Dumont, Hierarchy and Its Critics

Homo Hierarchicus (1966) is the most influential and most contested book ever written about caste. Its author, Louis Dumont (1911–1998), was a French structuralist anthropologist, and his central claim was that Western observers — including Indian sociologists trained in Western categories — could not understand caste, because they insisted on seeing a distribution of power where there was a system of ideas.

The argument he provoked is one of the best in the discipline, and it is not really about India. It is about whether a society can be understood in its own terms, and what happens when the analyst tries.

A question that sounds simple and is not.

Ask an Indian villager in 1960 who is at the top of the local hierarchy, and you may get two different answers depending on how you ask.

Ask about ritual status, and the answer is the Brahmin — even if he is poor, dependent, and owns no land.

Ask who actually decides things, and the answer is the dominant caste (6.2.1) — the landowning cultivators, who have the numbers, the fields and the political connections, and who are ritually below the Brahmin.

Now: which answer describes the hierarchy?

The obvious Western response is that the second is the real one and the first is ideology — a mystification concealing the true distribution of power. Brahmins have status because they legitimate an order that benefits landowners; the ritual is decoration.

Dumont's response is that this is precisely the error, and that making it means you have not understood what you are looking at.

His claim: in the Indian system, status genuinely encompasses power. It is not a mask over it. The Brahmin really is superior, in the terms of the system, and the fact that he is poorer and less powerful is not a paradox to be resolved. It is the structure.

And his diagnosis of why Westerners cannot see this: because they are individualists, for whom the only real relations are between individuals and the only real hierarchy is one of power. A society organised around a holistic idea — in which the whole precedes the parts and each part has value only by its place in the whole — is unintelligible to them.

The methodological problem Dumont poses is genuine, and it has an uncomfortable structure.

To understand another society, you must use concepts. Your concepts came from somewhere — from your own society, its history and its categories.

And if the other society is organised on a different principle, your concepts will systematically misdescribe it — not by getting details wrong but by reading its central features as versions of your own.

Dumont's specific claim: modern Western societies are organised around individualism and equality as fundamental values. Traditional India is organised around holism and hierarchy. These are not more and less of the same thing. They are different ideological orders.

So when a Western observer sees caste, they see: an unjust distribution of resources justified by superstition; individuals unfairly ranked; a system that ought to be understood in terms of exploitation.

And Dumont's argument is that this is a translation into the observer's own categories, and that what has been lost in translation is the object of study.

His method: proceed by contrast. Use the modern West as the explicit comparison — not as a norm but as the observer's own case, made visible so that its assumptions stop operating invisibly.

Which is, note, a serious methodological proposal. It is Bourdieu's participant objectivation (5.8.1) and Harding's strong objectivity (5.6.1) applied to a whole civilisation. Whether Dumont carried it out well is the argument of this lesson.

The argument

Dumont's core claim: caste is a system of ideas and values, not of power or economics, and its organising principle is the opposition of the pure and the impure.

Hierarchy — his technical term — is not stratification. It is "the principle by which the elements of a whole are ranked in relation to the whole."

And the key move: encompassment of the contrary. The superior term does not merely rank above the inferior; it includes it. Dumont's illustration is the biblical Eve, drawn from Adam's rib: man encompasses both the undifferentiated human and the male half of the differentiated pair. The superior contains and defines the inferior.

Applied to caste: the Brahmin encompasses the whole; the varnas are ranked by relative purity; and each level is defined by its relation to the whole rather than by its own attributes.

Purity and impurity are the single principle. Occupations that deal with death, bodily waste, leather, birth and blood are impure. Those who perform them are permanently so. The hierarchy is a gradient of purity, and everything — commensality, marriage, touch, temple access, occupation — follows from it.

And the crucial structural claim: the separation of status from power.

Dumont argues that in the Indian system, the priestly function was separated from the kingly function — and that status was made superior to power. The Brahmin ranks above the Kshatriya; the priest above the king; the pure above the powerful.

In the modern West, by contrast, power and status are fused in the individual and in wealth, which is why Westerners cannot conceive of a status hierarchy that is not a power hierarchy.

The criticisms

This is where the lesson earns its place, because the response to Dumont was substantial, sustained and largely successful.

One: it is the book view in structuralist dress (6.2.1).

Dumont's sources are heavily textual and Brahminical. The system he describes is the one the Brahminical texts describe, and his account of its principle is the account those texts give. He has taken one party's ideology as the structure of the whole.

Gerald Berreman's critique is the sharpest: Dumont describes caste "from the top down," and if you ask those at the bottom, you get a different account. Berreman's Himalayan fieldwork found that low-caste informants did not accept the purity hierarchy as legitimate; they experienced caste as exploitation and said so, in private.

Which connects directly to Scott's hidden transcripts (5.5.1): the appearance of consensus in the presence of the powerful is not evidence of consensus.

Two: it makes ideology the explanation.

If caste is a system of ideas, what maintains it? Dumont's account has no mechanism of enforcement — no violence, no economic dependence, no sanction. And caste is enforced, continuously and violently, which is documented in the atrocity statistics (6.2.1) and in every account of what happens when a boundary is crossed.

Three: it cannot see change.

A synchronic structural account of an ideological order (5.7.1's problem) has no way to register that caste has changed enormously, that Sanskritisation occurs, that dominant castes rise, that reform movements exist, or that Dalit politics has transformed the system.

Four: the West/India contrast is too clean.

Both terms are essentialised. "The modern West" as pure individualism ignores its own hierarchies of race, class and gender; "traditional India" as pure holism ignores its long traditions of dissent — Buddhism, the Bhakti movements, the Lingayats, Sikhism, and the anti-caste thinkers from Phule to Periyar to Ambedkar.

Dumont's structure requires that Indians be holists, and the moment one attends to those traditions the requirement fails.

And five, the political criticism: an account in which caste is a coherent value system that its members accept has been used, and can be used, to explain away demands for change as Western impositions on an authentic tradition. Dumont did not intend this and the reading is available in the text.

The most consequential response to Dumont did not come from anthropology. It came from those the system placed at the bottom.

The tradition is older than the discipline's engagement with it.

Jyotirao Phule (1827–1890) — from a Mali (gardener) caste, founded the Satyashodhak Samaj (Truth-Seekers' Society) in 1873, and with Savitribai Phule opened schools for girls and for lower-caste children from 1848. His Gulamgiri (Slavery, 1873) is explicitly dedicated to the American abolitionists, and it reads caste as a system of conquest and subjugation dressed as religion — a full generation before Ambedkar.

Periyar E. V. Ramasamy (1879–1973) — the Self-Respect Movement in Tamil Nadu, which attacked Brahminical authority directly, promoted self-respect marriages without priests, and reshaped Tamil politics permanently.

And the scholarship since Ambedkar has made two arguments that bear directly on Dumont.

The first: the purity/pollution principle looks like an organising ideology from above and like a technology of exclusion from below. Dalit accounts describe caste as a system of labour extraction, land denial and violence, in which purity ideology is the justification rather than the structure — which is precisely the reading Dumont rejects, made by the people the system is applied to.

The second: Dumont's system has no place for the excluded. The untouchable is outside the varna scheme entirely, which in a holistic account is a puzzle. Dumont handles this by making the impure the necessary complement of the pure — the untouchable's impurity is what makes Brahminical purity possible.

Structurally elegant, and it describes the Dalit as a function of someone else's status. Dalit scholarship's response is that this is exactly the operation being analysed: a framework in which the excluded exist to complete the picture of those who exclude them.

And the strongest empirical point: if Dumont were right that the hierarchy is a shared value order, one would expect those at the bottom to affirm it. Berreman found they do not; Dalit movements from Phule onwards have not; and the Bhakti, Lingayat and Buddhist traditions did not. Consensus is what the system looks like when you ask only those who benefit.

What survives

Dumont was substantially wrong and the book is still worth reading, which is a combination this course has met before (4.2.6, 4.3.7).

What survives:

The methodological warning is correct. Concepts carry the assumptions of the society that produced them, and applying them to another society without examining them produces systematic misdescription. Dumont's insistence on making the observer's own case explicit is the right instruction, even though he executed it by essentialising both terms.

The status/power distinction is real and useful. That ritual rank and secular power can diverge is well established (Srinivas's dominant caste, 6.2.1; Weber's class and status, 4.3.6). Dumont overstated it into a total separation; the divergence itself is a genuine finding.

And the concept of encompassment is analytically productive and has been used well beyond India — for the analysis of any hierarchy in which the superior term claims to represent the whole. "Man" including humanity; the majority as "the nation"; the standard case as the unmarked general case. That is a good concept.

What does not survive: the claim that caste is a system of ideas rather than of power; the treatment of Brahminical texts as the structure of the whole; the essentialised West/India contrast; and the absence of any account of enforcement or change.

And the general lesson is the one 6.2.1 stated: the book view is not merely an error but a systematic product — and it can be committed with enormous theoretical sophistication.

Go deeper

The other Indian traditions

Dumont's holism claim is defeated most simply by pointing at the Indian traditions that rejected caste from within.

Buddhism and Jainism (from the fifth century BCE) both rejected varna as a basis for spiritual standing. The Buddha's position — that a person is a Brahmin by conduct rather than birth — is stated repeatedly in the Pali canon.

The Bhakti movements (roughly the seventh to seventeenth centuries) produced saint-poets from low and untouchable castes whose devotional poetry is central to Indian literature: Kabir, a weaver; Ravidas, a leatherworker; Nandanar, a Dalit; Tukaram, a shudra; Mirabai, a woman who left a royal household. Their argument was that devotion supersedes birth, and their poetry is still sung.

The Lingayat / Virashaiva movement (twelfth century, Karnataka), under Basava, explicitly rejected caste, Brahminical ritual and temple worship, permitted inter-caste marriage and admitted women, and constituted a durable alternative community.

Sikhism (from the fifteenth century) rejected caste doctrinally and instituted the langar — the common kitchen in which all sit together and eat, which is a direct assault on commensality rules.

And the modern anti-caste tradition: Phule, Periyar, Ambedkar, and the movements they built.

Caste beyond Hinduism, and beyond India

Among Indian Muslims: the ashraf/ajlaf/arzal distinction, endogamous biradari groups, and documented discrimination — despite Islamic doctrine's rejection of hereditary hierarchy.

Among Indian Christians: separate seating, separate cemeteries in some places, and Dalit Christian movements campaigning for the reservations denied to them on the ground that Christianity has no caste.

Among Sikhs: caste-based gurdwaras and marriage endogamy, despite the doctrinal position.

And in the diaspora: caste discrimination has been documented in Britain, the United States, Canada and Australia, and has produced legal disputes — including the Equality Act 2010 provisions in Britain and litigation in California. The category travelled with the people, without the religion and without the state that enumerated it.

Which is the strongest evidence available for Ambedkar's mechanism over Dumont's: the ideology can be repudiated, the state removed, and the country left, and endogamy carries the system anyway.

And the general point for Part 6

"Dumont said caste is about religion." He said it is a system of values, organised by the pure/impure opposition, in which status encompasses power.

"Hierarchy means stratification." Dumont's technical usage is ranking by relation to the whole on a principle of value. He insists it is not a distribution of resources or a chain of command.

"Encompassment is a fancy word for domination." It is a specific structural relation in which the superior term includes the inferior. It is a genuinely useful concept and it is not synonymous with power.

"Dumont ignored power." He argued that power exists and is encompassed by status in the Indian system. The criticism is that this makes the ideology the structure and leaves enforcement unexplained.

"India was traditionally holistic." Buddhism, Jainism, Bhakti, Lingayats, Sikhism and the modern anti-caste movements are all internal rejections of caste, spanning two and a half millennia.

"Dumont's book is discredited, so it can be skipped." Its methodological warning is right, its status/power distinction is real, and encompassment is a good concept. Wrong in its central claim and worth reading, which is the normal condition of important books.

Why it matters

It states the hardest methodological problem in Part 6. To understand another society you need concepts; your concepts came from somewhere; and using them may reproduce your own society's categories as universal. Dumont saw the problem clearly, proposed the right remedy — make the observer's own case explicit — and failed anyway, which is more instructive than not seeing it.

It supplies encompassment, which is a genuinely good concept with reach far beyond caste: any hierarchy in which the superior term claims to represent the whole, and the inferior exists as its complement.

It demonstrates that the book view can be committed at the highest level of sophistication. Dumont was not careless. He had fieldwork, theory, self-awareness and enormous learning — and he produced a Brahminical account of caste that a generation took as the structure of Indian society.

And it sets the terms for the rest of Part 6. If the most sophisticated European account of an Asian society reproduced that society's dominant ideology as its structure, then the question is not how to be more careful. It is whether the categories can be built differently, and by whom — which is 6.8.

Dumont's Homo Hierarchicus (1966): caste is a system of ideas and values, organised by the pure/impure opposition, in which status encompasses power — the Brahmin genuinely superior to the king, which Westerners cannot see because they fuse status and power in the individual.

Holism (the whole precedes the parts) against individualism; hierarchy as ranking by relation to the whole, not stratification; and encompassment of the contrary, in which the superior term includes the inferior.

His method was to proceed by explicit contrast with the modern West — which is participant objectivation and strong objectivity applied to a civilisation, and a serious proposal.

The criticisms, largely successful: it is the book view in structuralist dress, taking Brahminical texts as the structure of the whole; Berreman's fieldwork found low-caste informants rejecting the purity hierarchy in private; it has no mechanism of enforcement in a system that is enforced violently; it cannot register change; and both terms of the West/India contrast are essentialised.

The Dalit response: purity ideology looks like structure from above and like justification from below; and a framework in which the untouchable's impurity exists to make Brahminical purity possible describes the excluded as a function of those who exclude them, which is the operation being analysed.

The holism claim is defeated by India's own traditions: Buddhism, Jainism, the Bhakti poets (Kabir, Ravidas, Tukaram), the Lingayats under Basava, Sikhism and the langar, and Phule, Periyar and Ambedkar.

And caste operates among Indian Muslims, Christians and Sikhs, and in the diaspora — which is the strongest evidence for Ambedkar's endogamy mechanism over Dumont's ideology: remove the doctrine, the state and the country, and endogamy carries the system anyway.

What survives: the methodological warning, the status/power distinction, and encompassment as a concept with reach.

Holism / individualism — Dumont's contrasting value systems, in which the whole or the individual is fundamental.

Hierarchy (Dumont) — ranking by relation to the whole on a principle of value, distinguished from stratification.

Encompassment of the contrary — the relation in which the superior term includes the inferior within itself.

Pure / impure — the opposition Dumont identifies as caste's single organising principle.

Separation of status and power — the claim that the priestly function was distinguished from and ranked above the kingly.

View from below — Berreman's methodological correction: the hierarchy looks different when the subordinated are asked in private.

Bhakti — the devotional movements whose low-caste saint-poets held that devotion supersedes birth.

Langar — the Sikh common kitchen, a direct institutional assault on commensality rules.

Ashraf / ajlaf / arzal — status distinctions among Indian Muslims, evidencing caste's operation outside Hinduism.

One — find an encompassment. Locate a case where a superior term claims to represent the whole and defines the subordinate as its complement — "man" for humanity, a majority as "the nation," the standard case as the general one. The concept works everywhere.

Two — ask from below. Take any hierarchy described as consensual and ask whether those at its bottom have been asked, in conditions where an honest answer is safe. Berreman's finding generalises.

Three — test the ideology claim. Take a system said to be held together by shared belief and ask what happens to someone who violates it. If the answer involves sanctions, the belief is not doing the work alone.

Four — check the diaspora. Look up caste discrimination cases in Britain or the United States. The system operating without the religion, the state or the country is a natural experiment on which mechanism is doing the work.

Five — apply Dumont's method to yourself. Write down three values you consider self-evidently correct. Then ask what a society organised on the opposite principle would look like — and whether you could describe it without treating it as a deficient version of your own.