership to the Scottish crown through the mediation of the aristocracy. The power of the tighearna, the Highland chiefs, rested on oighreachd, “royally-sanctioned title to land”. This meant that even people as apparently remote from cosmopolitan centres as the inhabitants of St Kilda were linked, by remarkably few steps, to the Scottish crown, the monarch him- or herself (Gow [2009]). Before the Reformation they were also, through the monarch and the clergy, linked very closely to a Europe-wide institution, the Catholic Church, an institution with a central concern with homogeneity.

The contrast with the traditional social forms of Native North America is dramatic. Here local communities were hardly isolated from each other, and were indeed engaged in complex and intense interactions over very long distances, but there was nothing even remotely resembling an overarching centralized authority or universal church. Each local community was its own central authority and its own universal church. Even among the chiefly societies, perhaps even aristocratic societies, of the Pacific Northwest Coast, local communities jealously guarded their autonomy from each other, and this autonomy was real. But autonomy is not isolation, for as Lévi-Strauss has noted, following such pioneers as Boas, there was a vigorous dialogue among such communities. But that dialogue was necessarily a two-way street, since in the absence of any centralized authority or universal church, there was no institutional means for any given local community to impose its particular views on any other.

It would seem to me that it is in terms of these divergent traditional social forms in Highland Scotland and in North America that we can better understand the contrast noted by Boas between the folklores of Europe and the Americas. In the former, complex stories are homogenous precisely because there are complex social mechanisms of homogeneity. In the latter, complex stories are heterogeneous because there are complex social mechanisms of heterogeneity. Why this difference in complex social mechanisms exists is an interesting historical problem, and may even be knowable, but I do not think that the relative complexity of mythological stories has much light to shine upon it.

The genuine interest of Boas’s insight about the difference between European and American folklore seems to lie elsewhere. In regimes of complex social mechanisms of heterogeneity, obvious and strongly marked cultural and linguistic thresholds abound. By contrast, in regimes of complex social mechanisms of homogeneity, obvious and strongly marked cultural and linguistic thresholds become increasingly fewer and further between, or go underground. As such, the former regimes become the privileged ground for exploring Lévi-Strauss’s canonical formula or ‘double twist’, while the latter would require much greater methodological subtlety to find.

Ethnography in a cultural key

To explore this problem, I want to look at a series of complex stories from South America. In an article on “Twin Heroes in South American Mythology”, published in 1946, the Swiss ethnologist Alfred Métraux summarises a myth he describes as of “extremely wide diffusion” [1946:122] as follows:

… the wife of the Creator or Culture Hero is killed by jaguars that find twins in her womb. The jaguar mother brings up the twins. Later they learn from some animal that the jaguars among whom they are living, are the murderers of their mother. They take revenge and then, after performing several miraculous deeds, climb to the sky by means of a chain of arrows and become Sun and Moon. [Metraux 1946:119]

Versions of this myth can be found in the area in which I have done most of my fieldwork, southeastern Peruvian Amazonia. The Piro or Yine people, speakers of an Arawakan language and living along the Bajo Urubamba river, tell a variant of this myth as one of their two major mythic narratives: it is the epic of the birth of the miraculous creator Tsla and his brothers, the Muchkajine, the “First White People”, from the womb of their mother who had been murdered by jaguars, and of their subsequent revenge on her killers (Gow [2001]). To the west, along the foothills of the Andes, the Yanesha’ people, speakers of a distantly related Arawakan language, tell their own variant as a cosmogonically central story, the origin of the Sun and his twin sister the Moon, Yompor Ror and Yachor Arrorr, “Grandfather Sun” and “Grandmother Moon”, who are major divinities (Santos Granero [1991]). Their mother is a secluded virgin who is killed by a female jaguar, who subsequently finds and raises the twins, but plots to kill and cook them for her sons. Instead, the twins kill and cook the female jaguar and feed her to her sons, who they then kill.

In between the Piro and the Yanesha’ live the Asháninka people, speakers of another Arawakan language distantly related to both of the languages of the former two. Gerald Weiss, an American cultural anthropologist and Boasian, in his monumental study of the Asháninka, Campa Cosmology [1975], demonstrates the centrality to that cosmology of the myth, or myth cycle, about the moon and the sun. In gross summary this myth tells of how Kashiri, the moon, came down to earth and seduced a young girl, who subsequently gave birth to Pawá or Inti, the sun, who in the act of being born burned his mother to death. Both the moon and the sun are tasorentsi, divinities, and Weiss’ work and all subsequent studies of the Asháninka people have shown the importance of this myth, and of these divinities, to the understanding of their cosmology.

Citing Métraux’s work on the twin heroes mythology, Weiss asked himself why this myth was absent among the Asháninka, given its key importance among their western and eastern neighbours. Actually, he notes, this myth is not absent among the Asháninka, for he recorded a curious myth about the sun and his sister from his informant Komémpe from the community of Osheriato on the Alto Tambo river. It goes as follows:

…the Sun… visits a community of jaguars in the form of a small boy with cricket characteristics, that is he takes the form of a cricket at a time when all creatures are human. He is accompanied by his “sister”, who is not otherwise identified. The jaguars welcome the apparently defenceless pair and plan to make a meal of them. They are left in the charge of the mother of the jaguars, while the rest go off to hunt. The jaguar mother prepares a boiling pot to receive the children, but instead she is thrown into it by the boy, who thus demonstrates his superior strength, for he is Pawa [the Sun]. The other jaguars return and begin to eat the meat waiting for them, presuming that it is the cooked flesh of the children, until the truth dawns on them. They try to capture the children to take revenge, but are unsuccessful. Finally they go off to their jaguar neighbours to recount the events, completely baffled as to the identity of the children who have shown such remarkable strength. [1975:483]

Weiss records that he thought that this myth was both somewhat incoherent (for example, the sister plays no role in the story beyond being mentioned) and to be a recent borrowing from the Yanesha’ people to the west, among whom, as we have seen, it is a major cosmogonic myth. As already noted, in the Yanesha’ version the murder and cooking of the jaguar mother is motivated by revenge for her murder of the twins’ own mother. Of this presumptive borrowing of a Yanesha’ myth by the Asháninka, Weiss wrote:

The indications are that the borrowing occurred recently and that familiarity with the myth is not widespread among the [Asháninka]. The obstacles in the path of assimilation to [Asháninka] cosmology of this myth in its usual form appear to be insurmountable. The [Asháninka] cannot identify the twins as the Sun and the Moon so long as they hold that the Moon is the father of the Sun, nor can they ascribe the death of the Sun’s mother to attack by a jaguar so long as they hold to the belief that the mother of the Sun was burned to death on giving birth to that deity. [1975:483]

This ethnographic data, and Weiss’s argument, raises some crucial methodological issues. As is so often the case, we simply have no idea whether this myth was indeed borrowed, whether it was indeed borrowed from the Yanesha’, or whether, if it was indeed borrowed from them, this borrowing was recent or ancient. Weiss’s argument only works if it was indeed recently borrowed from the Yanesha’, which we do not know. Indeed, the actual ethnographic evidence would seem to point in another direction. The narrator Komémpe presumably experienced this myth as an Asháninka one, and indeed it has some crucial differences to the Yanesha’ myth.

Komémpe lived in the community of Osheriato on the Alto Tambo river, roughly halfway by Asháninka standards between the territories of the Yanesha’ and the Piro. Her story contains an apparently unimportant detail, as follows:

[The jaguars] try to capture the children to take revenge but are unsuccessful. Finally they go off to their jaguar neighbours to recount the events, completely baffled as to the identity of the children who have shown such remarkable strength.

This apparently trivial ending is extremely marked in terms of the corresponding Yanesha’ and Piro versions, for both deal with the destruction of jaguar sociality. Both Yanesha’ and Piro versions are careful to note that only one female jaguar survived the massacre, a female pregnant with a male child, thereby implying that present-day jaguars, notable for their solitary habits, are the products of mother-son incest. So, Komémpe’s story is not a failed borrowing of a myth, but a dramatic creative transformation of myths from neighbouring peoples. It is, in fact, an excellent example of the Lévi-Straussian double twist: far from dealing with the end of the jaguars’ social lives, as in the Yanesha’ and Piro versions, the Asháninka version ends on an intensification of the social lives of jaguars, for these jaguars have neighbours who they visit.

Komémpe’s story is instructive about what exactly we do and do not know about the Asháninka and their neighbours. Despite the relatively long colonial intervention in Eastern Peru by the Spanish, and in particular by the Franciscan and Jesuit missionary orders, we know very little about the history of the mythologies of these peoples. Komémpe’s story might be a recent borrowing from Yanesha’ people, or it might be an ancient borrowing from Yanesha’ people, or it might not be a borrowing from Yanesha’ people at all. We simply do not know, and it seems to me very unlikely that we are ever going to find out. So, that problem is probably always going to languish among the files called, “Unsolved”.

What we do know, however, is that since the latter half of the sixteenth century, the Yanesha’ people have been the western neighbours of the Asháninka people, and the Piro people have been their eastern neighbours. That fact is not trivial, for it implies over four centuries of time depth to the cultural and linguistic thresholds between these peoples. Indeed, you can still see those cultural and linguistic thresholds to this day in Eastern Peru, albeit in highly modified forms. Given what we know about the history of these cultural and linguistic thresholds, it would seem much better to assume that they are places of intense and ongoing dialogue, places where Lévi-Strauss’s canonical formula or double twist will help us with our analysis of the available data, rather than, as Weiss’s argument suggests, odd and special cases within a larger frame ordered by different rules and thus amenable only to equally odd and special analysis.

Let us look again at the ethnographic evidence. From a wider point of view, Asháninka mythology looks quite strange. Not only does it not use the twin heroes mythology to any great cosmological purpose, it is also unusual in making a male moon the father of a male sun. The Asháninka are located in between two major variants of the treatment of the moon as a mythological character in Western South America. To the west, from the Yanesha’ up into the Central Andes and including Inka state ideology, the moon is female, the twin sister and divine incestuous spouse of the male sun. To the east, from the Piro people of the Bajo Urubamba far into Brazil and beyond, the moon is male and the nocturnal seducer of a young woman, often his sister: here the sun and the moon are not coordinated mythologically. Living in between these two major variants of the mythic treatment of the moon, it is as if the Asháninka were trying to have the best of both worlds, keeping both the male moon as seducer and the mythic coordination of the moon with the sun. To do so, however, they have been forced to make the moon and the sun into father and son. Or to use less voluntaristic language, and to move closer to the canonical formula or double twist, Asháninka mythology is structured as a transformer, in the electrical sense, between these two major, and radically incompatible, mythic systems.

In saying this, I am not suggesting that Asháninka mythology or cosmology is incoherent, or not tightly integrated as a self-consistent functional system. It most definitely is, as Weiss’ excellent study clearly shows. Instead, I am arguing that it is fully engaged with its own boundary conditions in the thresholds that constitute it as Asháninka mythology and cosmology, rather than the mythologies or cosmologies of its neighbours, the Yanesha’ and Piro peoples.

These data both confirm and challenge Boas’s observations on the heterogeneity of the complex story in American folklore. This heterogeneity is real. But it cannot reasonably be explained by the recency of the complex story in the Americas as opposed to its antiquity in Europe, since we have no evidence at all for that hypothesis. It is much more likely to be a function of the social heterogeneity of American societies, such as the Piro, Yanesha’ and Asháninka discussed here, and the consequent proliferation of linguistic and cultural thresholds between them, thresholds governing the specific nature of mythic transformations across them. Such transformations are characterized by the double twist of the canonical formula, as Piro and Yanesha’ stories about the end of jaguar sociality transform into an Asháninka story about their ongoing proliferation, or when a male sun almost completely loses his twin to acquire a seductive father in the moon. Such transformations may reflect events of borrowings between neighbouring societies, but they might also simply reflect the manner in which people from neighbouring societies maintain an elaborated knowledge of each others’ cosmologies and mythologies.

Ethnography in a social key

Social anthropologists who are basically content with structural-functionalism, the marriage between Malinowski’s field methods (the functional bit) and Radcliffe-Brown’s theoretical method of controlled comparison (the structural bit) are usually hostile to Lévi-Straussian structuralism, largely because it seems to remove the ‘functional bit’ from their analytical vocabulary.

Social anthropologists tend to see a key shift in Lévi-Strauss’s work from the earlier studies of kinship, which tend to be seen as ‘good’, to the later studies of myth, which tend to be seen as ‘bad’. Such judgments are obviously not made in terms of some objective decline in the quality of the work, but much more in terms of a resentful sense of abandonment.

For social anthropologists, the study of kinship was always seen as more real, more grounded, than the study of myth. Undoubtedly, this has had to do with the perceived isomorphism of kinship systems and of social systems as systems. Ever since Morgan, anthropologists had known that “systems of consanguinity and affinity” were systems, in the sense of sets of parts, the individual kin term or relation, connected together as a coordinated whole, the kinship system as an integrated totality. From there, it was easy for the structural-functionalists to see the isomorphism with the Durkheimian view of society, and to see kinship systems as the “royal road” to the understanding of the societies that they studied.

By contrast with kinship systems, the mythology of any given society does not seem to have this kind of systematic nature, since it is difficult to imagine that any ethnographer could be confident that he or she had collected every myth told in the community studied. There is always the real possibility that an informant, such as Komémpe of Osheriato, will come up with a myth we have never heard before. We would be rather more surprized if such an informant suddenly told us about an utterly idiosyncratic kinship terminology, unused by her fellows.

Lévi-Strauss himself seems to confirm the prejudices of the structural-functionalists in this regard, when he speaks in the “Overture” to the Mythologiques of the key difference he sees between his major projects on kinship and mythology. Kinship, he argues, is constrained by real objects in the world, and the order it reveals may therefore not be a pure expression of the nature of the human mind. Mythology, by contrast, is constrained by no real object in the world other than by itself, that is, by the nature of the human mind [1970:10]. This contrast, which led Lévi-Strauss to privilege the latter over the former, certainly helps to explain the regard in which the structural-functionalists held the former, and their distaste for the latter.

In the light of this contrast that Lévi-Strauss himself makes between kinship and mythology, one would not expect the canonical formula or double twist to be of any interest for the study of kinship. A myth, unconstrained by the real world, could follow that double twist, but presumably a kinship system, or even a kinship term, constrained by real objects in the world, could not. Yet, when he first proposed the canonical formula, he made a cryptic reference to his theory of generalized exchange, so maybe it is actually possible.

If we were going to find the double twist at work in kinship systems, where would we expect to find it? Precisely, I suggest, where we find it in the myths, at the linguistic and cultural thresholds. Now, we know that the cultural anthropologists were happy to interpret kinship systems as subject to borrowing, even although it is often difficult for social anthropologists to follow such arguments: consider Lévi-Strauss’s own incredulity at Boas’s very convoluted discussions of Kwakiutl kinship as a complex product of borrowing, and his own rather elegant reinterpretation of the latter’s data [1983]. Social anthropologists certainly would not disagree with the cultural anthropologists that kinship systems change over time, but they would certainly be wary of using data collected at the time scale normally available to ethnographers to prove it.

I suggest another possibility here, using the observations I have been making thus far about how Lévi-Strauss’s canonical formula or double twist can make us rethink controlled comparison. There would be nothing controversial in suggesting that the kinship systems of the Piro, Yanesha’ and Asháninka peoples are variants of what Viveiros de Castro has called the Amazonian Dravidianate, and it would be entirely possible to compare them as discrete objects, in the manner of the structural-functionalists. But they are more than discrete objects, for they are the kinship systems of neighbouring peoples, and therefore ripe for the sorts of transformations studied by the canonical formula.

To start with, in the area under consideration, there is the basic fact of multilingualism. The community I have studied, Santa Clara, is effectively trilingual in Piro, Asháninka (in two major dialects) and Ucayali Spanish. I say “effectively trilingual” advisedly, because it was common to hear many more languages spoken there on an everyday basis. Erudite Coastal Peruvian Spanish, Andalucian Spanish, Quechua, Machiguenga, Amahuaca and English would be obvious examples, but I have also heard two men spontaneously switching into Portuguese, a language that I did not then understand, when they discovered that they had both formerly lived in Brazil. Further, there is no expectation that two speakers will use a common language for any given conversation.

That said, across the three major languages spoken in Santa Clara, there are quite important differences in the internal structure of the kinship terminology. For example, there is a major difference between Piro and Asháninka in the treatment of consanguines in Ego’s own generation: the former emphasizes relative age, the latter the sex of both the speaker and the kinsperson referred to. Therefore, in referential kin term usage, Piro speakers must clearly state whether a sibling is older or younger, and in a less marked sense the sex of that sibling (in address, only relative age is marked). An Asháninka speaker must specify both the sex of the sibling and of the speaker, but need not specify relative age. Ucayali Spanish is different again.

Ever since Morgan, of course, anthropologists have known a lot about kinship terminologies, but the actual patterns of kin term usage in Santa Clara constantly present local people with the problem of translation between kinship terminologies which are internally structured in different ways. For example, Old Clotilde, a Piro woman, and Inez, an Asháninka woman, were neighbours and always conversed in Asháninka. They would call each other éétio, which means, “same sex-sibling of a female speaker”. When I asked Old Clotide’s son Artemio in Ucayali Spanish what this word meant, he told me that it meant hermana, “sister”, and that they called each other thus because Clotilde had called Inez’ mother in Asháninka, ina, “mother”. Similarly, Inez’s Asháninka husband Julian and his younger brother Marcial would call each other iye, “same-sex sibling of a male speaker”. This word, Artemio told me, was the Asháninka equivalent of Piro yeye, the vocative term for “older sibling”.

Now, Artemio told me that he could not speak Asháninka, by which he meant that he could not compose coherent, extended phrases in that language and hence could not converse easily in it. Indeed, he gave me a clear example of both his linguistic incompetence with Asháninka and its consequences when he told me that he wanted to learn to speak that language to acquire Asháninka women as lovers, which he currently could not do. But he had grown up hearing that language spoken constantly around him, not least by his mother, and certainly understood it with a very high degree of competence. As such he must have known that éétio does not really correspond to Ucayali Spanish hermana, since he can never have heard a man classifying a woman by the former term: as a term of reference, the Ucayali Spanish word is used by both male and female speakers. Similarly, he must have known that Asháninka iye and Piro yeye were not equivalent, because he had constantly heard Julian and Marcial reciprocally calling each other by the former term, a situation unimaginable for the latter term. But rather than speculate about these differences as an anthropologist might, Artemio simply answered my questions with these translations. Everyday, people in Santa Clara are confronted by such linguistic and cultural thresholds in the intimate details of ongoing social life, but they act as if such thresholds do not exist.

Artemio’s translations of these Asháninka kin terms has a very interesting and simple feature: he understood both of my questions, not as questions about the meaning of words per se, but as questions about relationships. Language differences certainly matter to people in Santa Clara, but the nature of social relationships matter more. Despite the linguistic and cultural complexities in a place like Santa Clara, local people’s usage of kin terms is actually very consistent. With the exception of small children learning to speak, local people invariably reply to an older interlocutor with the language-specific reciprocal of the term with which they have been addressed. They will often not even bother to speculate about why that term is used, assuming that the older person knows more about the nature of relationship than they do. Hence, Artemio’s translations of éétio and iye were technically correct, for they were made from the perspectives of the two younger people, Inez and Marcial, not from the perspectives of the two older people, his mother and Julian.

Artemio’s translations, and social life in general in places like Santa Clara, therefore seem to invert Lévi-Strauss’s canonical formula or double twist, for here linguistic and cultural thresholds are treated as if no transformations at all occur across them. Local people act as if the dramatic structural differences between the kinship terminologies of Piro, Asháninka and Ucayali Spanish do not exist, and I have never succeeded in gettting any of them to engage in a discussion of this apparently obvious problem. It seems to me, however, that this local inversion of Lévi-Strauss’s canonical formula or double twist is remarkably insightful, for it raises in a dramatic form a question that social anthropologists consider a non-problem: why are the Piro and Asháninka peoples different in the first place?

For social anthropologists, this is a non-problem because it is simply a fact: the linguistic and cultural thresholds between the Piro and Asháninka peoples are simply the boundary conditions of the two different societies, and as such largely unanalysable. But as I noted above, we know that the Piro and Asháninka peoples have been living next to each other for over four centuries, and probably much longer, and have long engaged in intermarriages, trading and other social relations. Why, then, have their kinship terminologies not progressively migrated towards a common form? My anthropological problem, the radical structural differences between criteria for the designation of siblings between Piro and Asháninka, must have been a constant issue for generations of these people. Why, instead of solving it, have they decided to treat it as a non-problem?

The answer is that local people do not experience difference between the Piro and the Asháninka as a difference between societies, but instead as a difference between “habitual ways” of doing things, as they would say. Local people certainly conceive of the Piro and the Asháninka as distinct populations, with distinct languages, cosmologies, historical trajectories and original territories, but they do not think of them as being distinct units within which privileged social relations occur, or between which social relations have a secondary character. Social relations, for local people, are about how particular people choose to act towards others.

For example, Clotilde Gordon was a very prestigious Piro woman, renowned for her knowledge of Piro ways of doing things. She had been initiated as a pubescent girl, and had the pierced nasal septum and lower lip to show for it. But she also had certain personal idiosyncracies. She would cook breakfast for her husband and others who depended on her cooking after the Piro style; but she would also cook a separate breakfast that she would always offer to others, but which was seldom taken up by them, and which she would then eat alone. This was baked manioc. Baked manioc, along with heated manioc beer, is a very distinctively Asháninka breakfast. It is not that Clotilde’s husband or children or other recipients of her food thought that baked manioc was unacceptable as food. They simply did not like it very much, while she did. Clotilde sometimes identified herself as Asháninka, and always greeted unfamiliar Asháninka visitors with nosháninka, “my kinsperson” in that language. None of this affected her high prestige as a Piro woman, and Piro people generally accept my knowledge of their “habitual ways” when I explain that I had lived Klotilte-yegi, “where Clotilde lived”.

Click to enlargeFieldwork photograph of Peter Gow with Sara Fasabi, his comadre and youngest daughter of Old Clotilde described in the text, taken by her nephew and Gow's godson Israel (son of Artemio, also discussed in the text): this is certainly the first time Israel had ever used a camera.

Fieldwork photograph of Peter Gow with Sara Fasabi, his comadre and youngest daughter of Old Clotilde described in the text, taken by her nephew and Gow’s godson Israel (son of Artemio, also discussed in the text): this is certainly the first time Israel had ever used a camera.

To return to the complex stories discussed before, we can see a connection to this issue. Even where the cultural and linguistic thresholds were manifest in everyday life, local people gave them little weight. They followed simple social algorithms to sort out the manifest differences between neighbouring kin terminologies, rather than debating which system was more ‘correct’, and they made great play of personal idiosyncracy to neutralize what might otherwise constitute divisive issues of cultural difference. The central emphasis was on “living well”. By contrast, local myths revelled in such differences. This is because myths are precisely not about “living well”, but rather about the cosmogonic potentials of “living badly”. It is, the myths assert, impossible to “live well” with such radical difference as people who consider each other, not as kin, but as food: the myths recount attempts to “live well” with jaguars, and their disastrous consequences.

Myths can therefore openly deal with thresholds as thresholds. Thus, the Piro myth of the ‘heroic twins’ deals with the theme of birth order and its consequences, so central to Piro kinship: the miraculous creator Tsla is unambiguously the older sibling of his multiple twins, the First White People. The Asháninka myth of the ‘heroic twins’ deals with the Asháninka kinship theme of opposite sex sibling relations. Weiss thought that the “sister” of the Sun was “an entirely superfluous character” in Komémpe’s story. Viewed from Santa Clara, this difference is saturated with meaning. Dialoguing back and forth between Piro and Asháninka people, and constantly addressing the differences that they know to exist between them, the “sister of the Sun” in Komémpe’s story doesn’t have to do much. It is enough that she is there.


In the final chapter of The Naked Man, itself the final volume of the Mythologiques, Lévi-Strauss writes:

It is high time anthropology freed itself from the illusion gratuitously invented by the functionalists, who mistake the practical limitations imposed upon them by the studies they advocate for the absolute properties of the objects with which they are dealing. An anthropologist may confine himself for a one or more years within a small social unit, group or village, and endeavour to grasp it as a totality, but this is no reason for imagining that the unit, at levels other than the one at which convenience or necessity has placed him, does not merge in varying degrees into larger entities, the existence of which remains, more often than not, unsuspected. [1981:609]

The French original of the Weightmans’s “larger entities” is ensembles. As my colleague Adam Reed has kindly pointed out, the word ensemble is interesting precisely because it points towards a coordinated whole made up of discrete parts. It is, in short, a system.

Lévi-Strauss himself has explored such ensembles in two places, in the chapters on his ethnographic researches in Tristes Tropiques, and in his analysis of masks and myths from the Pacific Northwest Coast in The Way of Masks, one of the three companion volumes to the Mythologiques. He concludes that book with this dramatic image:

By gathering scattered threads, I have only tried to reconstruct the backdrop for a stage some two thousand kilometres wide and perhaps three to four hundred kilometres deep, along whose entire stretch the actors of a play for which we do not have the script have left their footprints. [1983:228]

It seems to me that what Lévi-Strauss is pointing towards in his concept of the ensembles or “larger entities”, along with the canonical formula or double twist, is a way out of the impasse of controlled comparison as the raison d’être of ongoing ethnographic fieldwork for social anthropologists. If social anthropologists could abandon the conceit that linguistic and cultural thresholds are merely the boundary conditions of discrete objects, and see them as the locii of specific kinds of social relations, then the proliferation of ethnographic case studies need not end up in relatively sterile comparative exercises like Dialectical Societies, but might reveal the as yet unknown social properties of “larger entities” such as the Gê-Bororo peoples and their neighbours in Central Brazil.

The Lévi-Straussian ensembles are very poorly known phenomena, and here I have sketched out some of the features of one that I am just beginning to study explicitly as an ensemble in southwestern Amazonia. It might seem that my argument here would be substantively restricted to social anthropologists working in Amazonia, or perhaps the Americas more generally. I do not think that this is true – so to conclude, I want to briefly suggest the potentials of the argument for another ensemble, the one we are in right now.

I quoted above Boas on the genuine contrast between American and European folklore, his explanation for it, and my reservations about that argument. But Boas was referring to a genuine difference, and it seems very likely to me that it was with Boas’s argument in mind that Lévi-Strauss wrote, in the “Overture” to the Mythologiques:

From the start then, I ask the historian to look upon Indian America as a kind of Middle Ages which lacked a Rome… [1970:8]

By the same token, Europe is a kind of Indian America with a Rome, and therefore, Lévi-Strauss’s work in the Mythologiques may not be entirely irrelevant to our wider concerns.

Exactly the same contrast that Boas noted between American and European folklores is found in the differences between American and European kinship systems: the former are remarkable for their heterogeneity, the latter for their homogeneity. This fact is not trivial for social anthropology. It is the extreme homogeneity of European kinship systems that arguably blinded European thought to the fact of the extreme diversity of human kinship systems, and hence to a very important strand in anthropological thought. You could broaden your mind by travelling from one end of Europe to the other, as the Enlightenment thinkers did, and never confront that diversity. It took a mid-nineteenth century American thinker, Lewis Henry Morgan, to discover it in its full anthropological force among the Iroquois people in upstate New York and then to relentlessly pursue it across the whole world, and in the process found anthropology as a modern science. It is not that no European had ever noticed such diversity before, but they had never really bothered to think through what it meant.

Why it took European thought so long to really notice the diversity of human kinship systems is an anthropological question of some interest in its own right. The beginnings of an answer has been provided by a social anthropologist of impeccable structural-functionalist credentials, Jack Goody. In two fascinating books, Goody challenges every single spurious claim about the exceptionalism of European kinship systems, including Morgan’s own ([1983], [2000]). Instead, he focuses on the specific historical processes by which kinship systems which were anciently much more diverse became increasingly homogenized, indeed so homogenized that Europeans ceased to be aware that this homogeneity was unusual. Goody’s answer to this problem is, of course, Rome in both its senses: both the empire that began the process of vertical integration, and the church that relentlessly and aggressively pursued homogenization as a technique of control over its subject peoples, their bodies and their possessions.

To view Europe as an Indian America with a Rome is necessarily to acknowledge that long process of homogenization, which is still going on, but it does not mean that all linguistic or cultural thresholds have thereby been destroyed, and that the Lévi-Straussian double-twist has ceased to operate here. I suspect that it is alive and well, but masked under our obsession with forms of identity such as nationalism, what Freud called our narcissism with respect to minor differences, or even with identity itself, which, as Lévi-Strauss himself observed (in Benoist [1977]), is simply alterity reduced to zero.



The article was originally presented as an invited lecture at the Association of Portuguese Anthropology in September 2009, at the very kind invitation of Susana Viegas. I thank Anne-Christine Taylor, Aparecida Vilaça, Marcio Goldman and Tania Lima for their comments on the original, and Charles Stafford for the invitation to publish it in AOTC.



Benoist, Jean-Marie (ed.) 1977, L’Identité: Séminaire dirigé par Claude Lévi-Strauss, Bernard Grassett.

Boas, Franz 1916, Tsimshian Mythology, Smithsonian Institution, Bureau of American Ethnology, 31st Annual Report.

Caiuby Novaes, Sylvia 1997, The Play of Mirrors: The Representation of Self Mirrored in the Other, University of Texas Press.

Ewart, Elizabeth 2003, “Lines and circles: images of time in a Panará village”, Journal of the Royal Anthropological Institute 9 (2):261-279.

Ewart, Elizabeth 2013, Space and Society in Central Brazil: a Panará Ethnography, Bloomsbury/LSE Monographs on Social Anthropology.

Goody, Jack 1983, The Development of the Family and Marriage in Europe, Cambridge.

Goody, Jack 2000, The European Family: An Historico-Anthropological Essay, Blackwell Publishers.

Gow, Peter 2001,  An Amazonian Myth and its History, Oxford University Press.

Gow, Peter 2008, “The Cripple Tailor and the Deer: A Scottish myth and its history”, Inaugural professorial lecture, the University of St Andrews.

Gow, Peter 2009, “Answering Daimã’s Question: The Ontogeny of an Anthropological Epistemology in Eighteenth Century Scotland” Social Analysis, Vol. 53(2):19-39.

Lévi-Strauss, Claude 1963, Structural Anthropology, Basic Books, Inc.

Lévi-Strauss, Claude 1970,The Raw and the Cooked, Jonathan Cape.

Lévi-Strauss, Claude 1977, Structural Anthropology II, Allen Lane.

Lévi-Strauss, Claude 1981, The Naked Man, Jonathan Cape.

Lévi-Strauss, Claude 1983, The Way of Masks, Jonathan Cape.

Maybury-Lewis, David (ed.) 1979, Dialectical Societies: The Gê and Bororo of Central Brazil, Harvard University Press.

Métraux, Alfred 1946, “Twin Heroes in South American Mythology” The Journal of American Folklore, Vol. 59 (232):114-123.

Santos Granero, Fernando 1991, The Power of Love: The Moral Uses of Knowledge amongst the Amuesha of Central Peru, The Athlone Press.

Weiss, Gerald 1975, Campa Cosmology: The World of a Forest Tribe in South America, American Museum of Natural History.

  1. [1]Note the exceptions of Leach and Barth, and their focus on the maximizing individual.

Please join our mailing list to receive notification of new issues