Caste - The Book View and the Field View

Caste is the most studied topic in Indian sociology, one of the largest bodies of empirical social research anywhere, and the clearest available case of the argument in 6.1.1: a system described by outsiders from texts, hardened by administration, contested by those it governed, and still organising the lives of over a billion people.

This course has used the book view / field view distinction repeatedly (3.1.2, 4.3.10, 4.6.3). This lesson is where it gets its full treatment, because it is the single most important methodological lesson Indian sociology has given the discipline.

Two accounts of the same village.

Account one, drawn from the Dharmashastra texts, from Manu, and from the standard European reference works of the nineteenth century:

Indian society is divided into four varnas, in ranked order: Brahmin (priest), Kshatriya (warrior-ruler), Vaishya (merchant-cultivator), Shudra (servant), with a fifth category outside the scheme entirely — the untouchables, later Dalits. Each varna has prescribed occupations and duties (dharma). The order is sanctioned by scripture, justified by karma, hereditary and unchangeable, and enforced by rules of purity and pollution. Occupation follows birth. Marriage is within the group. The system is ancient, stable and uniform across the subcontinent.

Account two, from a village in Mysore in the late 1940s, recorded by a sociologist who lived there:

There are around twenty groups, called jatis, not four. Nobody can place several of them confidently in the varna scheme, and different informants rank them differently. The dominant group locally is not Brahmin; it is a cultivating caste with land and numbers. Several groups have changed their claimed status within living memory. A number of families are doing work their caste does not traditionally do. There are disputes about precedence that have been running for generations, and they matter enormously.

Both accounts describe caste. They are not describing the same thing.

M. N. Srinivas, who conducted the second, gave the distinction its name — and his point was not that the texts are wrong. It is that the texts describe a normative model, and that a normative model is not a description of practice.

The methodological problem, stated generally, because it is not confined to India.

When a society has an authoritative textual tradition describing itself, an outside observer faces a choice — and frequently does not notice they are making one.

The book is available. It is written, systematic, portable, translatable, and it presents itself as the account of how things are.

The field is not. It requires language, time, residence, access and patience, and what it yields is messy, local, contradictory and hard to generalise.

And there are structural reasons why outsiders take the book.

It is easier. It can be read in a library in another country.

It is authorised. It comes from those the society itself treats as the custodians of its knowledge — who are, invariably, a particular group with a particular position.

And it fits the administrative need (6.1.1). A state needs categories that are exhaustive, exclusive and uniform. The text supplies four; the field supplies four thousand.

So: the book view is not simply an error. It is the systematic product of a relation between an outside observer, an available authoritative text, and an administrative purpose.

And its content is not neutral. The Dharmashastra account is a Brahminical account — the description a particular group gives of a hierarchy in which it stands at the top. Taking it as the description of Indian society is taking one party's account of a dispute as the account of the dispute.

The basic distinction

Varna — the fourfold (plus one) scheme of the classical texts. Pan-Indian, textual, and largely a framework of reference rather than of practice.

Jati — the actual endogamous groups in which caste is lived. There are thousands — the 1931 census recorded several thousand, and the count depends entirely on how one counts. Jatis are regional: a jati significant in Tamil Nadu may be unknown in Punjab.

The relationship between them is the crux. Varna is not a list of which jatis exist. It is a scheme jatis are placed in — and the placement is contested, variable and consequential. Groups claim higher varna status; others dispute the claim; and the dispute is the social process.

Srinivas's formulation: varna gives "a common social language which holds good, or is thought to hold good, for India as a whole" — which is precisely why it travels and why it misleads.

Endogamy — marriage within the group. This is the defining feature, and it is Ambedkar's analytical starting point (4.6.3, 6.2.2): a caste is "an enclosed class," and endogamy is the enclosure.

Commensality rules — who may accept food and water from whom, and of what kind. Historically the most sensitive marker in daily life, and the one that made caste visible constantly.

Hereditary occupation — real as a tendency, substantially overstated as a rule, and far weaker in practice than the textual model implies.

Purity and pollution — a graded scheme in which contact, food, occupation and bodily substances carry ritual status. This is Dumont's central concept (6.2.3) and the most contested element.

And a jajmani relationship in many villages: a system of hereditary exchange in which service castes provide goods and labour to landholding families in return for grain at harvest. Studied extensively, romanticised in some accounts as reciprocal, and shown in others to be a relation of dependence and extraction.

Srinivas and the field view

M. N. Srinivas (1916–1999) is the most important figure in Indian sociology, and his contribution is principally methodological: the insistence that caste be studied where it happens.

Religion and Society among the Coorgs of South India (1952) and his village studies produced three concepts that reorganised the field.

Sanskritisation — "the process by which a 'low' Hindu caste, or tribal or other group, changes its customs, ritual, ideology and way of life in the direction of a high, and frequently 'twice-born', caste."

Its significance: it demonstrates mobility in a system described as immobile. Groups adopt vegetarianism, teetotalism, Brahminical ritual, and new origin myths, and over generations claim and sometimes secure a higher position.

And Srinivas's crucial qualification: Sanskritisation produces positional change, not structural change. One group rises; the hierarchy remains. It is mobility within a system that is thereby confirmed rather than challenged, which is why Dalit movements have generally rejected it as a strategy (6.2.2).

Dominant caste — the group that dominates a locality by virtue of numerical strength, land ownership, and political power, regardless of its ritual position. Srinivas's finding was that in most Indian villages, the locally dominant caste is not Brahmin but a substantial landowning cultivating caste.

This severs ritual rank from actual power, and it is a direct empirical refutation of any model in which Brahminical status determines everything.

Westernisation — the parallel process of adopting Western education, dress, occupation and values, which frequently competes with Sanskritisation as a mobility strategy and has different implications.

The field view is best understood through what it produced, so take the concrete findings.

Village studies from the 1950s and 60s — Srinivas, S. C. Dube, Adrian Mayer, F. G. Bailey, McKim Marriott, Oscar Lewis, Kathleen Gough — established, collectively, a set of facts that the book view could not accommodate.

Local hierarchies do not match the varna order. Rankings differ between villages fifty miles apart, and informants within a village disagree — particularly about the middle ranges, where most groups are.

Ritual rank and secular power diverge routinely. Poor Brahmins depend on wealthy Shudra landowners; the ritually pure are frequently not the powerful.

Occupational mobility is common at the margins, and the "traditional occupation" of a caste is frequently practised by a minority of its members.

Dispute over precedence is constant. Caste is not a settled hierarchy people occupy; it is an ongoing argument about ranking, conducted through ritual claims, marriage strategies, food rules, temple entry and, increasingly, politics.

And the system varies enormously by region. The North Indian pattern differs substantially from the South; Bengal, Kerala, Punjab and Tamil Nadu each present distinct configurations; and Kerala's history — with its own extremely severe pollution rules and its subsequent radical reform movements — fits no general model.

Louis Dumont's later criticism of the village studies is worth noting: he argued that they had produced excellent local detail and lost the system, and that focusing on villages obscured the pan-Indian ideological structure. That objection is the substance of 6.2.3.

And the honest assessment of the field view's own limits: the village studies were conducted overwhelmingly by upper-caste researchers, frequently living in the upper-caste part of the settlement, with the Dalit hamlet — usually physically separate — studied less closely. The field view corrected the book view and carried its own standpoint, which Dalit scholarship has since documented at length (6.2.2).

The colonial hardening

Treated in 6.1.1 and restated here because it is the bridge between the two views.

The census required singular, exhaustive, uniform categories across a subcontinent, and officials drew on Brahminical texts because those were written, systematic and available.

Risley's 1901 census attempted an all-India ranking by "social precedence" and used anthropometric measurement on a theory of racial origin.

The effects: caste associations formed to lobby for reclassification; identities that had been local became all-India; the categories hardened through decennial repetition; and caste became a currency in which claims on the state were made.

Dirks's Castes of Mind (2001) argues that caste as a unified all-India system of identity is substantially a colonial construction. Critics — including Susan Bayly — argue this overstates: pre-colonial states also classified, ranked and intervened in caste, and the eighteenth century saw considerable state-driven caste formation before the British.

The defensible position, and it is the one this course takes: caste is old, varied and real; the specific form it took by 1947 — enumerated, ranked, textually authorised, all-India and politically mobilised — bears a heavy colonial imprint; and neither "ancient unchanging system" nor "colonial invention" is accurate.

Go deeper

Caste after independence

The system has changed enormously since 1947, and in directions that no theory predicted well.

Constitutionally: untouchability was abolished by Article 17 of the Constitution — drafted under Ambedkar's chairmanship (6.2.2) — and reservations (quotas) established in legislatures, public employment and education for Scheduled Castes and Scheduled Tribes, extended in 1990 to Other Backward Classes following the Mandal Commission.

Politically: the most significant development. Caste has become the principal idiom of Indian democratic mobilisation. Rajni Kothari's formulation is the standard one: politics did not destroy caste; caste and politics transformed each other. Caste associations became vote banks; parties formed around caste coalitions; and the "silent revolution" (Christophe Jaffrelot's term) brought lower-caste politicians to power in several states.

Economically: liberalisation has produced substantial change in some sectors and very little in others. The evidence on caste discrimination in labour markets is strong — audit studies sending matched applications with different caste-marked names find substantial callback differences, replicating the design used for race in the US and UK.

In marriage: change is slow. Endogamy remains the most persistent feature. Survey evidence consistently finds that a large majority of marriages remain within caste, and inter-caste marriage — particularly involving Dalits — continues to provoke violence, including the killings euphemistically called "honour killings."

And in violence: the Scheduled Castes and Scheduled Tribes (Prevention of Atrocities) Act 1989 exists because atrocities are frequent and documented. Reported cases number in the tens of thousands annually, with substantial under-reporting acknowledged by researchers and by the state's own bodies.

Two things must be held together here, and simplifications in either direction are wrong.

Enormous change has occurred. Political representation, urbanisation, education, anonymity, legal protection and economic mobility have altered caste's operation profoundly. The village hierarchy Srinivas described is not the structure of urban India.

And caste has not disappeared; it has changed form. Endogamy persists. Occupational concentration persists at both ends — sanitation work remains overwhelmingly Dalit; elite professional and corporate positions remain overwhelmingly upper caste. Networks, credentials and cultural capital (5.8.1) carry caste into modern institutions without any explicit rule.

The academic dispute is about weighting, and it is heated. Some argue that caste is dissolving into class in urban India; others that it has been reconstituted in modern institutional forms. The data supports both partially, and which one is emphasised frequently tracks the writer's own position — which is itself a finding.

What Indian sociology gave the discipline

Beyond the substantive material, four contributions that travel.

The book view / field view distinction — the most useful, and applicable to any society with an authoritative self-description.

Sanskritisation — a general model of emulative mobility: a subordinate group adopting a dominant group's practices to claim status, producing positional change while confirming the hierarchy. The mechanism occurs everywhere, and it is what 4.3.6's status analysis needed.

Dominant caste — the analytical separation of ritual rank from secular power, which is Weber's class/status distinction (4.3.6) demonstrated empirically in a system supposedly organised by ritual alone.

And the study of endogamous closure — Ambedkar's contribution (6.2.2), which is Parkin's social closure with a specified mechanism, forty years earlier.

And a note on comparison

Caste is frequently compared with race, and the comparison has a long history: Ambedkar corresponded with Du Bois; Oliver Cromwell Cox argued at length in Caste, Class and Race (1948) that the two are fundamentally different; Gerald Berreman argued they are usefully comparable; and Isabel Wilkerson's Caste (2020) revived the comparison for a general audience.

The arguments against comparison: caste is not phenotypic; it carries a religious sanction race does not; its internal gradations are far finer; and mapping it onto race imports American categories.

The arguments for: both are systems of hereditary, endogamous, hierarchically ranked groups with unequal access to resources, enforced by violence and legitimated by an ideology of essential difference. Ambedkar and Du Bois each recognised the other's situation immediately, which is evidence of something.

The position this course takes: the comparison is analytically productive and must not be allowed to flatten either. What travels is the mechanism — endogamous closure, hereditary transmission, ideological legitimation, and violence at the boundary — not the content.

And Indian caste and American race are not the only cases: Japanese Burakumin, Korean Baekjeong, Somali occupational castes, West African griot and blacksmith castes, and European Roma exclusion all present variants. A general sociology of hereditary closure is possible and has barely been written.

"Caste means the four varnas." Varna is a textual scheme of reference. Caste is lived in thousands of regional jatis whose placement in the varna scheme is contested.

"Caste is fixed and unchanging." Sanskritisation, dominant caste, occupational mobility and constant disputes over precedence were all documented by the first generation of village studies.

"The book view is simply wrong." It accurately describes a normative model — a Brahminical one. The error is taking a normative model for a description of practice, and one party's account for the account.

"The British invented caste." Overstated. Caste is old and real; the enumerated, ranked, all-India form is substantially a colonial product, which is a different claim.

"Sanskritisation is resistance." It produces positional change without structural change, which is why Dalit movements have generally rejected it.

"Caste is disappearing with modernisation." Endogamy persists at high rates, occupational concentration persists at both ends, audit studies find labour market discrimination, and atrocity cases number in the tens of thousands annually. Its form has changed a great deal; its existence has not.

Why it matters

It supplies the discipline's best methodological warning. Any society with an authoritative textual self-description will offer the outside observer a book view — and the book will have an author with a position. Ask whose account it is before treating it as the account.

It demonstrates that the field corrects the book — and carries its own standpoint. The village studies overturned the textual model and were themselves conducted from a position, with the Dalit hamlet studied least. There is no view from nowhere, and the correction of one standpoint by another is how a field actually advances.

It gives three concepts that travel: emulative mobility that confirms the hierarchy it climbs; the separation of ritual rank from secular power; and endogamous closure as a general mechanism.

And it establishes what Part 6 is for. This is not an exotic topic appended to a general sociology. It is one of the largest bodies of empirical social research anywhere, it produced concepts the discipline needs, and it was substantially absent from the frameworks of Parts 4 and 5 — while Weber's textual account of caste, produced from a library by a man who never visited India, was the standard reference for most of a century.

The book view is drawn from the Dharmashastra texts: four varnas plus a fifth excluded category, hereditary, scripturally sanctioned, uniform and unchanging. The field view is what village studies found: thousands of regional jatis, contested rankings, dominant castes that are not Brahmin, occupational mobility, and constant dispute about precedence.

The book view is not simply an error. It is the product of a relation between an outside observer, an available authoritative text, and an administrative need for exhaustive uniform categories — and its content is one party's account of a hierarchy in which that party stands at the top.

Srinivas's three concepts: Sanskritisation — emulation of higher-caste practice producing positional change without structural change; dominant caste — local power resting on numbers, land and politics rather than ritual rank, which severs status from power; and Westernisation as a competing strategy.

The colonial hardening: the census requiring singular categories, Risley's 1901 ranking, caste associations lobbying for reclassification, and local identities becoming all-India. Neither "ancient unchanging system" nor "colonial invention" is accurate.

Since 1947: constitutional abolition of untouchability and reservations for SC, ST and OBC; caste as the principal idiom of democratic mobilisation (Kothari: caste and politics transformed each other); documented labour market discrimination in audit studies; persistent endogamy; and atrocity cases in the tens of thousands annually. Enormous change in form; no disappearance.

And the field view carried its own standpoint — conducted largely by upper-caste researchers, with the Dalit hamlet studied least, which is what 6.2.2 takes up.

What travels: the book view / field view distinction; emulative mobility; the separation of ritual and secular power; and endogamous closure, which makes a general sociology of hereditary exclusion possible — Burakumin, Baekjeong, Roma, West African occupational castes — and it has barely been written.

Varna — the classical fourfold textual scheme, plus the excluded fifth category.

Jati — the actual endogamous, regional groups in which caste is lived; thousands of them.

Book view / field view — the normative textual model versus observed practice.

Sanskritisation — a lower group's adoption of higher-caste practice to claim status.

Positional versus structural change — mobility within a hierarchy versus alteration of the hierarchy.

Dominant caste — the locally powerful group, by numbers, land and politics, irrespective of ritual rank.

Westernisation — the competing mobility strategy of Western education, occupation and values.

Jajmani — hereditary service-and-payment relationships between castes in a village economy.

Commensality — rules governing the acceptance of food and water between groups.

Scheduled Castes / Scheduled Tribes / Other Backward Classes — the constitutional categories governing reservations.

Endogamous closure — the reproduction of a group through marriage restriction; caste's defining mechanism.

One — find a book view in your own society. Take an official or scriptural self-description — a constitution, a professional code, a national story — and compare it with observed practice. Ask who wrote the book and where they stood.

Two — look for Sanskritisation. Find a group emulating the practices of one above it in order to claim status. Ask whether the emulation challenges the hierarchy or confirms it. It almost always confirms it.

Three — separate rank from power. In any institution, identify who has formal status and who actually decides. Srinivas's dominant caste finding is a general one.

Four — check an audit study. Look up the caste or race audit-study literature — matched applications, differing names. The design is simple and the findings are among the cleanest evidence of discrimination available.

Five — test the closure model. Take any hereditary, endogamous, ranked group system — caste, Burakumin, Roma, aristocracy — and ask what maintains the boundary. Marriage rules will be the answer in every case.