Caste is the first fact of social life in India and the first problem that any imported theory of society meets there. It assigns a newborn child a rank, a range of permissible occupations, a set of people who may be married and a set of people whose touch, food or shadow may be treated as polluting, and it does all of this by birth alone, before the child has acted, earned, chosen or believed anything at all. A reader who has lived inside that order knows its weight from within, in the surnames that announce it, the marriage negotiations that enforce it and the occupations that still follow its lines. A reader meeting it from outside should resist the temptation to treat it as a colourful survival, a piece of the past lingering into a modern economy. It is better understood as a working system of stratification that has organised labour, property, honour and violence in the subcontinent for a very long time, and that has adapted itself to every economic formation it has passed through, from agrarian kingdoms to a commercial republic with stock exchanges and software parks. The question this article examines is narrower than the system itself and harder than it first appears. It is the question of how caste stands in relation to class, posed inside the tradition that made class its master category: the Marxist tradition in India.

That question can be put in three forms, and the debate consists of the attempts to answer each of them. The first form is analytical: is caste a kind of class relation, wearing Indian dress, so that an account of who owns land and who works it also accounts for who eats with whom and who may draw water from which well? The second form is causal: if caste is not a class relation, is it an effect of one, a body of ideas and customs generated by an older economy and destined to fade as that economy is replaced? The third form is practical, and it is the form in which the question was first pressed with real force: can a politics organised around class, demanding solidarity among all who labour, succeed in a society where labourers are themselves divided into ranked and mutually exclusive orders, each taught to regard the one below it with contempt? Indian Marxists have given different answers across a century of argument, and the answers matter beyond the seminar room, because each one implies a different strategy for anyone who wants the hierarchy abolished rather than merely described. The sharpest statement of the question came not from inside the movement but from the man who became its most formidable Indian critic, Bhimrao Ramji Ambedkar, in an address written in 1936 that its invited audience refused to hear. His charge, quoted in this article from the address itself, was that the movement had misunderstood the very society it proposed to transform, and that until it took account of caste before the revolution, it would be forced to reckon with caste afterward, on far worse terms.

What caste is: the object the debate is about

Precision about the object has to come before any verdict about it, because much of the confusion in this debate is confusion about what caste actually does. Caste is a system of hereditary, ranked groups. Membership is acquired by birth and cannot be resigned, purchased or earned away within the classical rules of the system. The groups are endogamous, meaning that marriage is confined within the group and policed, in many communities with severe sanctions, so that the boundary of each group is reproduced biologically and socially in every generation. They are ranked, and the ranking is not a loose sense of prestige but a graded order, in which each group stands above some and below others, and knows its place in the sequence with great exactness. Occupation is attached to rank by custom: particular crafts, trades and services belong to particular groups, and the dirtiest and most degrading work, the removal of night soil, the handling of dead animals, the cleaning of sewers and latrines, was assigned to the groups placed at the very bottom, whose members were held to be polluted by the work and then held responsible for the pollution.

Untouchability is the extreme form of that logic, and it should be described without softening. The communities placed outside the fourfold varna order were treated as permanently impure: their touch was held to defile, their entry into temples was barred, their access to public wells and roads was restricted, and in parts of the country even their shadow or their approach within a measured distance was treated as contamination. These were not metaphors. They were enforced rules of daily life, backed by social boycott and by violence when boycott failed, and their victims were denied education, property and dignity as a matter of routine rather than of occasional cruelty. Any theory of caste that cannot look at that enforcement steadily, at the wells guarded and the children humiliated and the labourers beaten for asserting a right, has failed before it begins, and one of the standing indictments in this debate is that economic categories, applied carelessly, allowed educated men to talk about everything except the humiliation at the centre of the system.

What is graded inequality?

Ambedkar’s phrase for the ranking was graded inequality, and the grading is what separates caste from a simple two-storey division between high and low. In his analysis the order is an ascending scale of reverence and a descending scale of contempt, in which every group except the highest has someone beneath it whose subordination consoles it for its own.

Every group except the lowest has someone above it whose privileges it resents without joining those below to resist. A hierarchy arranged as a staircase rather than as a wall gives almost everyone a step to defend. That single observation explains a great deal of the system’s stability, because it dissolves the natural solidarity of the subordinated: the group on the second step from the bottom guards its distance from the bottom more jealously than it contests the distance above it.

Two further features complete the object. The first is that the ranking carries a religious warrant. In the classical justification, caste is not merely customary but sacred, ordained in scripture, enforced by doctrines of karma and rebirth that present a person’s station as the earned consequence of conduct in earlier lives, and guarded by texts whose authority the believer is forbidden to question. A social arrangement defended as divine law is a different opponent from one defended as convenient, because attacking it requires attacking the scripture that sanctifies it, an act with costs that purely political criticism never has to pay. The second feature is that caste is lived as community: it supplies marriage partners, credit, help in sickness, a network for finding work and a congregation for worship. It is at once a chain and a safety net, and any honest account has to hold both truths together, since the same endogamous network that excludes the outsider also insures the insider, which is one reason the system’s victims have sometimes defended the fragment of it that feeds them while hating the whole of it that degrades them.

Untouchability as the touchstone every theory must face

Every general claim in this debate should be tested against the institution at the order’s floor, because untouchability is where caste’s logic operates without the softening that commerce, ambiguity or distance provide higher up the scale. The communities subjected to it lived, by rule, outside the village’s inhabited core, in quarters whose very names in the regional languages announced their exclusion, and drew their water from separate wells or at separate hours, entered temples not at all, and buried their dead in separate grounds. Their hereditary occupations were the ones the order defined as polluting: the removal of human waste, the disposal of dead cattle and the working of their hides, the sweeping of streets and the carrying of night soil, labour indispensable to the village’s cleanliness and rewarded with the doctrine that those who performed it were themselves filth. The enforcement, when the subjected resisted the assignment, ran from boycott through fines imposed by caste councils to violence of a deliberateness that the historical record documents across every region and century the debate touches: the burned houses, the punished processions, the penalties for the crime of wearing shoes, riding a horse at a wedding or drawing water in defiance, each item a tuition fee in the order’s real curriculum.

Untouchability is the touchstone because each position in the debate handles it with a different degree of strain, and the strain is measurable. The superstructure doctrine strains most: an ideology, however tenacious, does not explain why the scavenger’s grandson is still summoned to the drain in the age of the municipal contract, nor why the summons is obeyed by men who despise it and enforced by men who profess to deplore it. The form-of-class position handles the economics of the institution, the hereditary assignment of degrading labour as a labour regime, and strains at its ritual surplus, the doctrine of the polluting touch, which allocates no surplus and organises no production and yet is enforced with the system’s full violence. The autonomy position handles the ritual surplus natively and strains at the institution’s material convenience for everyone above the floor, who receive, in return for the doctrine, a permanently cheap and permanently available workforce for the work nobody else will do. The mutual constitution thesis is, in large part, the refusal to choose between the labour regime and the doctrine, since the historical institution is visibly both at once, a jobs programme for the dominant and a theology for the dominated, each face of it keeping the other in repair. A reader who wants the debate in one test can do no better than to carry any proposed theory of caste to the village drain and ask it to explain, without remainder, why the same families are still being lowered into it.

A note on the words: varna, jati and the Portuguese import

The vocabulary of the subject is itself one of its traps, and a section on terms will repay the reader before the argument deepens. The word caste is not Indian at all: it derives from the Portuguese casta, meaning lineage or breed, applied by the first European traders to the ranked birth groups they encountered and naturalised thereafter in every European language. The Indian words it translates are two, and they do not mean the same thing. Varna is the textual scheme, the fourfold order of priests, warriors, traders and servants, with the untouchable communities placed beneath and outside it, set out in the law books and the creation hymns and known to every schoolchild as the system’s official diagram. Jati is the lived reality: the thousands of named birth groups, localised, occupationally tinged and ranked against their immediate neighbours, into which Indian society is actually sorted and within which marriages are actually arranged. Every Indian belongs to a jati; the varna scheme is the grid into which the jatis are notionally fitted, often so loosely that the same jati will claim one varna in one district and be assigned another in the next.

The distinction matters to the debate because the combatants were sometimes describing different objects with the same word. The classical position’s caste, the survival to be dissolved, was imagined largely at the level of varna, a diagram of society that economic change might reasonably be expected to redraw. The address’s caste was jati in its lived operation, the marriage circle, the hereditary calling and the graded etiquette of particular villages and towns, an object whose reproduction no change in the diagram necessarily touches. Scholars who stress the colonial census in the system’s history add a third warning: the decennial enumerations of the British period, which required every subject to be filed under a single caste name and ranked the names in order of precedence, pressed the fluid and locally variable jati order into a harder and more uniform mould, so that part of what modern India inherited as ancient caste was in fact standardised by the filing cabinets of a modern state. The point is contested among historians and should not be inflated into a claim that the colonisers invented the hierarchy, which the precolonial record refutes. Its disciplined version is enough for the debate: the object the theories claim to explain is itself partly a product of the instruments used to count it, and humility about the category is the beginning of theorising it well.

Counting the castes: enumeration as an actor in the debate

The debate’s object was not merely observed by the modern state; it was, in part, standardised by the state’s favourite instrument, the census, and the standardisation belongs in the record because it changed what the theories had to explain. The colonial enumerations of the late nineteenth and early twentieth centuries required the population to be returned under caste names, attempted in 1901, under the census commissioner H. H. Risley, to arrange those names in an official order of social precedence, and thereby invited every community in India to petition, memorialise and agitate for a higher filing than its neighbours had been granted. The files of those petitions are among the debate’s richest archives, since they show caste operating exactly as the mutual constitution thesis describes: communities spending real resources, organisational and financial, to purchase rank in a classification, because rank in the classification governed access to employment, education and honour in the world the classification administered. The census did not invent the hierarchy it counted. It gave the hierarchy a single national table, a bureaucracy to appeal to and a decennial rhythm of competition, and in doing so it hardened local and negotiable standings into identities of record.

The republic that succeeded the empire made the opposite choice, and the choice is equally instructive. The first census of independent India, taken in 1951, abandoned the general enumeration of caste, recording only the communities listed in the constitution’s schedules for the purposes of its safeguards, on the reasoning, much debated at the time and since, that a state pledged to the hierarchy’s abolition should not keep its ledger. The theoretical consequence was a curious blindness at the centre of the evidence base: for the decades in which the caste and class debate was most active, the Indian state declined to count the object of the debate at the national level, and scholars worked from village studies, sample surveys and the scheduled lists, reconstructing the national picture from fragments while the argument about the picture proceeded at full confidence. The episode is entered here as history and as a caution. The categories in which a society permits itself to be counted decide, in part, what can be known about it, and a debate conducted for half a century over an object its own state refused to enumerate is a debate whose confidence should everywhere be discounted by the reminder that its map was drawn, of necessity, from the territories its travellers happened to visit.

The categories the Marxist tradition brought with it

Marxism arrived in India with a theory of stratification already formed on European material, and the encounter between that theory and caste is the real subject of the debate. In the Marxist account, class is not a rank, an income band or a style of life. It is a relation, defined by a group’s position in the organisation of production: who owns the means of production, who must sell the capacity to work in order to live, who appropriates the surplus that labour creates. Because the definition is relational, class in this sense is treated fully in the series study of the Marxist theory of class, and this article links to that treatment rather than repeating it, since the point here is what happens when a relational category meets a hereditary one.

Alongside the definition of class came a picture of how societies hold together and change, the picture of a material base and a cultural and political superstructure raised upon it. In the compressed version that circulated most widely, the relations of production form the foundation, and law, religion, philosophy and custom form the edifice, explained in the last instance by the foundation they rest on. The fuller and more careful version of that metaphor, with its qualifications and its critics, is set out in the study of base and superstructure, and readers who want the metaphor’s exact wording should go there. For the caste debate, the operative question was always the classification of caste within that picture. Was caste part of the base in India, one of the relations through which production was actually organised, since it decided who ploughed, who wove, who lent and who cleaned? Or was it part of the superstructure, a body of religious ideas and inherited customs floating above the real relations of land and labour, destined to be explained by them and, in time, to be dissolved by their transformation?

Is caste a form of class?

caste and economic position overlap heavily in India, but caste also governs marriage, ritual, and social contact in ways economic change has not dissolved. The classical Marxist position treating caste as a superstructural effect is not supported, and current work treats the two as mutually constituting.

That answer, the considered one, took most of a century to reach, and the sections that follow reconstruct the route. The first position on the route classified caste firmly in the superstructure. The second, reacting against the first, folded caste into class as its Indian form. The third conceded caste an autonomy of its own in whole provinces of life while keeping production in command. The fourth, which the verdict of this article supports, abandons the contest for command altogether and treats the two orders as making each other, continuously, in both directions. Each position had serious holders and serious evidence, and each will be given its strongest form before its difficulties are examined, because a debate of this quality is not honoured by caricatures of the losing sides.

What Marx wrote about India, and what his Indian readers made of it

Behind the classical position stands a text, or rather a handful of newspaper articles, and the debate cannot be understood without them. In 1853, writing for an American newspaper on the British rule in India, Marx described the Indian village community as the solid foundation of an unchanging Asiatic order, and caste and slavery as the conditions that had degraded its inhabitants within it. He judged British rule with characteristic doubleness: the colonial power was an unconscious tool of history, destroying a stagnant social order whose foundations were caste and the self-sufficient village, and laying, through railways, the telegraph and a modern army, the material premises of a different society. The famous expectation followed: modern industry, arising from the railway system, would break up the hereditary division of labour upon which caste rested. The sentence mattered enormously to the debate, because it supplied the classical position with its founder’s own authority: the solvent of caste was to be economic development itself, arriving even by the agency of an empire the movement despised.

Read carefully, the articles of 1853 support the expectation and something more complicated besides. Marx’s village was not merely feudal in the European sense, and he knew it: he described communities that combined agriculture and manufacture in a self-sufficient round, sustained by an unchangeable division of labour that he treated as a law of the society’s reproduction. The description contains, in embryo, the observation the later debate would weaponise, that the Indian division of labour was hereditary in its personnel and not merely functional in its tasks, though Marx drew from it the conclusion of stagnation rather than the conclusion his Indian critic would draw, that the division’s reproduction had to be attacked directly. His later work complicated the picture further. The ethnological reading of his final years, the notebooks on communal forms and on India that scholars have since published, shows a mind returning to the question of pre-capitalist communities with fresh material and less confidence in any single solvent, and Indian scholars have quarried those notebooks for a Marx less committed to the railway prophecy than the Marx of the movement’s pamphlets.

What did Marx predict about caste and industry?

Writing on India in 1853, Marx predicted that modern industry, introduced through the railway system Britain was building, would dissolve the hereditary divisions of labour on which he believed caste rested. The expectation became the classical position’s charter in India: caste was to be ended by economic development as a by-product, needing no separate campaign.

The prediction’s failure across the following century and a half, as industry grew and caste adapted rather than dissolved, is the empirical wound from which the whole subsequent debate developed.

The use each side made of the founder’s India writings is instructive. The classical position cited the prophecy and ignored the notebooks. The critics cited the prophecy’s failure and sometimes ignored that Marx’s subject had been a colonial economy in the first shock of its transformation, not a developed capitalism of the kind whose Indian career the twentieth century would record. The mature scholarship treats the 1853 articles as what they were: brilliant journalism by a thinker working from the reports available in London, whose framework his Indian successors were obliged to test against an object he never saw. A tradition that cannot distinguish its founder’s journalism from its founder’s method will keep failing examinations of this kind, and the caste debate is the Indian record of that distinction being learned the expensive way, across three generations.

The classical position: caste as a survival of the feudal order

The earliest communist writing on India treated caste as a survival. In that treatment, caste belonged to the world of the village community and the feudal landlord, a pre-capitalist arrangement of hereditary occupation and ritual rank that the growth of markets, railways, factories and a mobile labour force would steadily undermine. Capitalism, on this view, was already dissolving caste as a by-product of its ordinary operation, because the mill owner hiring hands does not ask after their ritual status, the railway carriage mixes passengers whom the village would have kept apart, and the money wage pays no respect to hereditary calling. What remained of caste was ideology: a set of beliefs in the minds of the labouring poor, sustained by priests and landlords whose interest it served, and certain to wither as the economic relations that had produced it gave way to new ones. The political conclusion followed directly. A communist party did not need a separate campaign against caste, because the campaign against landlordism and capital was the campaign against the soil in which caste grew. To raise caste as a distinct question was, at best, a distraction from the class struggle and, at worst, a device for dividing workers whom employers and colonial officials were happy to see divided.

There was evidence available for this reading, and the men who held it were not fools. Marx himself, writing on India in the middle of the nineteenth century, had treated caste as one of the decisive impediments to Indian progress and had expected the modern industry that colonialism was importing, however brutally, to dissolve the hereditary divisions of labour on which caste rested. The early Indian communists inherited that expectation with the rest of the doctrine, and the visible facts of the early twentieth century could be arranged to support it. Caste panchayats were losing their coercive grip in the cities. New occupations, from mill work to clerical service to soldiering, recruited across caste lines in ways the village economy never had. Reform movements inside Hindu society were attacking untouchability in the name of religion itself. A theory that predicted the slow erosion of caste by economic change could point to the erosion beginning, and a party short of cadres, hunted by a colonial police state and obliged to choose its fronts, had every organisational reason to accept a theory that told it the caste question would be answered by the struggles it was already fighting.

What did the economic interpretation of history claim about caste?

Applied to India, the economic interpretation of history claimed that the material organisation of production explained the rest of the social order, caste included. Its Indian exponents argued that caste had arisen as a division of functions in an early economy, hardened under feudal relations of land and service, and survived as an ideology of hierarchy whose roots were disappearing.

From that diagnosis the remedy followed of itself: transform the relations of production, and the ideology would lose its foundation, first growing anachronistic and then untenable, as religious justifications of other dead arrangements had done elsewhere.

Underneath the diagnosis sat a picture of the human subject that the critic of 1936 would seize upon. Ambedkar, summarising the socialists he opposed, observed that they propounded an account in which “man is an economic creature”, whose activities and aspirations are bound by economic facts and for whom property is the source of power. Whether or not that summary was fair to Marx, it was a recognisable portrait of the doctrine as it was actually preached in Indian party schools and pamphlets of the period: a doctrine in which the worker’s hunger was real and his humiliation was secondary, in which the landlord’s rent mattered and the landlord’s contempt was an ornament on the rent, in which what a man was prevented from being by the contempt of his neighbours would be cured automatically once he was no longer prevented from eating by their ownership of the land. The classical position’s weakness was not that it attended to property. Its weakness was that it had classified contempt as a shadow cast by property, and shadows, in its scheme, required no independent strategy, only the removal of the object casting them.

Why intelligent people held the classical position

Reconstructing the strength of a defeated position is a condition of understanding why its defeat took so long, and the classical position had three distinct strengths. Its first strength was analogical. European history, as the nineteenth century had written it, seemed to show religion and inherited rank dissolving under the acids of commerce and industry, and the Indian case was read through that template: caste was India’s version of the estates and guilds, unusually elaborate, unusually durable, but the same kind of thing, and therefore subject to the same solvent. The analogy was not absurd. It failed, in the end, on a specific difference that the critic would name: European ranks had not been a graded religious order dividing the labourers themselves into sealed compartments, each with a scriptural warrant and each guarding its distance from the compartment below. An analogy that misses the mechanism of a thing’s reproduction will mispredict the thing’s death, and caste reproduced itself through marriage and ritual, mechanisms that the market left almost untouched.

Its second strength was strategic, and it deserves to be stated as its holders felt it. The Indian working class of the period was small, concentrated in a few cities and industries, and surrounded by an ocean of peasantry. The movement’s great fear was fragmentation: a labour movement splitting along communal lines, as Indian politics repeatedly did, would be useless to everyone. Insisting that the worker was a worker first, and that every other identity was a distraction fostered by the enemy, was a discipline imposed against a real and visible danger. When communal riots could empty a mill district in a day and set Hindu against Muslim among men who had struck together the week before, the refusal to organise on the lines of inherited identity felt less like blindness than like the first condition of survival. The error was to extend a discipline about what the party would organise around into a theory about what society was made of, as if declining to mobilise caste meant caste had thereby been explained.

Why did the early communists expect caste to dissolve?

They expected dissolution because the evidence of their formative decades pointed that way and because their theory named the solvent. Urban employment was visibly loosening hereditary occupation, colonial law had displaced caste courts, and the factory whistle summoned men whom the village order would have seated apart.

In the inherited doctrine, such arrangements belonged to the superstructure of a feudal economy, and a superstructure deprived of its base was expected to crumble, slowly and unevenly, but as surely as a roof outlasts the walls that held it by years rather than by centuries.

The third strength was a genuine insight that later positions had to preserve. Caste in the Indian village was not only a ritual order. It was also, undeniably, a way of organising the division of labour and the distribution of its product: the jajmani arrangements of hereditary service, the attachment of labouring families to dominant landholding families, the reservation of clean and lucrative functions to the high and of degrading functions to the low, tied ritual rank to material position so tightly that in most villages a man’s caste predicted his land, his credit and his diet with grim reliability. A theory that noticed this fit was noticing something real, and the correspondence between caste and class position, which the empirical sections of this article examine in detail, is the permanent deposit the classical position left inside all its successors. What the classical theorists could not explain was the traffic in the other direction: the ways in which caste did not merely reflect an economic order but actively produced and policed one, assigning the labour before the market could bid for it, and surviving every change of the surrounding economy by re-establishing the ranking inside each new arrangement it met.

Exit as evidence: what the converts teach the debate

Among the bodies of fact that accumulated against the classical position, the history of conversion deserves a section of its own, because it constitutes something close to a controlled experiment in the debate’s central question. Across many centuries, communities at the order’s floor and margins exited the Hindu fold in large movements: to Islam, whose Indian expansion drew heavily, in region after region, from those the hierarchy valued least; to Christianity, whose missions in the south and the northeast gathered congregations disproportionately from the oppressed castes and the tribal peoples; to Sikhism, whose founding teaching denied caste in terms as explicit as any in the record; and, in the twentieth century, to the Buddhism of the Nagpur conversion. In each case the exit renounced, formally and completely, the ideology that the classical position blamed for caste’s survival: the converts ceased to believe in karma as the warrant of their station, ceased to honour the scriptures that ranked them, ceased, in the Sikh and Buddhist cases, to belong to a religion with any doctrine of graded birth at all. If caste were a superstructural effect of that ideology, the experiment’s prediction is unambiguous. It should have dissolved in the convert communities within a generation or two, leaving behind, at most, a memory.

The experiment’s result is the result the address’s anatomy predicts, and it is double. The ideology’s departure did not dissolve the structure: the convert communities carried caste with them through the door of exit, ranking their own converts against one another, endogamous within the inherited circles, worshipping in separate congregations where the old order’s converts predominated, and reproducing, inside religions of formal equality, the gradations they had fled. At the same time, and this is the result’s other half, the exit was never nothing. It armed the converts with an alternative account of their worth, with institutions, schools, missions, congregations, printing presses, that answered to them rather than to the hierarchy, and with a standing refutation, embodied in their own persons, of the doctrine that their degradation was sacred. The communities that exited did not escape caste. They acquired, in the exit, weapons against it that staying had not provided, and the history of the anti-caste movement is in large part the history of those weapons being taken up. For the debate, the double result is decisive against every single-grammar theory. Belief abandoned did not free the believers, which refutes the idealist reading; belief abandoned armed them, which refutes the reading for which belief is a shadow that changes nothing. What the converts teach is the fourth position’s lesson in its most experimental form: the order lives simultaneously in the conscience and in the arrangements, and it is dismantled, where it is dismantled, by attacks that refuse to choose between its addresses.

The evidence that accumulated against the classical view

Positions of this kind rarely die by refutation in argument. They die by accumulation, as the facts they cannot digest pile up at the edge of the theory until a generation arrives for whom the pile is the most obvious thing in the room. Four bodies of fact did most of the work against the superstructure reading of caste. The first was the behaviour of caste inside the very institutions that were supposed to dissolve it. The Bombay textile mills, the largest concentration of industrial workers in the country, recruited through networks of village and kin, assigned departments in patterns that tracked caste, and left the dirtiest processes disproportionately to the lowest; the railway workshop and the mine showed variants of the same pattern. Men who had been expected to discover a common identity at the factory gate discovered instead that the gate admitted them to a floor plan drawn by the old order. Class organisation grew in those mills, sometimes heroically, but it grew through caste networks and around caste boundaries, not over their graves.

The second body of fact was conversion. Communities that left Hinduism altogether, for Buddhism, Christianity, Islam or Sikh traditions, carried caste distinctions with them into religions whose doctrines gave those distinctions no warrant at all, ranking convert communities against one another, worshipping apart and marrying within the inherited circle. A superstructural effect ought to weaken when the ideology that sustains it is formally renounced. Caste survived renunciation, which suggested that whatever held it in place was not exhausted by belief: it was embedded in marriage, in property transmission, in the organisation of local dominance, and in the material advantages that even a modest rank conferred on those who held it over those placed beneath them.

The third body of fact was political. Universal franchise, introduced at independence, was widely expected to submerge caste in larger national and class appeals. Instead, electoral politics gave caste a new career as a principle of mobilisation, since a hereditary community with a shared name, a shared memory of insult and a dense internal network is a formidable instrument for assembling votes, and parties of every description learned to assemble them. Caste associations, which the classical reading had filed as survivals, modernised themselves into federations, conferences, journals and pressure groups, using print, rail travel and the ballot box with complete competence. The survival had learned to operate the machinery of the society that was supposed to replace it, and a survival that operates the machinery is no longer well described as a survival at all.

What evidence ended the classical position?

No single study ended it; a pattern did. Industrial employment reproduced caste segmentation instead of abolishing it, converts carried caste into religions that denied it any sanction, democratic politics rewarded caste organisation rather than dissolving it, and the lowest communities themselves, organising under their own leaders, insisted that their enemy had a name the class vocabulary did not capture.

Each fact was compatible with caste being weakened at the margins, and the classical position could absorb any one of them as a lag or a residue. Together they reversed the burden of proof: persistence stopped looking like a residue of the past and started looking like a property of the present.

The fourth body of fact was the testimony of the oppressed themselves, organised politically under leadership drawn from their own communities. From the late nineteenth century onward, movements of the lowest castes had insisted, in print and in conference, that their subordination was not a side effect of poverty which poverty’s cure would heal. A poor Brahmin remained a Brahmin, honoured at the temple door that stayed shut to a prosperous Mahar; a labourer who doubled his wage did not thereby gain entry to the well, the school bench or the marriage market of his supposed class allies. This distinction, between what money could buy and what birth forbade, was the experiential core of the anti-caste tradition, and it is what gave the address of 1936 its authority when it finally appeared in print: it was not a seminar objection but the theoretical form of a grievance that its intended audience had spent lifetimes administering and its author had spent a lifetime receiving.

The tradition the address inherited: a century of anti-caste argument before it

The address of 1936 did not invent its subject, and part of its authority came from the tradition it condensed. The critique of caste as a system, argued from the standpoint of those it crushed, begins in modern India with Jotirao Phule, the gardener-caste reformer of Pune who founded the Satyashodhak Samaj in 1873 and devoted his life to the proposition that the priestly order was a fraud practised on the cultivating and labouring majority. In Gulamgiri, published in the year of the Samaj’s founding and dedicated, with deliberate provocation, to the people of the United States who had abolished slavery, Phule retold the sacred myths as a conqueror’s archive: the incarnations of the gods became, in his reading, the successive campaigns by which an incoming priestly and warrior order subdued the indigenous cultivators, demonised their kings and installed a hierarchy that taxed the many for the leisure of the few. Strip the mythology of its costume and the argument is recognisably materialist in shape: a ruling stratum maintains itself by an ideology that sanctifies its extraction, and the extraction is of grain, labour and dignity together. Phule had not read Marx, whose works were unavailable to him; he arrived at the family resemblance independently, from the furrow rather than from the library, which is itself a datum in this debate, since it suggests that the analysis of caste as a system of exploitation was generated by Indian conditions before it was imported by Indian communists.

From Phule’s Maharashtra the argument travelled along several roads the address would later gather. The non-Brahmin movements of the Bombay and Madras presidencies converted the critique of priestly monopoly into organised politics, demanding education, office and representation for the cultivating castes and treating the Brahmin not as a theological problem but as a competitor entrenched in the schools and the secretariats. In the south, the Self-Respect movement associated with Periyar, E. V. Ramasamy, carried the attack to the point of atheism, staging the burning of the law books and the ridicule of the gods with a theatricality Phule would have recognised. Within the princely states and the presidencies, the communities placed below the line of untouchability began, in the opening decades of the twentieth century, to organise in their own name and under their own leadership, in conferences and associations that refused the tutelage of caste Hindu reformers and demanded rights as citizens rather than charity as penitence. It was from this last stream, the autonomous organisation of the oppressed themselves, that Ambedkar emerged, and the emergence explains the address’s distinctive tone: it addresses caste Hindus, reformers and socialists alike from outside their shared assumption that the emancipation of the lowest was a gift to be bestowed by their betters, at a pace their betters would set.

His formation gave the critique an equipment no earlier statement had possessed. The movements had produced the grievance and the organisation; the universities of two continents gave him the comparative and legal apparatus to state the grievance as theory. He had examined, as a student of economics, the finances of the colonial state; he had examined, as a barrister, the difference between a right proclaimed and a right enforceable; and he had examined, as the leader of the Mahad satyagraha of 1927, where thousands of his followers asserted the civic right to draw water from a public tank, the exact distance between the law’s promise and the village’s practice. When the Round Table Conferences carried him to London as the representative of the depressed classes, and when the award of separate electorates for those classes was traded away in the Poona Pact of 1932 under the pressure of Gandhi’s fast, he learned the lesson that underlies the address’s political science: that emancipation conceded by the dominant, on terms the dominant dictate, reproduces the dominance inside the concession. The address was written four years later, and it is best read as the theoretical deposit of that whole education, the point at which a movement’s experience was converted into a doctrine that any theory of Indian society, Marxist theories included, would thereafter have to answer.

An address its audience refused to hear

In December 1935 a reform organisation in Lahore, the Jat-Pat Todak Mandal, a body of caste Hindu reformers devoted to breaking down the barriers between castes, invited Ambedkar to preside over its annual conference and to address it on the caste system. He accepted, and prepared his address with the care of a man who knew exactly what he intended to say and to whom. The draft was sent to the organisers, as courtesy required. The organisers read it and recoiled: the address did not propose to soften caste, to purify it or to reinterpret it, but to annihilate it, and it located the condition of annihilation in the destruction of the religious authority that sanctified the system, a conclusion the reformers judged unbearable for their conference. Correspondence failed to move either side. The invitation was withdrawn, the conference lost its president, and the address lost its audience, in circumstances that its author then described in a prologue when he published the text himself in 1936 as Annihilation of Caste, printing at his own expense the speech that no platform would carry.

The circumstances of non-delivery matter to the debate because they exhibit the argument in action before the argument is even opened. A body of reformers committed, in their own description, to breaking caste barriers found the price of the address too high at exactly the point where the address touched religion, which was precisely the point the address existed to make: that caste could not be reformed by men who intended to keep the scriptures that authorised it, and that the distance between breaking barriers and annihilating the order was the whole of the question. The text’s later career completed the irony. The undelivered speech became the most widely read anti-caste document in the language, reprinted in edition after edition, argued over by Gandhi in print within weeks of its appearance, and installed as the founding statement of a tradition of criticism that no Indian theory of society has since been able to ignore. Its author, born into the Mahar community and placed by the order he anatomised among the untouchables, had taken a doctorate at Columbia and another in London, had read the economists and the jurists, and had led mass movements of his own people for a decade before he wrote it. The address was therefore not an outsider’s polemic against Marxism. It was a considered judgement on the socialist programme in India, delivered by a contemporary who shared many of that programme’s ends and intended, among other things, to explain why its means would fail on the ground he knew best.

Why was the address never delivered?

The inviting organisation, a society of caste Hindu reformers, read the prepared text and found its conclusion unbearable: the address argued that caste could not be ended while the sacred authority of the scriptures that sanctioned it remained intact, and the reformers had not bargained for an attack on religion itself.

Negotiation over the text failed, the invitation was withdrawn, and Ambedkar published the speech at his own expense, so that the most influential anti-caste argument in modern India reached the public as a book rather than as a speech, carrying a prologue that records the refusal.

A division of labourers, not of labour

The address opens its theoretical attack with a distinction that has never been answered on its own terms. Every complex economy divides labour: some grow food, some make cloth, some build, some teach, and a theory of society may properly ask how that division is organised and who appropriates its fruits. Caste, Ambedkar argued, does something categorically different and worse. It divides the labourers. It sorts the people themselves into sealed and hereditary compartments, assigns the compartments to the tasks in advance of any aptitude or choice, ranks the compartments against one another, and forbids movement between them, so that the division of work is frozen into a division of human beings. His sentence for the distinction has become the most quoted line in the whole debate, and it earns the quotation: the caste system, he wrote, “is not merely a division of labour. It is also a division of labourers” into watertight compartments. A division of labour can be rearranged by economic change, by technology, by education, by the decisions of planners or the movements of markets. A division of labourers reproduces itself in the marriage bed and the cradle, and no change in the organisation of production touches it unless the change also reaches the rules by which people are born, ranked and wed.

The analytical consequences of the distinction run deep, and later Marxist writing on India has had to absorb them one by one. If labourers are divided before they reach the labour market, then the market does not meet a supply of undifferentiated workers and proceed to sort them by skill and capital, as the simpler class models assume. It meets a population pre-sorted by birth into those who may and may not do each kind of work, and the sorting is enforced by sanctions that have nothing to do with price. The sweeper’s son is not merely unlikely to become a clerk; the order intends that he shall not, and his community’s internal discipline, the dominant communities’ violence and the state’s indifference combine to keep the intention good. Class formation, in such a setting, does not begin from a blank sheet. It begins inside compartments, and the compartments decide in advance much of what class analysis then discovers and mistakes for its own findings: which groups are landless, which are tenants, which hold the plough and which hold the pen.

What follows from dividing labourers rather than labour?

Three consequences follow, and each one damages a different part of the classical position. Reproduction is the first: compartments sealed by endogamy renew the division in every generation without any economic mechanism needing to act, and without any decision that a later generation can repeal.

Valuation is the second: the worth of the person precedes the worth of the work, so degrading work is assigned to degraded people and then cited as proof of their degradation. Solidarity is the third: workers ranked against one another in a sacred order arrive at the factory or the field already instructed in contempt and deference, so the unity that class politics presupposes has to be built against the grain of the workers’ own formation, not merely announced.

Ambedkar pressed the valuation point with particular force, because it inverted the causal story the classical position told. In that story, degrading work existed and someone had to do it; the people who did it were then despised, as an unfortunate cultural residue, for doing it. In the address’s story, the order comes first and the assignment follows: a class of persons is designated as degraded, the degrading work is reserved for them as the expression of their nature, and the work’s degradation is then used, in a perfect circle, to justify the designation. The circle matters because it explains the otherwise baffling failure of economic change to launder the stigma. When sanitation work is mechanised in part, the stigma does not attach to the machine; it follows the community that the order has marked, into whatever work remains to it, and men continue to die in sewers and septic tanks in the age of the municipal corporation, lowered into poisoned darkness by a hereditary summons that no labour legislation has yet managed to repeal in practice. A theory that treats the stigma as a shadow of the task cannot explain why the shadow survives the task’s transformation. A theory that treats the stigma as a property of the ranked order, imposed on persons before tasks, explains it at once.

Endogamy: the mechanism the critic named first

Twenty years before the address, the mechanism at the centre of it had been identified by its author in an academic paper, and recovering that paper changes the standing of the whole critique. In 1916, as a doctoral student at Columbia, Ambedkar presented a paper on the mechanism, genesis and development of caste in India to an anthropology seminar, and published it the following year. The paper is a young scholar’s performance, learned, combative and systematic, and its central claim has survived a century of anthropology remarkably intact: the essence of caste is endogamy, the superimposition of the rule of marrying within the group upon a wider society that had once married outside it, begun, on his reconstruction, by the priestly caste’s decision to close its own ranks and then imitated, at descending levels of success, by every group below. Caste, in the paper’s analysis, is not primarily a theory of occupation or of purity. It is a system of enclosed classes, and the enclosure is matrimonial.

The analysis carries a consequence that the older paper states with an engineer’s pleasure and that the later debate kept rediscovering. An endogamous group must solve a bookkeeping problem no exogamous society faces: the numbers of marriageable men and women inside the enclosure never match, and the surplus of each generation must be disposed of without breaching the wall. The paper argues that the notorious institutions of Hindu social life are the disposal mechanisms: the burning of the widow with her husband removed the surplus woman who could not be remarried outside; enforced widowhood, with its shaven head and its civic death, neutralised her where burning was forborne; and the marriage of girls before puberty, to husbands chosen early and sometimes grotesquely older, managed the surplus man by capturing brides before the market in them could disturb the enclosure. A reader need not accept every step of the reconstruction to feel its force. It explains caste from a mechanism that operates every generation, in the marriage negotiations of ordinary families, rather than from a conspiracy of priests or a stage of production, and it locates the system’s heart exactly where the address would later locate it: in the control of marriage, and through marriage, of women, whose sexuality is the wall’s mortar and whose subjection is therefore not an incident of caste but its load-bearing structure.

Why is endogamy the mechanism that matters most?

Because endogamy is the only feature of caste that manufactures the next generation of the system. Occupations can be abandoned, doctrines doubted, interdining tolerated and temples opened, and the order will survive each concession intact if the marriage circle holds, since the circle alone guarantees that rank, property and identity descend together.

Every other feature of caste is a practice that persons perform; endogamy is the process by which the persons themselves are produced as members. The address’s author had identified that priority as a student, two decades before the address made him famous for it, and the scholarly literature’s slow arrival at the same priority is one of the debate’s clearer verdicts on his quality as a theorist.

For this debate, the endogamy analysis does decisive work in both directions, which is the mark of a real contribution. Against the classical position, it identifies a reproduction mechanism that no transformation of production touches directly: a society may socialise its factories and collectivise its land and still marry its children inside the inherited circles, regenerating the hierarchy from the cradle with every generation, a point the twentieth century’s actually existing socialisms illustrated in their own idioms wherever inherited status orders survived their revolutions. Against a purely religious account of caste, the analysis is equally disobliging, since the mechanism it names is demographic and material rather than theological: the wall is maintained by the disposal of surplus persons, by the policing of widows and the timing of marriages, practices that scripture blesses but does not create, and that persist, in softened and negotiated forms, among communities whose theology has long ceased to command them. The mechanism is social, material and self-renewing, and it belongs to neither the base nor the superstructure of the old model, one more exhibit in the case for the fourth position’s retirement of that model for this object.

The gendered character of the machinery deserves its weight in the verdict, because it is here that the Indian debate touches the theme the synthesis scholarship would later formalise as the interlocking of caste, class and gender. A system whose reproduction requires the surveillance of women’s marriages requires, as its corollary, the surveillance of women: the seclusion, the early marriage, the widow’s degradation and the honour economy in which a family’s rank is hostage to its daughters’ conduct are not decorations on caste but its operating costs, charged overwhelmingly to women, and charged at rates graded by the family’s position, most savagely where the claim to rank is most precious and least secure. Any account of caste that omits this machinery has omitted the room in which the system’s future is decided, and any politics that proposes to annihilate caste while leaving the marriage market and the honour economy untouched has mistaken the building for its façade, a mistake the student of 1916, whatever else he had still to learn, had already declined to make.

The staircase with no common landing

The grading of the order is the second element of the critique, and it is aimed at the political heart of Marxism rather than at its economics. A class society, in the classical Marxist picture, simplifies as it develops: the ranks of the intermediate strata thin out, and society polarises toward two great camps whose confrontation organises politics. Caste does the opposite. It multiplies ranks and distributes them so finely that almost every group has both a superior to envy and an inferior to command, and Ambedkar’s description of the result, reverence ascending and contempt descending storey by storey, captures the psychological engineering of the arrangement. The Brahmin is revered by all; the Kshatriya reveres the Brahmin and is revered by the Vaishya; and so downward, each caste nursing its grievance against the one above while policing its precedence over the one below, until the scale reaches the untouchable, who has no one beneath him within the order and whose function is to be the floor on which the whole structure of relative dignity stands.

A staircase so arranged is a machine for preventing combination. The class theorist who calls on the lower orders to unite confronts the fact that the lower orders experience one another, in the first instance, as rivals in a competition for rank that the order itself administers, and that the nearest enemy of each group, the group whose proximity threatens its fragile precedence, is the group immediately adjacent to it rather than the group at the summit. History supplied abundant illustration: resistance to the claims of the lowest castes has repeatedly come with the greatest violence not from the summit of the order but from the landed and cultivating castes immediately above them in the local scale, whose relative standing was what the claims endangered. The critique’s point is not that combination is impossible. It is that combination across caste has a price and a difficulty that the classical position, by treating caste as a fading ideology, systematically refused to budget for, and that a movement which will not budget for the difficulty will repeatedly mistake its own failure for the backwardness of the workers.

There is a subtler strand of the same argument concerning interest. In the classical model, men are moved, in the decisive instance, by their material interests, and shared exploitation generates shared interest generates shared organisation, in a sequence so reliable that the party’s task is mainly to articulate what the situation already teaches. Graded inequality breaks the sequence at its second link. A group’s material interest in abolishing the order competes with its status interest in preserving the order’s lower storeys, on which its own modest dignity depends, and the status interest is not a delusion to be dispelled by pamphlets. It is a real good, distributed by the system to its participants in careful proportion to their submission, and it is collected daily, in the deference of inferiors, the services of the hereditary servant, the knowledge that however poor a man may be, there are those whom the village obliges to call him master. Ambedkar’s analysis asks the class theorist to take the status wage as seriously as the money wage, and it is a measure of the critique’s quality that modern scholarship on hierarchy, in several disciplines, has had to reinvent that request under other names.

The status wage: honour as a currency men are paid in

The staircase argument deserves one further development, because it contains an economic insight that later theory, in another country, would state in almost the same terms. If the graded order distributes dignity as carefully as it distributes grain, then dignity functions as a wage: a real payment, received daily and valued accordingly, which the system makes to nearly every participant in return for policing the participants below. The labourer who owns nothing may still own his precedence over the scavenger, and he will defend that asset with the tenacity of any small proprietor, because it is, in the coin of the village, his capital. Class politics that offers him solidarity in exchange for surrendering the asset is asking him to make a present and certain sacrifice against a future and uncertain gain, and his refusal, which the movement habitually read as ignorance or priestcraft, is on this analysis a rational defence of the only property the order has allowed him. The address’s political pessimism rests substantially on this calculation, and it explains why its author insisted that the order had to be annihilated as a system of honour, and not merely circumvented as an arrangement of production.

The parallel across the ocean is exact enough to name. Writing of the American South in Black Reconstruction, published in 1935, one year before the address, W. E. B. Du Bois argued that the white labourer, though poorly paid in money, was compensated in part by a public and psychological wage: deference, admission to public functions, the courtesy titles of race, the knowledge that the colour line placed a whole people beneath him whatever his own station. Two thinkers, working in ignorance of each other on opposite sides of the world, arrived at the same mechanism to explain the same puzzle, the loyalty of the exploited to a hierarchy that exploits them, and located the mechanism in the same place: a wage paid in status, distributed down the hierarchy in graded amounts, and forfeited by any solidarity that levels the ranks. The convergence is among the strongest reasons for treating the mechanism as real rather than rhetorical. Systems of graded birth and systems of race differ profoundly in their histories and their justifications, and the Indian order’s religious sanction and matrimonial machinery have no American equivalent. But at the joint where the theory of class meets the recalcitrance of the divided labourers, the Indian critic and the American historian are describing one problem, and the Marxist traditions of both countries spent much of the twentieth century learning to price a wage their founding models had not listed among the payments men work for.

The development completes the staircase argument’s work in the debate. It converts the critique’s psychology into political economy without reducing the one to the other, since the status wage is a genuine material interest that a purely economic vocabulary can state but a purely economic programme cannot buy out. A movement may offer the labourer higher wages, land, education and power, and he may calculate, correctly within the order’s terms, that equality with those beneath him costs him the one distinction his poverty had spared. Overcoming that calculation requires changing the terms, abolishing the currency in which the wage is paid, and that is an operation on the order of honour itself, on its scriptures, its marriages and its daily etiquette, of exactly the kind the address prescribed and the classical position deferred. The point will return in the verdict, because it is the sharpest available statement of why the annihilation of caste could not be a by-product of any struggle that declined to aim at it.

Sanctioned by scripture: why caste is not an economic arrangement with religious decoration

The third element of the critique concerns the source of caste’s authority, and it is here that the distance from any purely economic account becomes unbridgeable. Caste, in the classical Hindu justification that the address takes as its target, is not presented to its subjects as a human convention that might be renegotiated when it becomes inconvenient. It is presented as sacred law. The order of the varnas is woven into the creation accounts of the scriptures; the duties of each station are prescribed in the law books, the Shastras, whose authority the tradition treats as beyond human revision; and the doctrine of karma supplies the moral bookkeeping, assuring the sweeper that his station is the just wage of deeds in a former life and assuring the priest that his eminence is equally earned. A hierarchy so warranted does not present itself to the believer as an arrangement at all. It presents itself as the structure of the moral universe, and to disobey it is not to break a rule but to commit a sin whose price will be collected across incarnations.

Ambedkar drew the conclusion without flinching, and the conclusion is what cost him his Lahore audience. If caste is sustained by the sanctity of the texts that command it, then caste cannot be destroyed while that sanctity stands; reform confined to conduct, to interdining and temple entry campaigns and the softening of manners, leaves the warrant untouched and therefore leaves the system its means of regeneration. The real enemy, in his formulation, was the religion of the Shrutis and the Shastras themselves, and the address insists that the believer must be brought to see that what he calls religion is in fact law, codified by a priestly tradition in its own interest, before the hierarchy built upon it can be dismantled. Whatever a reader’s theology, the analytical point is separable and severe: a stratification system whose legitimacy is sacred cannot be reduced to the economic relations of the epoch that happens to host it, because its subjects experience its commands as obligations of conscience, and men will starve for a conscience who would never starve for a custom. Economic change alters what men do for a living. It does not, of itself, alter what they believe they owe to God, to ancestors and to the order of the world, and the address’s whole case is that in India the second of those is where caste keeps its garrison.

What does religious sanction add to caste’s durability?

Religious sanction converts compliance into piety and dissent into sin, and it thereby gives the hierarchy a garrison inside the conscience of every believer, including the believers it degrades. Karma and rebirth recast inherited station as moral desert, so its victims are given an explanation blaming their former selves, and its beneficiaries an assurance that their eminence is earned.

A hierarchy defended by interests can be bargained with; a hierarchy defended by the structure of the moral universe must first be disbelieved, which is why the address treats an attack on scriptural authority as the precondition of annihilation rather than as an optional radical flourish.

The durability evidence supports the distinction. Caste outlasted the Buddhist centuries that denied its warrant, the bhakti movements that softened its devotions, the Muslim and British sovereignties that were indifferent or hostile to its theology, and the arrival of an economy that had no doctrinal use for it at all. In each passage the order adapted its surface and preserved its grammar: hereditary rank, endogamous closure, graded honour, sanctioned pollution. Political economy can narrate those passages as changes in the base with lags in the superstructure, but the lags lengthen into millennia, and at some point a lag that outlasts every base it is blamed on has to be promoted, in honesty, into a structure with a life of its own. That promotion, performed inside Indian Marxism over the following decades, is the story of the responses examined later in this article; the address performed it in a single step, by the simple method of asking what, in the believer’s own account, actually held the system up.

The monster across the path of every reform

The address gathers its strands into a single warning to the socialists, and the warning is the part of the critique that the movement’s later history would test most directly. Suppose, Ambedkar argued, that the socialists decline to deal with the social order and proceed to their revolution on economic lines alone. They will find, first, that the proletariat they summon is divided against itself by caste and creed, and will not combine on the assurance of leaders that such things do not matter. They will find, second, that if the revolution succeeds by fortune, the new state will wake on its first morning to the prejudices of high and low, clean and unclean, operating inside every office and village, and will be compelled to grapple with the social order it postponed, now with the whole machinery of power entangled in the grappling. “He will be compelled to take account of Caste after the revolution”, the address concludes, “if he does not take account of it before the revolution”, and the sentence that follows has passed into the language: turn in any direction you like, “caste is the monster that crosses your path”, and neither political nor economic reform can be had until the monster is killed.

Stated with the address’s own force, the objection to class-first politics is not that class is unreal, nor that economic exploitation is a fiction, nor that the landlords and employers the communists fought were innocent. The objection is that the instrument is broken before it is lifted. A politics organised around class in Indian conditions asks for an alliance that caste has already made impossible on the terms the politics assumes: it asks the scavenger and the mill hand of respectable caste to discover themselves as one class, while everything in their formation, their marriage prospects, their ritual standing and their daily treatment of one another instructs them that they are not one thing and must not become one thing. It asks the movement to postpone the fight against the very divisions that prevent its army from assembling, promising that the divisions will be attended to after the victory that the divisions prevent. And it locates the authority for that postponement in a theory imported from societies whose workers were divided by much else, but not by a sacred and graded order of birth. That is the critique at full strength, and it is worth pausing on the fact that nothing in it depends on denying the reality of exploitation: the address’s author accepted the socialist indictment of property, as the next sections show, and aimed his fire not at the destination but at the map.

Does caste block the road to class unity in the address’s argument?

It blocks it twice over, in the address’s construction. Before any revolution, the graded order divides the labourers who would have to combine, so the united front that economic reform presupposes cannot be assembled on class terms alone, however complete the programme’s economics and however sincere its leaders.

After any revolution made without settling the social order, the same prejudices reappear inside the new institutions, whose functionaries and citizens carry the hierarchy in their own attitudes, and the state is compelled to fight the postponed battle from within its own machinery. The monster crosses the path at both ends of the journey, which is why the address insists on taking account of caste first.

The reformist answer: Gandhi’s defence and the rejoinder it provoked

The address did not lack for respondents, and the first and most consequential of them was not a Marxist at all, a fact that situates the debate correctly in the India of its time. Within weeks of the text’s publication, Gandhi reviewed it in his journal Harijan, in a series of articles later printed with the address itself as an appendix, so that the two documents have travelled together ever since in an unwilling partnership. His defence was the reformist Hindu position at its most distinguished. Untouchability he condemned without reserve, as a blot and a sin to be removed; but the varna order itself he defended as a law of human nature wrongly confused with the innumerable castes that had corrupted it, a division of hereditary callings that, purified of hierarchy and of the pollution rules, he held to be compatible with the equal dignity of all labour. The remedy, on this view, was the repentance and reform of caste Hindus, the opening of temples and hearts, not the annihilation of the order and certainly not the destruction of the scriptures’ authority, which he received as the address’s most dangerous demand.

The rejoinder, published in the same volume, is a model of the address’s method applied to its most formidable opponent, and its structure is the debate in miniature. Ambedkar replied that the distinction between varna and caste could not bear the weight placed on it, since the ideal division of callings, hereditary and sanctified, was the parent of the actual system and not its corruption; that a varna without hierarchy was a fantasy, because the scheme’s whole content was the ranking of its parts; that the saintly reformer who defended the scriptures while condemning their fruit was asking the tree to be judged by everything except its produce; and that the test of any doctrine of Hindu society was what it offered the untouchable, for whom the difference between being despised under a corrupt system and being assigned, kindly and eternally, to the scavenging varna under a purified one was not a difference he was obliged to celebrate. The exchange matters here because it fixes the third party of the debate, the reformist position that would annihilate untouchability while preserving the sacred order, between the classical Marxism that would postpone caste and the annihilationism that would end it. Indian argument about caste has been conducted among those three poles ever since, and the Marxist positions examined in this article were all formulated in a field where the reformist pole, armed with the Mahatma’s authority, competed for the same constituency of the concerned. A history of the debate that confined itself to Marxists and their critic would miss the triangle in which the actual arguments moved, and miss, too, the address’s most underrated achievement: it forced both of the other poles to state, in public and in print, exactly what they proposed to do about the well, the temple and the scavenger, and to be judged by the answers.

The third party to every argument: the national movement and the price of unity

No position in this debate was formulated in a vacuum, and the vacuum’s name was the national movement, whose overriding need, Hindu unity against the colonial power, taxed every Indian argument about caste for half a century. The tax fell on all three of the debate’s poles in different currencies. The reformists paid it as restraint: the campaign against untouchability was to be conducted so as never to divide the national ranks, which meant, in practice, that the oppressed were asked to accept the pace of their oppressors’ repentance as the pace of their emancipation, and to regard the demand for separate political safeguards as a blow struck for the empire. The communists paid it as subordination of a second kind, for a movement competing for the leadership of the anti-colonial struggle could not afford to be found lecturing the nation on its internal hierarchies while the national leadership monopolised the anti-colonial emotion, and the caste question’s postponement inside the movement cannot be understood apart from that competition, though it cannot be excused by it either, since the address’s whole point was that the hierarchies were the nation’s, and a freedom that left them intact would be the hierarchies’ freedom. The anti-caste movement paid the highest rate of all, in the coin of legitimacy: its refusal to wait was read, by the national mainstream and by much of the left alike, as collaboration with the power that profited from Indian division, and its leader’s presence at the Round Table Conferences, arguing his people’s case before the imperial arbitrator because no Indian forum would hear it, was held against him for the rest of the century by men whose own forums had refused him a hearing.

The triangle explains a peculiarity of the debate that puzzles readers who meet it in the syllabus, where it appears as a seminar disagreement among theorists. In its own time the debate was conducted under the discipline of a struggle for the state, and every theoretical position carried an immediate price in allies, in unity and in the suspicion of treason, which is why the positions were held with a tenacity that purely intellectual conviction rarely sustains and abandoned, when they were abandoned, slowly and with elaborate face-saving. Independence changed the triangle’s terms without dissolving it. The republic the national movement founded wrote the anti-caste settlement into its constitution and inherited the hierarchy as its electorate in the same act, and the debate continued inside the republic’s institutions, in its universities, its parties and its courts, under the new discipline of electoral arithmetic rather than the old discipline of anti-colonial unity. The forms of the tax changed. The taxed argument, that the republic’s deepest division was the one its founding categories were built to transcend and its founding compromises were built to postpone, remained recognisably the address’s argument, wearing the dress of each succeeding decade.

The charge against the movement itself

The address’s criticism of doctrine was accompanied by a criticism of the movement that carried the doctrine, and this part of the critique must be stated as its author stated it, without the softening that sympathetic historians have sometimes applied to it after the fact. Ambedkar charged that the communist movement in India was led by men drawn overwhelmingly from the privileged castes, that its leadership shared the social formation of the order it claimed to oppose, and that it neither understood untouchability from within nor felt it as the burning question of Indian society. The charge was not that individual communists were insincere in their hatred of poverty. It was that a movement whose leading cadres had never been refused the well, never been made to sit apart at the school wall, never inherited an occupation of filth, would reliably misjudge what the caste order was, treating as a superstition of the masses what was in fact the organising principle of the society around them, and postponing, with the clean conscience of men who had never paid its price, a battle that was not theirs to postpone.

He put the point about the rank and file in the form of two questions that the address addresses to the socialists directly, and they retain their edge because they are empirical. Can it be said that the proletariat of India, poor as it is, “recognises no distinctions except that of the rich and the poor”? Can it be said that the poor of India recognise no such distinctions of caste or creed, high or low? If the fact is that they do, he asked, what unity of front can be expected from such a proletariat in its action against the rich? The assurance of a socialist leader “that he does not believe in Caste” would not suffice, because the assurance had to proceed from a much deeper foundation, from the mental attitude of the compatriots toward one another, and that attitude was the one thing the movement’s economic programme did not propose to change. Between the leader who does not believe in caste and the follower who will not take water from the hand of the man beside him on the picket line, the address locates the whole failure of the classical position in practice.

The record against which that charge was laid is part of the history of the movement, treated in full in the series history of the communist movement in India, and this article states it without varnish, as the brief of the debate requires. The early communist leadership in Bombay and elsewhere was recruited heavily from the educated strata of the privileged castes, as was the leadership of nearly every Indian political formation of the period, including the nationalist one. The movement organised mill workers, tenants and labourers across caste lines and fought, in many places, real battles against landlord power in which the lowest castes stood in its ranks; it also repeatedly subordinated caste-specific grievances, such as temple entry, access to water and the abolition of hereditary service, to the priorities of trade union and electoral strategy, and its theoretical statements on caste remained, for decades, thin restatements of the superstructure doctrine long after its own organisers’ experience had outrun them. Ambedkar’s specific accusation, that the movement lacked understanding of untouchability because untouchability was not the experience of the men who led it, was directed at that gap between the movement’s ranks and its command, and it has never been met by the simple method of refutation, because the demographic premise beneath it was, in the period he described, substantially true.

What precisely did the critic charge the communists with?

The charge had three counts. The first was composition: the movement was led by men of the privileged castes, formed by the very order the movement proposed to transcend. The second was comprehension: never having lived under untouchability, that leadership mistook a sacred and graded hierarchy for a removable prejudice, and so underestimated what abolishing it would require.

The third was priority: the movement asked the oppressed to defer their specific emancipation until after a general revolution, a sequence whose convenience for those not suffering the specific oppression the address exposes with some dryness. Together the counts amount to the claim that the movement’s theory of caste and the movement’s sociology were the same fact viewed from two sides.

It is part of the discipline of this article that the charge is presented here at full strength and is not answered in the movement’s defence, because the brief of the debate assigns the defence elsewhere and the record is what it is. What can be registered, as a matter of the debate’s own history rather than as a rebuttal, is the use later Indian Marxists themselves made of the indictment. When, decades afterward, serious Marxist scholarship in India reconstructed its account of caste, it did so in a landscape where the address’s questions had become unavoidable, cited approvingly by scholars inside the tradition who treated the critique not as an enemy’s taunt but as a standing examination the tradition had to pass. The strongest objection section of this article returns to that reckoning. For the moment the charge stands as its author left it: unanswered by the generation it addressed, and directed at any successor generation that proposes to inherit the movement’s programme without inheriting its examination.

The premise examined: the sociology of the movement’s command

The charge’s first count, the composition of the movement’s leadership, is an empirical claim, and the debate is served by treating it as one rather than as a slogan either side may wave. The studies of the Indian communist movement’s social composition, conducted by historians of the movement across several decades, support the premise in its period form. The founding and interwar leadership in the movement’s principal centres was recruited overwhelmingly from the educated strata of the privileged castes, schoolteachers, clerks, journalists and students whose literacy was itself a caste inheritance in a society where literacy had been a near-monopoly of the twice-born. The pattern was not a communist peculiarity. It reproduced, inside the movement, the sociology of every literate politics in the India of the time, nationalist, reformist and loyalist alike, since the pool of the educated from which all leaderships were drawn had been filtered by the caste order long before any party existed to recruit from it. The address’s charge, precisely understood, was therefore double: that the movement shared the period’s universal limitation, and that a movement proposing to abolish hierarchy was obliged, as the others were not, to have noticed and corrected the limitation in itself, in its recruitment, its education and its distribution of authority, and had not.

The correction, insofar as the record shows it, arrived slowly and from below, as the charge would have predicted. The movements that rooted the parties in the countryside, in Kerala, in Bengal, in the Andhra country, drew their local leadership from the cultivating and labouring communities the struggles mobilised, and the texture of the parties changed accordingly at the base while changing more slowly at the summit, a lag the movement’s own internal documents periodically lamented without curing. Women and members of the oppressed communities remained, across the century the debate surveys, rarer in the commanding committees than in the crowds those committees addressed, a distribution the movement shared with the republic it opposed and criticised in others with justice it did not always apply at home. None of this is entered here as a verdict on individuals, many of whom spent their lives and liberty in the struggles this article records. It is entered because the debate’s central empirical claim, that a hierarchy reproduces itself inside the organisations founded to destroy it unless those organisations make its destruction an internal discipline and not only an external programme, is either true or false of the record, and the record, examined without tenderness, says that it is true.

The remedy the critique prescribed: intermarriage and the destruction of the warrant

A critique is half a theory until it states its remedy, and the address’s remedy follows from its mechanism with the same rigour as its diagnosis. If caste is reproduced by endogamy, then the solvent of caste is intermarriage: the fusion of the circles, beginning with the interdining that custom forbids and ending with the intermarriage that custom abhors, until the compartments leak so freely that the compartments cease to exist as compartments. The address names intermarriage as the real remedy with a bluntness that still startles readers who expected a programme of legislation, and the reasoning is the 1916 paper’s reasoning arrived at its practical end. Laws can forbid discrimination in the street and the office; they cannot manufacture the kinship that makes discrimination unthinkable, and a society of groups that will not marry one another will keep its hierarchies in storage, in the family, against every public reform. Annihilation, in this vocabulary, is not a metaphor for strong disapproval. It is the name of a specific social process, the dissolution of the endogamous group, and it is to be distinguished precisely from the amelioration that the reformist position offered and the postponement that the classical position practised.

The second element of the remedy has been described already, in the section on religious sanction: the warrant must be destroyed, the authority of the Shastras denied, the believer released from the doctrine that his station is his soul’s biography. The third element is the one that connects the remedy back to politics and to the debate with the socialists, and it is the insistence that social reform of this order is not the handmaid of political and economic transformation but its precondition. The address’s author had arrived at that sequence through the Poona Pact’s lesson and the Lahore invitation’s collapse, and he states it as a law of Indian public life: a people divided by a sacred hierarchy cannot operate democratic institutions as anything but the hierarchy’s instruments, cannot sustain a revolution as anything but the hierarchy’s reassertion, and cannot be argued out of the sequence by pointing to societies where the precondition happened, for contingent historical reasons, to be less binding. The Indian Marxist tradition spent the rest of the century negotiating with that law, conceding its force in practice long before conceding it in print, and the negotiation, rather than any formal surrender, is the form the critique’s victory took.

What would annihilation require, on the address’s own account?

It would require the dissolution of the endogamous group through intermarriage freely entered and socially honoured, the destruction of the scriptural authority that sanctifies hereditary rank, and the prior or simultaneous reconstruction of the social order that political and economic programmes usually promise to deliver afterward.

Nothing less is annihilation in the address’s usage, and the definition is deliberately severe: it is meant to expose the distance between the order’s actual abolition and the milder projects, reform of manners, abolition of untouchability alone, economic uplift alone, that borrow its rhetoric. A position in this debate can be located with unusual precision by asking what it makes of the definition, and the grid later in this article uses exactly that test.

The labour movement’s own record, examined as evidence

The charge against the movement invites a question the polemic on both sides often dodges: what did organised labour in India actually do about caste, in the unions and parties where the doctrine met the workforce? The record is mixed in ways that instruct both camps, and the debate is best served by examining it as evidence rather than as ammunition. The great textile unions of Bombay built, across the first half of the twentieth century, organisations of genuine class solidarity among workers drawn from many castes and regions, and their strikes held, at their best, across lines the village would have treated as uncrossable; a mill gate picket is a severe teacher, and men who have faced mounted police together do not easily resume the etiquette of the well. Yet the same industry’s internal order bore caste’s fingerprints in its hiring channels, its departmental assignments and its residential geography, and the unions’ leadership, for all its internationalism, rarely made the abolition of caste inside the workforce a stated object of struggle with resources attached. Solidarity was achieved, repeatedly, as a wartime measure of the class struggle, and allowed to lapse, repeatedly, as a peacetime irrelevance, which is the exact pattern the address had predicted for a movement that treated caste as somebody else’s subject.

The counter-experiment was run by the critic himself, and its results belong in the evidence. In 1936, the year of the address, Ambedkar founded the Independent Labour Party to contest elections in Bombay Presidency on a programme that joined the vocabularies deliberately: land to the tiller and the nationalisation of key industries on one page, the abolition of hereditary offices and the annihilation of caste on the next, addressed to an electorate of workers and oppressed peasants whom the programme refused to sort into separate constituencies of grievance. The party’s electoral debut was a striking success by the standards of a new formation under a restricted franchise, returning a substantial bloc of members to the provincial legislature and demonstrating, in the only laboratory the period offered, that a politics fusing the class demand and the caste demand could assemble the constituency that each, separately, had struggled to hold. Its later history was brief and entangled in the great events of the decade that followed, and its founder moved on to federal labour office and to the constitutional work that consumed his final years. As an experiment, though, its result stands in the record beside the mills’ result, and the two together frame the debate’s practical verdict before its theoretical one was written: class organisation that ignored caste achieved solidarity as an episode, and organisation that joined the questions achieved it, for a while, as a programme.

The agricultural field tells the same double story at greater length. Where communist-led peasant movements organised the landless and the tenants, in the Andhra country of the Telangana struggle, in the tebhaga sharecroppers’ movement in Bengal, in the long Kerala campaigns, the social base of the mobilisation was drawn overwhelmingly from the oppressed castes and tribal communities, and the movements’ victories, rent reduced, land redistributed, hereditary service refused, were experienced by their participants as caste emancipation and class emancipation in a single act, indistinguishable in the living of them. The party literature of those movements, examined afterward, often described the victories in the class vocabulary alone, filing under landlordism a complex of dominations its own cadres had fought in the round. Later scholarship on the movements, much of it sympathetic, has restored the caste composition of both the exploited and the exploiters to the centre of the account, and the restoration has a moral for the debate: the practice of the movement at its best was already the fourth position, and what lagged was the theory’s permission to say so. The distance between the movement’s practice and its doctrine, measured in those campaigns, is the most precise available measure of the classical position’s cost, since a theory that cannot describe its own side’s finest hours will not long direct them.

A critic who had read the texts he criticised

No part of this debate is more often falsified, in popular retelling, than the relation of its central critic to the doctrine he attacked. The falsification runs in both directions. To one camp, Ambedkar figures as the man who refuted Marxism, as if the encounter were a duel with a winner; to the other, he figures as a liberal reformer who never engaged the Marxist analysis seriously enough for his objections to count. The record supports neither portrait. Ambedkar had studied political economy at two of the world’s great universities, had taught the subject, and read Marx with the attention of a professional economist and the suspicion of a man checking a contract for the clause that would bind his own people. What he produced, across his writings and supremely in a lecture of his final year, was not a rejection of the whole but an itemised division of it, accepting some propositions at full strength, rejecting others with reasons, and proposing a different road to ends that he and the Marxists substantially shared.

The acceptances were not grudging. The address and the later writings grant the core of the Marxist diagnosis of capitalism: that private property in the means of production is a source of exploitation and of suffering for the many, that the worker is robbed of a part of what he produces by the arrangements that confront him as an ownerless necessity, and that a social order should be judged by whether it ends that robbery rather than by whether it administers it politely. Ambedkar described himself, on more than one occasion, as a believer in the socialist ideal, and his constitutional work carried schemes of state ownership far beyond anything the liberal parties of his day would entertain, a fact that embarrasses every attempt to recruit him posthumously as a mascot against socialism. His quarrel was never with the proposition that property relations shape human misery. It was with three other propositions that the movement had welded to that one: that caste and religion were epiphenomena which an economic revolution would sweep up in its passage; that violence and dictatorship were acceptable instruments for reaching the destination; and that religion as such was an opiate to be abolished rather than a moral discipline to be judged by its fruits.

Did Ambedkar reject the Marxist analysis of exploitation?

He did not, and the point is established by his own division of the doctrine. He accepted that private property generates exploitation and suffering, that the labourer’s product is appropriated under arrangements he does not control, and that these evils require a social remedy rather than private charity.

What he rejected were the movement’s accompanying claims and instruments: the reduction of caste and religion to reflections of the economy, the use of violence as a method, the dictatorship proposed as a transition, and the abolition of religion as an aim. Treating the acceptance and the rejections together is the only accurate way to describe his position, and the fuller comparison belongs to this article’s companion study of the man and Marx.

The late lecture: Marx and the Buddha weighed in one balance

Twenty years after the undelivered address, in the last year of his life, Ambedkar set out the comparison that the earlier polemic had implied, in a lecture of 1956 devoted to Marx and the Buddha. The form of the piece is judicial. He states the creed he attributes to Marx as a series of propositions: that philosophy should aim at reconstructing the world rather than merely interpreting it; that nothing is final or eternal; that the world is governed by forces in conflict; that private property brings power to one class and suffering to another; that the exploitation of the many by the few is an evil to be ended. He then sets the Buddha’s teaching beside each proposition and asks, point by point, where the two agree, where they differ, and by what road each proposes to reach the society in which exploitation has ceased. The lecture is the fullest evidence against the cartoon of Ambedkar as Marxism’s simple enemy, because its author conducts the comparison as a man might compare two medicines for a disease he is determined to cure.

The agreements, in his weighing, were substantial. The Buddha, he argued, had taught the reality of suffering and its roots in a world of conflict and craving; had denied the sanctity of private property in the monastic order he founded and organised that order on communal lines; had preached the equality of men against the doctrine of graded birth two and a half millennia before the modern movements; and had made the ending of suffering, rather than the pleasing of a god, the purpose of the moral life. Where the Marxist creed said that the function of philosophy was to reconstruct the world, the lecture finds the Buddha already committed to a reconstruction, conducted by the transformation of men’s minds and the building of a society on liberty, equality and fraternity. The ends, in other words, were held in common: a world without exploitation, without the mastery of man by man. On the lecture’s account it is the means, and only the means, that divide the two teachers, and the division over means is total.

What did the late lecture accept from Marx?

It accepted the diagnosis. Private property, in the lecture’s weighing, does bring power to one class and suffering to another; exploitation is a real evil and not a misunderstanding; society is a field of conflict rather than a divinely harmonised order; and the purpose of thought is reconstruction, the remaking of the social world, rather than contemplation of it.

The acceptance is stated in the lecture as a series of points on which the Buddha is held to have anticipated or to match the Marxist creed, which is why the comparison reads as an adjudication between two routes to a shared destination rather than as a dismissal of one of them.

The rejections were aimed at the instruments. Violence, the lecture argues, can destroy an old order but cannot persuade men to live justly in the new one; a society founded on force must keep force in reserve against its own members, and the dictatorship proposed as a temporary expedient concentrates power in hands that history gives no reason to trust with its surrender. The dictatorship of the proletariat is therefore refused, not because the ends are wrong but because the instrument contradicts them: men who are to be treated as ends in the destination cannot be treated as material on the road. And religion, far from being abolished, is reassigned. A society requires a moral discipline that men hold sacred if its just arrangements are to be kept when keeping them is costly; the question is not whether there will be religion but whether the religion will sanctify hierarchy, as the Shastras did, or sanctify equality, as a reconstructed Buddhism, in his reading, could. Fraternity, the term missing from the Marxist vocabulary as he received it, becomes in the lecture the load-bearing member of the whole structure: liberty and equality without fraternity, he held, would require a constable to enforce them, and fraternity is precisely the sentiment that caste, in its graded contempt, exists to prevent.

What did the late lecture refuse in Marxism?

It refused the instruments and the metaphysics while keeping the diagnosis. Violence is refused because a method that selects for the ruthless cannot be trusted to build the gentle society it promises. Dictatorship is refused because power surrendered temporarily has a record of declining to be surrendered back.

The abolition of religion is refused because a just order needs a sacred moral discipline to hold it when obedience becomes expensive, and the task is to replace a religion of graded birth with a religion of equality. And the reduction of mind to matter is refused throughout: unless men are changed inwardly, the lecture argues, no rearrangement of property will change them.

The significance of the 1956 comparison for the caste and class debate is that it fixes the critic’s standing with precision. He was not an anti-socialist who happened to mention caste, nor a Marxist who happened to dissent. He was an independent thinker with a rival theory of the Indian social order, who had tested the imported theory against the object he knew best, found its Indian application wanting in stated respects, and built an alternative whose central categories were annihilation of caste, fraternity and constitutional morality. That standing is what makes the address a theoretical event rather than a historical curiosity. Marxism in India was not criticised from the mission house or the landlords’ press; it was criticised, at the height of its intellectual prestige, by the one Indian thinker whose knowledge of the society’s deepest structure exceeded its own, and the criticism had to be answered in kind, as theory, or not answered at all. The answers, when they came, are the subject of the sections that follow, and they begin with an attempt to save the unity of the theory by re-describing caste itself.

The critic’s economics: the memorandum and the conversion

Two further documents complete the evidence about the critic’s standing, and both embarrass the caricature of him as a liberal gradualist who happened to dislike Marxism. The first is the memorandum he submitted in 1947, on behalf of his political organisation, to the Constituent Assembly, published as States and Minorities, in which the safeguards demanded for the oppressed communities are set inside an economic programme of remarkable radicalism: state ownership of land, with agriculture organised as a state industry; state ownership or control of key and basic industries; a state monopoly of insurance; and the whole entrenched in the constitution itself, placed beyond the reach of ordinary legislative majorities, so that the social revolution would not depend on the continuing goodwill of the classes it dispossessed. The document is the constitutional form of the 1956 lecture’s division of the doctrine. The ends of the Marxist programme, the socialisation of the means of production, are accepted and indeed hardened, made fundamental law rather than party aspiration. The means rejected in the lecture, violence and dictatorship, are replaced by the one instrument its author trusted: law that binds the powerful in advance, drafted by a man who had spent a lifetime watching the powerful escape every restraint that goodwill alone imposed.

The second document is an act rather than a text. In October 1956, at Nagpur, weeks before his death and days before the lecture on Marx and the Buddha reached its audience, Ambedkar formally converted to Buddhism, together with a mass gathering of his followers, in a ceremony that administered new vows renouncing the Hindu gods and the doctrines of caste by birth. The conversion executed, at the scale of a community, the remedy the address had prescribed at the level of theory: the warrant was refused, publicly and collectively, and a religion was adopted whose founding teaching, as its convert read it, knew nothing of graded birth and made fraternity a doctrine rather than a hope. For the caste and class debate, the conversion is the experiment that tests the religious-sanction thesis from inside a life. Its author wagered that the hierarchy’s garrison in the conscience could be evacuated by an act of collective religious will, and the subsequent history of the community he founded, its persistence in education, organisation and self-respect across the generations since, is part of the evidence any verdict on that thesis must weigh. It is equally part of the record that the conversion did not carry its converts out of the economy they inhabited, and that the community’s struggles for land, wages and office continued on the Monday after the ceremony, a sequence the fourth position would have predicted it and the first position never could.

The constitutional settlement: the debate enters the law

The debate’s most durable institutional residue is a legal text, and its clauses deserve reading as the debate’s partial verdict delivered by the state. The constitution of the Indian republic, adopted in 1950 with the address’s author chairing its drafting committee, abolished untouchability by name and forbade its practice in any form, threw open places of public worship and resort, prohibited discrimination on grounds of caste among its other grounds, and authorised the safeguards and reservations whose prehistory in the separate electorates struggle the genealogy section recorded. Whatever is said elsewhere about those arrangements as policy, their presence in the fundamental law is the historical trace of the debate’s central contention having prevailed at the level of the state: that caste was not a private prejudice with which law had no concern, nor a survival that development would launder unassisted, but a structure of public power requiring public prohibition, entrenched beyond the reach of the majorities the hierarchy could ordinarily assemble. The classical position had expected the economy to do this work in its own time. The constitution did it by act of law in the republic’s first year, on the insistence, above all, of the man who had argued that the sequence ran from social annihilation to political possibility and not the other way around.

The sequel is equally part of the record, and it is the sequel that keeps the debate alive as scholarship rather than as commemoration. Legal abolition did not abolish: the wells and temples of the constitution’s promise remained, across much of the country, guarded by the same customs the law had forbidden, and the enforcement of the prohibition became itself a field of struggle in which the oppressed communities learned to wield the text as an instrument, filing, litigating, citing, converting the parchment into leverage village by village. The gap between the settlement’s promise and its enforcement is not, for this article’s purposes, a scandal to be denounced or a hypocrisy to be exposed. It is the debate’s oldest datum in its newest form, the discovery the Mahad satyagrahis made in 1927 and the address theorised in 1936, that a right proclaimed and a right practised are separated by the whole machinery of the social order, and that closing the separation is the work the order’s opponents are actually for. A theory that predicted law would be unnecessary and a theory that predicted law would be sufficient stand refuted together by the same seventy years, which is as compact a statement as the evidence affords of why the fourth position, with its two grammars and its refusal of every shortcut, commands this article’s assent.

One tradition, several voices: the anti-caste inheritance is not a monolith

A caution is owed against a symmetry error into which accounts of this debate, this one included, perpetually threaten to fall: the treatment of the anti-caste side as a single voice answering a single doctrine. The inheritance that runs from Phule through the address to the Panthers contains real disagreements, and the disagreements matter because they determined which parts of the inheritance the Marxist synthesis could absorb. Phule’s quarrel was with the priesthood and its mythology, and his constructive programme was the education and organisation of the cultivating majority, imagined as a unity of the oppressed that included castes the later movement would classify among the oppressors. The Self-Respect tradition in the Tamil country carried the argument to atheism and to a politics of cultural revolution, burning the law books and marrying couples without priests, and held organised religion itself, rather than any one religion’s hierarchy, to be the enemy. The address’s author shared the atheists’ conclusion about the Shastras and refused their instrument, staking his people’s future on constitutional law, on education and, at the end, on a reconstructed religion of equality, a sequence his rationalist critics found timid and his constitutional critics found visionary.

The Panthers, in turn, quarrelled with the inheritance they claimed, accusing the established leadership of the oppressed communities of having converted a movement of annihilation into a machinery of representation, and their split, within a few years of the founding, rehearsed in miniature the debate’s master problem: whether the aim was a share of the existing order’s goods, distributed more justly among its ranked groups, or the abolition of the ranking itself. The two aims are not the same, and the tension between them runs through every politics the debate touches, including the Marxist one, whose parties have faced the identical choice between organising the oppressed as they are, ranked and networked, and organising them against the ranking that constitutes them. Stating the tension inside both traditions, rather than awarding it to one as a vice, is the condition of understanding why the synthesis was so long delayed and why, when it arrived, it arrived as a practice before it arrived as a theory: the practice of movements that needed the land and the well in the same procession could not afford the luxury of choosing which tradition’s purity to preserve.

The first reply: caste as the specific form class takes in India

The most consequential Marxist answer to the critique did not deny the facts the critique had assembled. It re-described them. On this second position, the error of the classical view was to treat caste as a costume worn by class relations that existed independently underneath, when in truth, in India, caste was the form in which class relations themselves existed and had always existed. The correspondence on the ground was too systematic to be coincidental: the castes that held the land were, village by village, the castes that held the ritual rank; the castes that laboured without land were the castes the order placed at the bottom; hereditary occupation meant that a man’s caste announced his relation to the means of production as reliably as any deed or wage book. If class is a relation to production, and if in India that relation was assigned, enforced and inherited through caste, then the struggle against caste hierarchy and the struggle against class exploitation were not two struggles competing for the movement’s calendar. They were one struggle described in two vocabularies, and the village notable who resisted land reform in the name of custom and the village notable who resisted it in the name of property were the same man defending the same position from two directions.

This position had distinguished scholarly carriers. The historical work associated with D. D. Kosambi treated caste as the characteristic social form of class at a primitive level of the productive forces, a mechanism by which tribes absorbed into an expanding agrarian order were frozen into ranked occupational strata, so that the history of caste became legible as a chapter of the history of production itself. The wider school of Indian Marxist historiography that followed, whose account of caste’s origins, the feudalism debate and the periodisation of Indian history belongs to the companion study of the Indian Marxist historians, worked variations on the same thought: caste was no decorative survival but the load-bearing structure of the pre-capitalist formation, reproduced by the relations of land and surplus extraction it organised. In the political literature the position licensed a distinctive practice. Where land reform was fought, where sharecroppers were organised, where the hereditary servants of a village demanded wages in cash rather than grain at the dominant caste’s pleasure, the movement could say, with some justice, that it was fighting caste in the only coin that finally counted, by breaking the material arrangements that made ritual rank payable.

What does it mean to call caste the form of class in India?

It means that the correspondence between ritual rank and material position is not an accident of two separate systems happening to align, but the working of a single system in which birth assigns both together, each certifying the other in the eyes of the village.

On this account, land, occupation and surplus in the Indian village were distributed through caste, so that to describe a man’s caste was to describe his class position with unusual completeness, and to attack the distribution of land and work was therefore to attack the caste order at its foundation rather than at its foliage. The position preserves the unity of the theory by relocating caste from the superstructure into the relations of production themselves.

The strength of the form-of-class position is the strength of the correspondence itself, which the empirical sections below confirm in detail, and no verdict in this article should be read as denying it. Land reform in the regions where it was seriously attempted did weaken the ritual authority of the old dominant castes; the spread of wage labour did give the lowest communities an exit from hereditary service that sermons never had; and the political mobilisation of the labouring castes was built, in the movement’s strongholds, on the material grievances the position predicts. Its weakness appears exactly where the address said it would: in the provinces of life where the correspondence stops running. A theory that identifies caste with the Indian form of class must predict that the man who escapes his class position escapes, in the same motion, his caste position, and the prediction fails at the marriage registrar, at the temple door, at the dining arrangements of men who have worked the same machines for thirty years. It fails equally for the poor of the privileged castes, whose ritual rank survives their poverty intact and whose honour at the village well is not a wage that anyone pays them. The form-of-class position explains why caste and class so often move together. It cannot explain the occasions, numerous and systematic, when they move apart, and an explanation of caste that fails at marriage and worship is failing in the rooms where caste most characteristically lives.

The second reply: autonomy conceded, primacy retained

A third position, reached by a later generation of Indian Marxist writing, conceded much of the critique’s evidence while defending a version of the theory’s architecture. On this account, caste is genuinely autonomous in specified domains: in marriage, where its rules operate with a completeness that economic incentives barely dent; in ritual, where rank is assigned by criteria no market computes; and in the networks of kin and community through which much of Indian economic life in fact flows. Within those domains, caste is to be studied and fought on its own terms, with its own specific history, and the old language of residues and survivals is withdrawn as an embarrassment. What is retained is a claim about primacy at the level of the whole: that the organisation of production sets the constraints within which the other orders operate, that caste’s material effects are ultimately conditioned by the economy that hosts it, and that a politics which forgot production in favour of a purely cultural war on hierarchy would leave the landlords in possession of the village while winning arguments about its customs.

The concession position marks the distance the tradition had travelled, and its importance is easy to understate from outside. The classical doctrine had not merely assigned caste a low priority; it had denied, in principle, that caste could be an object of theory in its own right, a structure with mechanisms of its own that a serious analysis must reconstruct rather than await the dissolution of. To concede autonomy in marriage, ritual and network, and to concede it in print, inside the tradition’s own journals and party schools, was to admit the address of 1936 into the tradition’s working library by the side door, often without the citation that candour would have supplied. The position also generated better empirical work than either of its predecessors, because a scholar licensed to study caste as a real structure, even a subordinate one, asks different questions from a scholar instructed to watch it fade: questions about how endogamy is enforced in a metropolis, about how caste networks allocate credit and jobs in an informal economy, about how rank is renegotiated when a subordinated community grows rich. Those questions, pursued across the later twentieth century, produced the evidence base on which any verdict must now rest.

Where does caste operate independently of class, on the concession account?

In the concession account the independent domains are the ones organised by kinship and sanctity rather than by production. Marriage is the clearest: the choice of a spouse is governed by caste rules whose enforcement survives wealth, education and migration, so that a family’s class ascent leaves its marriage circle almost unaltered.

Ritual standing is the second: rank at the temple, the well and the funeral ground is assigned by birth and cannot be purchased directly by any quantity of income. Social network is the third: the community through which credit, employment information and emergency help actually flow in much of India is the caste community, an institution that economic position influences but does not constitute.

What the concession position could not settle was the meaning of its own retained claim. Primacy, asserted at the level of the whole while autonomy is conceded in every particular domain that can be inspected, risks becoming an article of faith rather than a proposition: if caste is autonomous in marriage, autonomous in ritual, autonomous in network and autonomous in the assignment of degrading labour, the domains in which production’s primacy is exercised grow difficult to point to, and the claim that it nevertheless rules in the last instance invites the question of what evidence could ever count against it. A claim that no evidence can disturb has left the company of empirical propositions. It was this discomfort, felt inside the tradition as much as outside it, that prepared the ground for the synthesis attempts and, eventually, for the fourth position, which drops the grammar of primacy and subordination altogether and asks instead how the two orders produce each other.

The politics of a name: what the oppressed are called

The debate was fought over a word as well as over a theory, and the word’s history compresses the argument. The colonial administration filed the communities beneath the caste order as the Depressed Classes, a bureaucrat’s euphemism that described their condition while concealing its cause. Gandhi, campaigning against untouchability in the 1930s, renamed them Harijans, children of God, a term of tenderness that its intended beneficiaries came to resent as the kindness of the very order that had named them polluted: to be rechristened by the reformer was, in the resentment’s logic, to remain an object of somebody else’s moral programme. The constitution of the republic listed them, for the purposes of its safeguards, as the Scheduled Castes, a term of law, exact and bloodless. The name that broke the sequence was the one the communities chose for themselves in the militancy of the 1970s: Dalit, from a root meaning broken, ground down, oppressed, adopted by the Panthers and the literature that followed them precisely because it named a condition inflicted rather than an essence inherited, and because it was capable of extension, gathering into one political identity all those the order had broken, by caste first of all but by every machinery of oppression the broken recognised as kin.

The naming quarrel is the debate in miniature because each name encodes a theory. Depressed Classes described a misfortune and implied a charity. Harijan described a ward and implied a guardian. Scheduled Castes described a legal category and implied a schedule. Dalit described a verdict and implied a struggle, and the analysis compressed in it, that the condition was made by identifiable arrangements and could be unmade by those it was made upon, is the premise the anti-caste tradition and the corrected Marxism came, by their different roads, to share. Scholars of the debate have learned to read the vocabulary of any text on the subject as its first argument, and the habit is worth acquiring by every reader of this article: before an author has stated a single proposition about caste, the word chosen for its victims has already announced whether the author thinks the condition is a fate, a wound or a crime.

The southern variation: anti-caste politics without the Marxist frame

The debate’s geography contains a control case of great value, in the Tamil country of the deep south, where an anti-caste politics achieved organisation, office and cultural revolution across the twentieth century while owing almost nothing to the Marxist tradition, and where the comparison disciplines every claim made elsewhere in this article. The non-Brahmin movement of the Madras Presidency, organised politically from 1916 in the Justice Party and culturally, a decade later, in Periyar’s Self-Respect movement, identified its enemy with a precision the northern debates rarely matched: the Brahmin monopoly of education, office and sacred authority, exercised over a cultivating and trading majority that the movement taught to call itself Dravidian and to regard the sacred order as an Aryan import imposed on an older and more equal civilisation. Its methods were the address’s methods arrived at independently and pursued further in some respects: the ridicule and public burning of the sacred texts, marriages conducted without priests, the demand for proportional shares of state employment for the non-Brahmin communities, and an atheism preached in the street rather than confessed in the study.

The southern record tests the positions impartially. Against the classical Marxist expectation, the region’s transformation was led not by class organisation but by a politics of caste against caste, which captured the provincial state, rewrote the rules of recruitment to its services and permanently broke the priestly monopoly of the secretariat and the college, achievements the class-first parties of the region conspicuously failed to match in the same decades. Against any conclusion that caste politics suffices, the same record shows the movement’s limits from within: its constituency was the cultivating and trading castes of the middle, its governments were theirs, and the communities at the very bottom of the order, the landless labourers and the untouchable castes of the Tamil country, found the Dravidian revolution’s benefits arriving late, partially and sometimes not at all, a discovery that reproduced, inside the anti-caste camp, the graded order’s oldest pattern, each storey emancipating itself and then defending its landing against the storey below. Ambedkar’s staircase predicts the southern outcome in detail, which is one reason the southern movements’ own intellectuals read him with an attention they gave no other northern thinker. And the communist movement’s southern history completes the lesson, for in Kerala, the region’s other half, the class parties achieved their greatest Indian successes precisely by fusing the grammars, organising the lowest castes’ material and ritual grievances in a single campaign, exactly as the synthesis prescribed. Two neighbouring regions, two routes, and the comparative verdict is the fourth position’s: the grammar that was omitted, in each case, is the one whose absence the record charges.

The synthesis attempts: movements and theories that refused the choice

From the 1970s, the argument between the two traditions stopped being conducted only in books and began to be conducted in organisations, as a generation of activists concluded that the choice between the class vocabulary and the anti-caste vocabulary was itself the obstacle. The emblematic case arose in Bombay in 1972, where young writers and activists, many of them from the communities the order placed lowest, founded the Dalit Panthers, taking their name and some of their manner from the Black Panthers of the United States and their intellectual charter from an explicit refusal to choose between Ambedkar and Marx. Their manifesto defined the enemy in both vocabularies at once: the landlord and the moneylender, certainly, but also the sacred order that had taught the labourer to accept his station, and it defined the constituency correspondingly, as all those oppressed by the combined operation of caste and class power. The organisation burned brightly and split, as intense organisations do, but its theoretical gesture outlasted its organisational life: it demonstrated that the synthesis was not a seminar compromise but a fighting position, held by the very people in whose name both traditions claimed to speak.

The gesture had a prehistory in the regions where the communist movement was strongest. In Kerala and in Bengal, organisers working among agricultural labourers drawn heavily from the oppressed castes had long practised a politics that attacked the landlord’s rent and the landlord’s ritual authority in the same campaign, whatever the official doctrine said about priorities, because the labourers’ own experience refused to separate the hand that took the grain from the hand that barred the temple path. Scholars of those movements have noted how the practice ran ahead of the theory that was supposed to guide it, and how the theory, when it finally caught up, was in part a transcription of what the organisers had already learned. The synthesis, that is to say, was not invented in the 1970s. It was recognised then, named then, and given manifestos then, by a generation that included intellectuals formed inside the oppressed communities themselves, whose presence in the debate changed its sociology as much as its arguments: the objects of the theory had become its authors.

What did the Dalit Panthers attempt to combine?

The Panthers attempted to combine the two traditions organisationally and rhetorically in a single movement of the oppressed. From the Marxist side they took the analysis of exploitation, the identification of landlords, moneylenders and capital as enemies, and the internationalism that linked their struggle to oppressed peoples elsewhere.

From the anti-caste side they took the annihilation of caste as a non-negotiable aim, the authority of Ambedkar as a theorist in his own right, and the insistence that religious sanction and graded dignity were material forces and not decorations. Their experiment’s importance for this debate is that the synthesis was attempted by activists drawn from the communities concerned, not negotiated between two leaderships.

In the universities, the same decades produced theoretical work attempting the combination with more apparatus and less fire. Writers formed in the Marxist tradition began to publish sustained studies of caste as a structure of oppression with its own logic, to be theorised alongside class rather than beneath it; scholars of the anti-caste tradition, among whom Gail Omvedt’s work holds a central place, constructed accounts in which caste, class and gender appear as interlocking systems of exploitation, each requiring the others for its full description. The Dalit movement’s own intellectuals pressed the synthesis from the other side, insisting that any Marxism unable to theorise the annihilation of caste as a primary aim, rather than as a corollary of land reform, had failed the Indian test regardless of its European credentials. By the end of the century the boundary between the traditions had become genuinely porous in the scholarly literature, with Marxist sociologists citing the address of 1936 as a theoretical source and anti-caste writers deploying the categories of surplus and accumulation without apology. What had not yet emerged was a single agreed statement of how the two orders relate, and the fourth position, examined after the evidence, is best understood as the statement toward which that porous literature was converging.

The agrarian debate: the universities argue over the village

The caste and class debate had a second front in the universities, conducted in the driest vocabulary and over the highest stakes, and a reader who knows only the polemics has missed the terrain where the form-of-class position was tested most rigorously. From the end of the 1960s, the pages of the Economic and Political Weekly and the seminar rooms of Delhi and Calcutta hosted the Indian mode of production debate, an argument over how to characterise the agrarian relations of the Indian countryside: feudal, semi-feudal, capitalist, or some colonial formation that answered to none of the classical names. Caste was implicated at every step, because the evidence for the feudal reading consisted largely of the relations this article has been describing: labour attached to landholding households by hereditary obligation rather than by contract, remuneration in customary shares rather than wages, the fusion of economic compulsion with caste duty and caste sanction, so that the labourer’s unfreedom appeared to the feudal school’s analysts as the signature of a pre-capitalist order surviving into the present.

The debate’s dramatis personae divided along lines the grid predicts. Defenders of the semi-feudal reading, among whom the economist Amit Bhaduri’s model of semi-feudalism in eastern India became canonical, argued that the interlocking of land, credit and labour in the person of the dominant landholder, an interlocking that caste organised and sanctified, blocked the emergence of a free labour market and with it the capitalist transformation of agriculture: the attached labourer could not sell his labour to the highest bidder because he was bound, by debt, by custom and by caste, to a single master, and the master had no interest in releasing him into productivity. Critics of the reading, Ashok Rudra prominent among them, replied with survey evidence that the supposedly feudal relations were shot through with wage payment, seasonal migration and hard bargaining, and that the countryside’s problem was not a surviving feudalism but a capitalism developing in a stunted and distorted form. Between the schools stood the uncomfortable fact that both were describing the same villages, and that the variable separating their descriptions was often the weight each assigned to caste: count the hereditary attachments as relations of production and the village is semi-feudal; count them as customary survivals draped over an emerging capitalism and the village is in transition. The caste question, in other words, was not one topic inside the agrarian debate. It was the hinge on which the agrarian debate turned, since the classification of caste decided the classification of the countryside, and with it the strategy of every party that proposed to transform the countryside, a dependence the debaters themselves only intermittently acknowledged.

What did the semi-feudalism thesis claim about caste-bound labour?

The thesis claimed that the attachment of labourers to landholding households, hereditary, debt-bound and sanctioned by caste, constituted a relation of production in its own right, not a cultural survival, so that eastern India’s countryside was governed by a semi-feudal order in which the landholder extracted surplus through usury and control of labour rather than through capitalist investment.

Its critics answered with evidence of wages, migration and bargaining inside the same relations. For the caste debate, the episode’s importance is that both sides treated caste attachment as economically constitutive, a concession the classical superstructure doctrine could never have made.

The long outcome of the agrarian debate prefigures this article’s verdict with unusual fidelity. No school succeeded in classifying the countryside to the others’ satisfaction, and the literature’s mature position, assembled piecemeal across the following decades, is that the Indian agrarian order combined, in proportions that varied by region and decade, the extraction of surplus through caste-bound customary relations and through thoroughly capitalist markets, each mode penetrating and using the other: the dominant landholder who was also a moneylender and a contractor, the labourer who was simultaneously a hereditary servant and a seasonal migrant wage worker, the caste panchayat that functioned as a labour exchange. A formation that refuses the classification offered it is usually announcing that the classificatory scheme, and not the formation, is at fault. The agrarian debate took thirty years to reach that announcement about the Indian village. The caste debate, with which it was secretly one debate, reached the same announcement by the road of the address, the replies and the evidence, and the two literatures’ convergence on a constitutive rather than a classificatory answer is among the strongest corroborations the verdict can offer.

The land question as the debate’s testing ground

If any single arena can adjudicate between the positions, it is land, because land in India is where the two orders are most tightly fused and where their fusion was most seriously attacked. The form-of-class position makes a clear prediction about land reform: break the landlord’s hold, redistribute the acres and record the tenant’s right, and the caste order loses the material spine that ritual merely clothes, so that dominance withers with the dominance’s estate. The autonomy position makes the opposite prediction: redistribute the land and the hierarchy will reassert itself through the channels land reform cannot reach, through marriage, credit, ritual service and the capture of the new institutions by the old locally dominant. The Indian record, across the states that attempted serious reform and the many that attempted its simulacrum, reads like a laboratory report written to frustrate both predictions impartially and to instruct the fourth position in their place.

Where reform was real, in the communist-governed states above all and in pockets elsewhere, its caste effects were exactly as large as the form-of-class position promised and exactly as bounded as the autonomy position warned. The abolition of tenancy and the recording of sharecroppers broke the hereditary dependence of the lowest communities on the dominant households in a way no sermon had achieved; the labourer who held his own plot and his recorded share could refuse the customary services, send his children to school and stand upright in the village in a sense his father would have recognised as revolution. Dominant castes deprived of their tenants’ obedience discovered how much of their ritual precedence had been a property relation all along. And yet the same villages record the hierarchy’s migration into the reform’s own machinery: the panchayat chaired, decade after decade, by the old dominant families in new party dress; the credit and the contracts flowing along the old channels; the marriage circle unmoved by the land records; the temple honours reallocated by acreage nowhere. Land reform changed who could command labour and who could refuse it, which is to say it changed the class structure profoundly, and it changed the caste structure’s material basis with it. It did not change the caste structure’s rules of reproduction, and the structure, given a generation, rebuilt a material basis suited to its rules out of whatever the reform had left standing.

What does the land reform record prove, and what does it leave open?

It proves that the caste order rests in part on material foundations that political action can destroy, since the breaking of tenancy and landlordism measurably broke hereditary dependence wherever the breaking was honestly done, and did so within living memory of the dependence itself.

It leaves open, and in fact sharpens, the question of the order’s other foundations, since the same record shows hierarchy surviving its dispossession by migrating into institutions, networks and marriage markets that the reform never touched. Read whole, the record is the fourth position’s best single exhibit: each simpler theory predicts one half of it and is embarrassed by the other half, and a relation of mutual constitution predicts the whole.

The regional comparison drives the lesson home, because the states supply the control case the debate would otherwise lack. Where land reform was legislated and evaded, which is the larger part of the Indian record, the caste order’s material spine survived intact beneath a generation of class rhetoric, and the rhetoric’s emptiness was registered by the lowest communities with an accuracy the theorists might envy: the demand that rose from those communities, in state after state, was characteristically double, for the land that the class programme promised and for the dignity, the water, the entry and the representation that it postponed. The sequence of Indian politics in the later twentieth century, the rise of parties and movements organised explicitly by the oppressed castes and speaking the language of honour as fluently as the language of shares, is unintelligible unless that doubleness is grasped. It was not a diversion from the material struggle. It was the material struggle’s missing half announcing itself, and the announcement had been drafted, in full, in an undelivered address half a century before.

What counts as evidence: the methods by which the debate was fought

The debate’s positions were carried by methods as much as by doctrines, and a reader who watches the methods can predict the positions with uncomfortable accuracy. The classical theorists worked from the text and the schema: the founder’s writings, the stages of history, the deduction of the Indian case from the general law, a method whose strength is coherence and whose weakness is that the object is consulted last, in the illustrations, after the conclusion has been furnished. The anthropologists worked from the village and the informant, and their caste is correspondingly vivid in its ritual and etiquette and correspondingly innocent of the ledger and the land record, until the materialist fieldworkers corrected the account from below. The historians worked from the archive, revenue settlements, court records, the paperwork of extraction, and produced the caste of the form-of-class position, a structure of land and surplus whose ritual dress appears in their pages mainly when it signs a document. The address’s author worked from a method none of the others could replicate: the testimony of the subjected, organised into theory by a mind trained in the subjected’s own institutions and in the universities of two continents, and the address’s peculiar authority in the debate derives from that method’s scarcity value. It was the one account of the object written from the object’s receiving end with full theoretical equipment.

The mature debate is the one in which the methods check one another, and the field’s best modern work is recognisable by its refusal to let any single method monopolise the object. The correspondence experiment measures what the monograph can only suspect; the monograph explains what the experiment cannot interpret; the archive dates what memory misplaces; and the testimony of the organised oppressed, which the synthesis generation restored to the centre, corrects the systematic bias of every archive and every informant list toward the view from above. A student entering the debate should therefore be suspicious of any conclusion that arrives with only one method’s luggage. The superstructure doctrine was a text’s conclusion. The pure ideology doctrine was an informant’s conclusion. The verdict this article defends is the one that remains standing when text, village, archive, experiment and testimony have each been allowed to cross-examine the others, which is a slower way of working and the only way that has ever worked here.

The overlap: caste and economic position in the evidence

Any verdict in this debate must answer to the empirical record, and the record begins with a correspondence so strong that no honest treatment can minimise it. Across the research literature on Indian villages, towns and labour markets, the same pattern recurs with monotonous regularity: the communities ranked highest in the ritual order are disproportionately the communities that hold the land, command the credit, staff the professions and own the productive assets, while the communities ranked lowest are disproportionately landless, confined to wage labour and hereditary menial service, and excluded from the education that converts labour into income. Village studies from every region describe the mechanism in its local dress. The dominant caste of a village, the community whose numbers, land and muscle set the terms of local life, converts ritual precedence into control of the panchayat, the cooperative and the police station’s ear, and converts that control back into the protection of its land. The lowest communities enter the record as attached labourers, as tenants at will, as the hereditary servants of the dominant households, paid in grain at customary rates that mock the word wage, and disciplined, when they reach for more, by boycott and by the violence that the literature records with a restraint the events do not always deserve.

The urban evidence, gathered later and by different methods, tells a cognate story. Studies of recruitment in the organised sector find caste operating through the referral networks by which jobs actually change hands, so that formal equality of application coexists with a labour market segmented by community from within. Field experiments of the kind conducted by Sukhadeo Thorat and Paul Attewell, in which matched applications differing only in the community signalled by the applicant’s name were sent to employers, found the applications bearing names associated with the oppressed communities and with religious minorities returned at markedly lower rates, a result the authors present as evidence of discrimination operating at the threshold of the modern labour market itself, not at its feudal margins. Wage studies in the same literature find gaps between communities that survive the statistical removal of education and occupation, the residue that analysts attribute to the direct operation of caste in hiring and payment. None of these findings is stated here with figures, in keeping with the protocol of this cluster, and none needs figures to do its work: the direction of the evidence is uniform, and its uniformity across methods, regions and decades is itself the finding.

How strong is the overlap between caste and wealth?

The overlap is strong, systematic and documented by every method the research literature has applied to it, from the village monograph to the field experiment, and it is the permanent truth inside the form-of-class position. Rank predicts landholding, occupation, education and income closely enough that a map of one is a map of the other across the country.

The overlap is nevertheless not identity, and the exceptions are not noise: they cluster in exactly the domains the critique identified, marriage, ritual standing and social honour, where economic position loses its predictive power, which is why the correspondence, however strong, cannot carry the conclusion that caste is class under another name.

The village as laboratory: dominance, service and the hereditary round

The debate’s abstractions were all, at bottom, claims about the Indian village, and the village studies that accumulated across the twentieth century gave the debate its most concrete evidence and its most useful concept. The concept is the dominant caste, formulated by the sociologist M. N. Srinivas from his fieldwork in a Mysore village and elaborated across his studies of social change. A caste is dominant in a locality, in Srinivas’s account, when it combines numerical strength with ownership of the land and, increasingly, with education and political connection, and dominance so defined is a material condition before it is a ritual one: the dominant caste’s precedence is enforced by its ploughs, its grain and its strong arms, and ratified, rather than created, by its standing in the sacred order. The concept performed a quiet revolution inside the debate, because it showed the traffic between the orders running in plain sight. Ritual rank without land and numbers decayed into a courtesy title; land and numbers without ritual rank purchased, within a generation or two, the customs, the priests and the genealogies of a higher standing. Srinivas named that second process Sanskritisation, the adoption by a rising group of the practices of the twice-born, vegetarianism, teetotalism, the seclusion of women, the employment of Brahmin priests, as the group’s application for promotion within the hierarchy it did not dream of abolishing.

Beneath dominance lay the service order the older literature called jajmani, the hereditary round by which the village’s artisans and servants, the carpenter, the potter, the washerman, the barber and the scavenger, served the landholding families in exchange for customary shares of the harvest and customary obligations on both sides. Described by its apologists as a harmony of reciprocity, the round appears in the testimony of its lowest participants as a system of attached dependence, in which the servant family’s share was fixed by the custom its masters interpreted, its right to withdraw its labour was the one right the system most carefully denied, and its degradation was renewed at every festival where the order of precedence was publicly enacted. For the caste and class debate the jajmani village is the perfect exhibit of mutual constitution in miniature. The economic relations, who eats and who serves, are caste relations, assigned by birth and enforced by ritual sanction; the ritual relations are economic relations, since precedence at the feast follows the granary and the granary follows precedence in a circle no participant could have drawn as a line. A theorist who asks which of the two is the base has brought a European question to a village that declines to parse it.

What does the idea of a dominant caste add to the debate?

It adds the missing middle term between ritual rank and material power. Srinivas showed that local supremacy belongs to the caste that unites numbers, land and organisation, whatever its place in the sacred scale, and that ritual standing follows such supremacy as often as it precedes it.

For the debate, the concept disciplines both extremes: against the superstructure doctrine, it shows caste power resting on land and muscle rather than on belief alone, and against the form-of-class doctrine, it shows the resulting dominance being spent, characteristically, on the purchase of ritual rank, a good no class analysis predicts men should want at that price.

The village evidence also corrects a temptation common to all four positions, the temptation to treat caste as a single national system with a single logic. The ladder’s rungs, the identity of the dominant, the severity of the pollution rules and the avenues of escape varied by region and by local history to a degree that embarrasses every general theory in this article, the verdict’s included. In some regions the order’s summit was a priesthood without land, dependent on the patronage of warrior and cultivating castes whose dominance it legitimised; in others, movements of the lowest communities had, by the middle of the twentieth century, broken the service order’s back in fact while its forms persisted in courtesy. The variation is not an objection to theory. It is the data from which any honest theory of caste and class must be built, and it explains why the debate’s best work has so often been done by scholars, of whatever doctrine, who began in a particular village and let the village argue back.

Where the overlap stops: the domains that resist the economic reading

Against the correspondence stands a set of facts, equally systematic, that mark the boundary of the economic interpretation. The first concerns marriage, the institution through which caste reproduces itself. Matrimonial practice in India remains overwhelmingly endogamous, and this is not a pattern confined to the poor, the rural or the uneducated, among whom a survival might be expected to linger. It persists in the matrimonial columns of metropolitan newspapers, in the matchmaking of families whose wealth is new and whose politics are modern, and in the preferences recorded by the marriage market’s own data wherever scholars have examined it. A class relation dissolves, in principle, when its material basis dissolves: the bankrupt landlord’s son is a proletarian, whatever his grandfather was. A caste relation does not dissolve when its material correlates change. The prosperous contractor from an oppressed community who advertises for a bride discovers that his money is welcome in the market for goods and nearly worthless in the market for marriages beyond his own community’s circle, and the discovery, repeated in millions of family negotiations, is the daily referendum in which the autonomy of caste is confirmed.

The second fact concerns the direction of dependence between honour and wealth. In a purely economic hierarchy, honour follows money with a lag: the rich man’s manners are vulgar for a generation and distinguished thereafter. In the caste order, ritual rank can defy money indefinitely in both directions. The impoverished Brahmin priest, dependent on the alms of merchants far richer than himself, retains a precedence at the ceremony that his creditors do not dream of claiming, and his poverty is pitied in a vocabulary entirely different from the contempt reserved for the poverty of the low. At the other end, the wealthy family of an oppressed caste finds that its wealth purchases land, vehicles, education and political office, all of them real, and does not purchase commensality: the invitations, the temple honours, the seat at the common feast remain governed by the older ledger. Honour that can neither be bought by the low nor lost by the high is not an economic variable wearing fancy dress. It is the currency of a second system, exchangeable with the first at rates that vary, but not reducible to it, and a theory with only one currency in its model will misprice every transaction in which the second is used.

Can money buy ritual rank?

It cannot buy it directly, and the indirect routes prove the rule by their ingenuity. Wealthy families of subordinated communities have historically sought rank by adopting the customs of the higher orders, by commissioning genealogies, by endowing temples and by pressing claims to a grander origin, a process the sociological literature calls Sanskritisation.

The process is slow, contested, collective and frequently refused, which is exactly how a boundary behaves when it is not a price. What money buys at once is land, schooling, office and comfort. What it buys across generations, if at all, is a contested re-ranking of an entire community, never the simple transfer of honour to an individual purchaser.

The third fact is the most theoretically interesting, because it complicates the verdict in both directions at once: caste is itself an economic institution, a working piece of the economy’s machinery, and not merely a prejudice floating above it. The caste network is, for millions of Indians, the operative unit of economic life: the channel through which credit is raised without banks, through which a migrant finds his first job and lodging in a strange city, through which apprenticeships are arranged, trading diasporas are staffed and risks are pooled. Whole industries in the subcontinent have been built inside caste networks, whose internal trust lowered the cost of every transaction they conducted, so that the community a liberal economics would classify as a pre-modern survival functions, on inspection, as a competitive advantage in a modern market. This fact wounds the classical position from an unexpected side, since a superstructure is not supposed to allocate capital, and it equally wounds any purely cultural account of caste as a system of ideas, since the network’s members defend it with the tenacity of men defending an asset. Caste persists partly because it is believed, partly because it is enforced, and partly because it works, for those positioned to draw on it, as the most reliable financial instrument they possess. An adequate theory has to hold the belief, the enforcement and the instrument in a single frame, which is precisely what the mutual constitution position attempts and what each of the single-direction theories, base to superstructure or culture to economy, cannot do.

A fourth fact belongs to the politics of dignity, and it is the one the subjects of the system have always ranked first. When the oppressed communities of India organised, their demands were never confined to land and wages, indispensable as those were. They demanded the right to draw water from the common well, to walk the public road, to enter the temple, to send their children to the common school and seat them inside it, to wear the garments and use the honorifics that custom denied them, to be addressed as human beings by other human beings. These demands, pressed from the nineteenth-century movements through the campaigns Ambedkar led at Mahad and elsewhere to the assertions of the present scholarly record’s historical chapters, are unintelligible as class demands in the classical sense, since every one of them could be granted without transferring a rupee of property, and every one of them was resisted as if property itself were at stake. The resisting castes understood the stakes correctly. What was being demanded was the surrender of the status wage, the daily tribute of deference that the graded order paid to everyone above the floor, and the fury with which a procession, a pair of shoes or a wedding horse could be met measures the value of the currency in which that wage was paid. A theory of Indian society that cannot state why a pair of shoes on the wrong feet is an economic event and a theological event and a police event at once has not yet described its object.

The view from the top and the view from the floor: the scholarly quarrel over what caste is

While the political traditions argued, the universities conducted a parallel quarrel that sharpened the terms of both, and no account of the debate is complete without it. Its most influential statement came from the French anthropologist Louis Dumont, whose Homo Hierarchicus, published in French in 1966 and in English four years later, proposed that caste be understood as an ideological system entire: a hierarchy founded on the opposition of pure and impure, in which the religious principle of rank encompasses and subordinates the political and economic, so that power and wealth in India appear as functions of ritual status rather than as its rivals. Dumont’s India was a civilisation that had made a distinctive choice, subordinating the individual and the market to the whole and its sacred order, and his book gave generations of readers a caste that was, above all, a system of ideas in men’s heads, coherent, total and serene.

The materialist rejoinders arrived quickly and from the field. Gerald Berreman, an American anthropologist who had done fieldwork in the Himalayan foothills as well as comparative work on race in the United States, objected that Dumont had described caste as it appeared to those it exalted, and that the view from the bottom of the system was strikingly different: those at the bottom, in Berreman’s fieldwork, did not experience the order as a sacred whole in which they found their honoured place, but as a system of power and deprivation maintained by the superior force of those it benefited, resented in private and complied with in public for want of an alternative. The parallel with the segregated American South was deliberate: in both cases, the ideology of the whole was the anthropology of the dominant, and the subordinated kept their counsel. Joan Mencher, working among agricultural labourers in Tamil country, pressed the point into its sharpest form in an essay on the caste system viewed from below: for the labourer who receives the worst land, the worst wages and the daily instruction of contempt, caste is not experienced as a ritual mystery at all. It is experienced as the mechanism by which his labour is cheapened and his resistance divided, which is to say, as something functionally indistinguishable from class exploitation, whatever the priests may say about purity.

The quarrel matters to the Marxist debate because it reproduced, inside anthropology, the debate’s own alternatives, and because each side captured a fact the other needed. Dumont’s encompassing hierarchy explains what the economic reading cannot: why the order’s categories organise marriage, worship and honour in domains where no extraction is occurring, and why its subjects, including its victims, so often describe it in the vocabulary of purity rather than of interest. Berreman and Mencher explain what the ideological reading cannot: why compliance is so often sullen rather than serene, why the order requires the regular application of boycott and beating to keep its lowest members in their places, and why the labourer’s own account of his situation sounds, when anyone records it, like the account a class theorist would give. A hierarchy held together purely by belief would not need its sanctions; a hierarchy held together purely by force would not need its theology. The Indian order, possessed of both in unusual strength, requires of any adequate theory a double account, and the pressure toward the fourth position in this debate, the position of mutual constitution, is in part the pressure of this scholarly stalemate finding its resolution: ideology and extraction are not rival explanations of caste, competing for the same territory, but the two hands of a single system, each unintelligible without the other.

There is a further lesson in the quarrel about the sociology of knowledge, and the debate’s participants were not slow to draw it. The account of caste a scholar produces correlates suspiciously well with the informants a scholar can reach: the Brahmin literati who opened their texts and their philosophy to the visiting anthropologist were the order’s authorised interpreters, and a science built on their testimony risked transcribing the ideology as the analysis, a risk symmetrical to the one Ambedkar had charged against the communist leadership, whose distance from the bottom of the order had furnished it with a theory in which the bottom’s experience did not appear. The methodological moral generalises beyond this debate. Every theory of a hierarchy is written from somewhere inside it, or from somewhere that the hierarchy has arranged for visitors, and the effort to correct for the vantage point, by seeking the testimony of those the system injures, is not a courtesy extended to the oppressed but a condition of getting the object right. The Marxist tradition possessed, in its own theory of ideology, every tool needed to state that moral. That it applied the tools to the anthropologists sooner than to itself is one of the ironies the verdict will have to carry.

How each order makes the other: the mechanisms of mutual constitution

The fourth position earns its name only if the constitution it alleges can be shown at work, mechanism by mechanism, in both directions. Four mechanisms carry most of the traffic from caste to class. The first is assignment: caste allocates persons to economic positions before any market operates, through hereditary occupation, through the reservation of education and of clean work, and through the exclusion of the lowest from landownership by custom enforced as violence, so that the class structure a census records is, in large part, the caste structure photographed at the level of property. The second is segmentation: caste divides the labour supply against itself, attaching different communities to different tasks at different customary rates and teaching each to guard its niche against the others, which cheapens labour as a whole and stabilises the wage structure an employer confronts. The third is legitimation: the doctrines of karma and of sacred rank persuade the exploited that their position is a moral fact rather than a political arrangement, performing for the Indian propertied order the service that every propertied order requires and usually has to purchase from its ideologists. The fourth is network: caste supplies the trust, credit and information infrastructure through which much production and exchange actually flows, so that the economy does not merely sit beneath caste like a foundation under a house but runs through it, along channels caste dug.

The traffic in the reverse direction is no less concrete. Economic change reshapes caste practice continuously, though it does not abolish caste itself, and four mechanisms again do most of the work. The first is exit: wage employment, migration and the anonymous city loosen the hereditary assignment of work, giving individuals a livelihood outside the village order’s jurisdiction and communities the resources to refuse degrading obligations, as the lowest castes’ abandonment of customary services in region after region attests. The second is accumulation: wealth gathered by a subordinated community is converted, across generations, into temples endowed, customs adopted from the higher orders, schools founded and claims to higher rank pressed, the Sanskritising traffic by which the order’s contents are perpetually renegotiated even while its form survives. The third is organisation: economic resources fund the associations, journals, parties and litigation through which castes act as collective political agents, transforming a ritual category into an electoral and bargaining constituency with demands its members formulate in the language of rights and shares. The fourth is law’s lever: employment in the state, education in its institutions and the protections of a constitution that forbids untouchability give the lowest communities instruments whose use alters, village by village, what the hierarchy can still enforce with impunity.

What separates mutual constitution from mere interaction?

Interaction suggests two finished things colliding, each complete before the encounter; mutual constitution means neither is finished without the other. On the constitutive account, no caste order exists apart from the relations of land, labour and surplus it organises, and no class structure exists apart from the hereditary ranks through which its personnel are recruited and its discipline maintained.

Remove the counterfactual test and the difference sharpens: an interaction claim survives the thought experiment of varying one system while holding the other fixed, while a constitution claim predicts that the experiment cannot be performed, because each system supplies part of the other’s working parts.

Set the eight mechanisms side by side and the character of the fourth position becomes plain. It is not a compromise that awards half the truth to each of the older camps, nor a mere insistence that both factors matter, which no participant in the debate ever denied. It is a claim about the architecture of the object: that in India the relations of production were built with caste components, and the caste order was maintained with economic components, so that an analysis which begins by separating the two and asking which determines the other has made its decisive error in the first step, at the dissecting table, before any evidence is examined. The claim also explains the debate’s own history with uncomfortable economy. Each of the older positions fastened on real mechanisms and mistook its selection for the whole: the classical view on legitimation, the form-of-class view on assignment, the concession view on the domains where assignment fails. The evidence never contradicted any of them in its chosen territory. It contradicted each in the territory it had agreed not to inspect.

The strongest objection, gathered into one statement and answered

Debates conducted across a century accumulate their force in fragments, and it serves clarity to gather the objection against class-first politics into the single statement its author would have recognised, before answering it in the terms this article’s evidence allows. The objection runs as follows. Caste is not a division of labour but a division of labourers, hereditary, graded and sanctified by religion, and it therefore prevents the very thing that Marxist politics in India requires as its premise: a working class capable of recognising itself as one class and acting as one. The labourers are divided not by misunderstanding but by a real distribution of dignity, in which almost every labourer holds a fragment of superiority over another and will defend that fragment against the solidarity that asks him to surrender it. The movement that proposes to lead these labourers compounds the difficulty, being itself commanded by men of the privileged castes who have never undergone the system’s characteristic injury and who ask the injured to wait, for the sake of a unity the uninjured define, until after a victory that the divisions render unattainable. Therefore a politics organised around class, in Indian conditions, is not merely incomplete. It is disabled at its foundation, and caste must be annihilated as a primary and independent aim, by an attack on its religious warrant and its social practice, before or alongside any politics of property, not after it as a dividend of victory.

Three parts of that objection must simply be conceded, because the evidence assembled in this article concedes them. The description of caste is conceded: hereditary, graded, sanctified, dividing labourers, surviving economic transformation in its core domains, the address’s anatomy has been confirmed by a century of scholarship that began by resisting it. The disabling effect on solidarity is conceded as a fact of Indian labour history: caste segmentation inside mills, mines and fields is documented wherever the record has been examined, and the burden of mobilising across it has been paid, in defeat after defeat, by every movement that pretended the burden did not exist. The weakness of the historical communist movement on the question is conceded in the specific form the address charged: a leadership drawn predominantly from the privileged castes, a theory that lagged its own organisers’ practice for decades, and a record of postponing caste-specific emancipation to a tomorrow that its strategy never reached. A defence of the movement against those counts is not available in honesty, and this article does not offer one.

What can be answered is the objection’s implied conclusion, that class analysis and class politics are thereby disqualified in India, and the answer rests on the same evidence. The overlap between caste and economic position is not an illusion of the older theory but the most robust finding in the entire literature: rank predicts property, work, schooling and income across every region and method, and the movements that have most damaged the hierarchy’s material power, the land struggles, the agricultural labour movements, the campaigns against hereditary service, have characteristically been fought with class instruments wielded by organisations that caste analysis alone never built. The combined movements of the synthesis generation, and the regional practices that anticipated them, show the two vocabularies achieving together what neither achieved separately: the demand for the well and the demand for the land arriving in the same procession, each making the other harder to refuse. The objection establishes that class politics alone does not dissolve caste. It does not establish that the annihilation of caste can proceed without confronting the property relations through which caste’s commands are enforced and paid for, and its author’s own career concedes the point in practice: the man who wrote the address spent his later decades drafting constitutions, demanding state ownership of land and industry in his memoranda, and building the legal machinery of a republic, which is to say, fighting on the economic and political fronts his polemic had sometimes seemed to subordinate. The verdict that does justice to the whole record is therefore double-edged by necessity. Against the classical position, the objection is decisive and is not answered. Against the possibility of an Indian Marxism that has learned from it, the objection functions not as a refutation but as a syllabus, and the fourth position is what a Marxism looks like after it has taken the examination and passed parts of it it once failed.

From emulation to assertion: the order’s subjects change their strategy

The scholarship of the later twentieth century registered a change in the strategy of the order’s subjects that every position in the debate had to absorb, and the change is usefully read as an experiment conducted by the oppressed themselves upon the theories offered to them. The older strategy, the one Srinivas codified as Sanskritisation, accepted the hierarchy’s scale of values and sought promotion within it: the rising community adopted the customs of the higher orders, renounced its stigmatised foods and occupations, elaborated a genealogy of worthier origin and pressed its claim to a higher filing in the common table. The strategy worked, after its fashion, for particular communities in particular districts, and its working confirmed the hierarchy even as it relieved the climbers, since every successful emulation paid the scale the tribute of imitation and left the scale itself, with its floor of the polluted and its summit of the pure, exactly where it stood.

The newer strategy, which the literature of the period describes under the heading of assertion, refused the scale. Its marks, recorded by scholars across the regions, were the public celebration of identities the hierarchy had stigmatised, the recovery and honouring of the oppressed communities’ own heroes, deities and histories against the sacred archive that had demonised them, the conversion of the stigma’s symbols into emblems, and the demand for shares of land, office and education framed not as a petition for promotion but as the right of citizens and the arrears of justice. Where Sanskritisation asked the order for a better place, assertion informed the order that its places were forfeit. The shift in strategy corresponds, in the debate’s terms, to the shift in theory from the second position to the fourth: a politics that experiences caste and class as one formation does not apply for promotion within either, and the intellectuals of the assertion generation, many of them formed in the universities the constitution’s safeguards had opened, supplied the synthesis movements with their cadres and the scholarly debate with its most uncompromising contributors. The order’s subjects had run the experiment the theorists were still arguing about, and the result, visible in every region by the century’s end, was that emancipation proceeded fastest where the demand was framed in both grammars at once and slowest wherever either grammar was trusted alone.

The debate’s completion: gender and the reproduction of the order

The synthesis scholarship of the later twentieth century added to the debate the term its founding documents had handled as machinery but rarely as a subject in its own right: the women through whom the order reproduces itself and on whom its costs are disproportionately charged. The historian Uma Chakravarti gave the analysis its durable name in an essay of the early 1990s that subsequent scholarship has built upon, the concept of Brahmanical patriarchy: a structure in which caste and gender are not two systems in alliance but one system with a single reproductive logic, since the purity of the caste, on which its rank depends, can be maintained only by the control of women’s sexuality, and the control of women’s sexuality can be enforced only by the caste’s authority over marriage, kinship and honour. In that structure the woman of the higher castes is guarded as the vessel of the line’s purity, secluded, married early and widowed into civic death, while the woman of the lowest castes is exposed, by the same logic that guards her counterpart, to the labour market’s hardest work and the dominant’s predation, her dishonour being, in the order’s arithmetic, no charge against its account. The graded inequality of caste is thus underwritten by a graded inequality of gender, each caste’s women policed in proportion to the rank the policing defends.

For the debate over caste and class, Chakravarti’s intervention completed a circuit that the founding texts had left open. The address’s author had seen the machinery in 1916, in the disposal of surplus women and the capture of brides, and had named endogamy the essence when his contemporaries named occupation or purity; what the later scholarship added was the systematic statement that the machinery is not a by-product of caste but its means of production, the process by which caste, uniquely among stratification systems, manufactures its own next generation in the required proportions and ranks. A class system can recruit. A caste system must breed, and the breeding is organised through women whose subordination is therefore not an additional oppression laid alongside caste but the load-bearing wall of the edifice. The mutual constitution thesis absorbs the point without strain, since it predicts that the systems it studies will be found interlocked at the level of reproduction itself, and the feminist scholarship of caste stands in retrospect as the debate’s third great correction, delivered, like the first two, by the testimony of those the master categories had treated as a detail: first the labourer divided by caste, then the scholar from the oppressed communities, then the women whose guarded marriages were the wall everyone else had described as scenery.

Two literatures, one structure: the parallel with intersectionality

There remains the comparison that this article exists, in part, to state, because it connects the Indian debate to a literature that developed independently, in another country and another vocabulary, and arrived at a structurally identical destination. In the United States, the legal scholar Kimberlé Crenshaw, writing at the end of the 1980s, gave the name intersectionality to a criticism of single-axis analysis: the practice of treating race and gender as separate and additive systems, so that the situation of a person standing at their crossing, a Black woman in her founding examples, disappears from both analyses, too specific for the theory of race, which had modelled itself on men, and too specific for the theory of gender, which had modelled itself on the white and the propertied. The remedy Crenshaw proposed was not to add the axes together more carefully. It was to recognise that the systems constitute one another’s operation, that the crossing is not a sum but a distinct location with its own structure, and that any analysis which begins by separating the axes will misdescribe the location no matter how refined the separated analyses become.

The structural resemblance to the Indian debate is exact, and it is not superficial. The additive habit that intersectionality criticises, in which race is taken first and gender added as a modifier, is the same habit the classical position exhibited in India, in which class was taken first and caste added as a survival, a modifier, a lag. The constitutive alternative is likewise the same: the claim that the systems make each other, that there is no Indian class relation to be found underneath its caste form awaiting separate description, just as there is no American gender relation that exists in abstraction from the racial order through which it is lived. Even the polemical histories rhyme. In both cases the constitutive claim was forced into the theory from below, by the testimony of those stationed at the crossing, who reported that the single-axis account did not describe their lives: the Black women whose lawsuits failed in American courts because their experience matched no protected category as the categories were drawn, and the Indian labourers whose humiliation matched no class category as the categories were preached. And in both cases the older custodians of the master category received the correction, at first, as a distraction from the real struggle, a phrase whose career in the two literatures would repay a study of its own.

How far does the parallel with intersectionality run?

It runs through the logic of the positions and stops at the materials. Logically, the two debates are one: single-axis analysis, additive correction, constitutive conclusion, forced in each case by the testimony of people located where the axes cross.

Materially, the systems differ in everything that fills the logic: caste is hereditary, endogamous and religiously sanctioned in a way that neither race nor gender in the American case exactly reproduces, and its mechanisms of assignment and enforcement are its own. The parallel is therefore a parallel of structure, not an identity of subjects, and it is stated here as the series treats intersectionality and class in its own right; the value of placing the two side by side is that each literature supplies the other’s missing articulation of a shared mistake.

The two literatures, as the brief of this debate observes, do not cite each other, and the silence is worth a paragraph of diagnosis because it is itself a finding about how theory travels. The American debate was formed in law reviews and in the feminisms of the United States, behind the linguistic and institutional walls of the Anglophone academy at its most metropolitan; the Indian debate was formed in party schools, in the journals of a subcontinental left and in the writings of a movement whose central texts were printed cheaply in Bombay and Delhi for readers the metropolitan academy never imagined. Each side had reasons, of audience and of polemical convenience, to treat its own case as the universal one, and the theory of intersecting structures was accordingly discovered twice, in mutual ignorance, by traditions that would have profited from the recognition that they were working the same seam. There is a standing lesson in that duplication for any reader of this series. The correction of a master category by the people it misdescribes is not a local event. It is a recurring episode in the history of emancipatory theory, and the traditions that survive it are the ones that learn to treat the correction as evidence about the category, rather than as an attack on the cause.

Lines and gardens: caste in the plantation and the mill town

The industrial evidence deserves a second examination at closer range, because two workplaces became the debate’s favourite exhibits and each repays a slower reading. The tea plantations of Assam recruited their labour, across the nineteenth and early twentieth centuries, through chains of contractors reaching into the tribal and low-caste districts of central India, and housed the recruited families in lines, the barrack rows of the garden, where the community of origin supplied the line’s internal order, its marriages and its discipline, while the management supplied the wage, the task and the coercion. Planters and recruiters alike understood the workforce in caste and community terms and organised it in those terms, so that the most thoroughly capitalist enterprise in the Indian countryside, producing for a world market with wage labour, ran its labour regime through the hereditary categories the classical position expected capitalism to dissolve. The labourers’ own resistance, when it came, organised through the same channels, and the garden’s history is unintelligible in either vocabulary alone: the wage relation explains the grievance, and the community explains the solidarity, and the two explanations meet in every recorded strike.

The mill town supplies the mirror image. In the textile cities, the workforce was recruited through the jobber, the intermediary whose authority over hiring rested on his place in the webs of village, kin and caste that connected the mill to its labour catchments, so that entry to the most modern employment in the economy was governed by the oldest networks in the society. Departments within the mills acquired caste complexions that persisted across generations of technology, the cleaner and better-paid processes reserving themselves, in effect, to communities of higher standing, while the dirtiest work of the industrial city, the sweeping, scavenging and carrying on which its sanitation depended, remained the hereditary portion of the communities assigned to it in the village, now wearing municipal uniforms. Historians of Indian labour have documented these patterns across the major centres, and their cumulative force is the same as the plantations’: the Indian working class was not formed by the dissolution of caste in the crucible of industry and then, regrettably, recaptured by it. It was formed through caste, recruited by it, housed by it, divided and united by it, from the first shift onward, and a class analysis that begins after the forming has missed the forming, which is where the address of 1936 had insisted the whole question lay.

Segmentation: the corroboration from an unexpected quarter

The mutual constitution thesis does not stand on the Indian literature alone, and one corroboration deserves record because it arrived from a tradition with no stake in the quarrel. Labour economics, in its mainstream development across the later twentieth century, abandoned the picture of a single labour market clearing at a single wage and replaced it with accounts of segmentation: labour markets divided into compartments with different rules of entry, different securities and different prospects, between which movement is restricted and within which wages are set by custom and power as much as by productivity. The vocabulary was developed for the industrial economies of the West, for the division between the secure core and the precarious periphery of their workforces. Read against the Indian evidence, it describes a special and extreme case with embarrassing precision. The Indian labour market is segmented by a principle older than any employer, hereditary and sanctified, which assigns the compartments in advance, restricts movement between them by sanctions no personnel department could devise, and legitimises the wage differentials the segmentation produces as the moral order of the world.

The corroboration cuts in a direction neither camp in the Indian debate fully anticipated. Against the classical position, it shows that segmentation is not a friction that growth erodes but a structure that growth operates through, in the advanced economies as in the Indian one, so that the expectation of dissolution by development was never supported even by the economics of the societies development was supposed to imitate. Against a purely cultural account of caste, it shows that the mechanisms the Indian order perfected, restricted entry, customary wages, network hiring, hereditary niches, are the ordinary machinery of labour markets everywhere, differing in their warrant and their rigidity rather than in their kind. The Indian peculiarity, on this comparative view, is not that its labour market is segmented but that its segmentation was theorised, sanctified and inherited, given a scripture and a marriage system where other societies improvise theirs from prejudice and convenience. A debate that began with the question whether caste was a superstructure ends with the discovery that caste was, among its other identities, the most complete labour market institution of the pre-modern world, and that the science of labour markets caught up with the address’s author some decades after he was buried.

The city as a solvent that worked by halves

Every position in this debate made a prediction about the Indian city, and the city’s career is therefore a convenient final examination of them all. The classical expectation was metropolitan: the city would anonymise. The migrant who arrived by train entered a labour market that asked for his hands rather than his horoscope, a housing market that sorted by rent rather than by rank, a politics that counted heads rather than pedigrees, and he would shed, in a generation, the hierarchies of the village he had fled. Something like that shedding genuinely occurred, and the record should not be grudged: urban employment gave the oppressed their first incomes independent of the dominant households, urban schools educated their children beyond the village’s permission, urban anonymity let a man eat, travel and worship without the daily audit of his neighbours, and the great movements of the oppressed were disproportionately conceived in cities, printed in cities and financed from cities. The solvent was real, and the classical position’s error was not to have imagined it but to have mistaken a solvent for an acid bath.

For the city also rebuilt what it dissolved, in forms adapted to its own machinery. Residential segregation reproduced the village’s geography in the metropolis’s wards, with the oppressed communities concentrated in the districts of worst services and the hereditary occupations of filth reappearing as municipal employments staffed, generation after generation, by the same communities the village had assigned to them. The marriage market industrialised without liberalising, its newspaper columns and, in the digital successor the scholarly literature now examines, its platforms sorting suitors by caste with a precision the village matchmaker might envy. Caste associations found in the city their headquarters, their printing presses and their scholarship funds, and the federations that lobbied the modern state for their communities’ shares were urban organisations staffed by the educated sons of the very modernisation that was supposed to have dissolved them. The city’s verdict on the debate is the fourth position’s verdict in concrete: economic transformation changed caste’s instruments, weakened its cruellest enforcements and left its structure of reproduction intact, which is neither the classical prediction of dissolution nor the autonomy position’s prediction of indifference, but the reciprocal remaking of each order by the other that the mutual constitution thesis describes.

The school as the debate’s second laboratory

Beside the land, the debate’s other great laboratory was the school, because education in India was never a neutral good distributed by merit. It was, historically, a caste monopoly defended as such: the sacred learning reserved to the twice-born, the village schoolmaster drawn from the literate castes, the child of the oppressed communities seated, where admitted at all, on the veranda outside the classroom, within earshot of a lesson the order forbade him to join. The first demand of every anti-caste movement from Phule onward was therefore a school of their own, and the demand’s priority in the sequence, educate before agitate and organise in the famous triad the address’s author gave his followers, was a theoretical position in itself: that the order’s reproduction ran through the distribution of knowledge, and that a community armed with literacy possessed the one instrument with which every other instrument could be seized. Phule opened schools for the lowest castes and for girls in the teeth of orthodox fury in the middle of the nineteenth century; the princely reformers of Kolhapur and Travancore, whose roles the series examines elsewhere, built state provision against the monopoly; and the mission schools, whatever their evangelising purpose, taught generations whom the village school had turned away, an irony the debate records without embarrassment, that the order’s educational monopoly was first breached at scale by institutions answerable to neither the classical theory nor the national movement.

The Marxist tradition’s relation to the school recapitulates its relation to the whole question. Its cadres were themselves products of the literate castes’ near-monopoly, as the sociology of the movement’s command established, and its programme treated education, like caste, as a good that the transformation of production would universalise in due course. The record of the movement’s strongholds complicates the postponement in the usual direction: where its governments held office for long periods, the expansion of schooling to the oppressed communities was among their most concrete achievements, pursued as a caste measure and a class measure indistinguishably, because in the village the empty bench and the empty stomach were the same family’s twin deprivations. And the school’s own internal order supplied the debate with its most intimate evidence, for the educated of the oppressed communities, the teachers, clerks and graduates whom the schools produced, became the synthesis generation’s intellectuals and the assertion movements’ organisers, the living proof that the instrument, once seized, reproduced not the hierarchy but its opposition. A theory of Indian society that cannot explain why a school bench was worth a satyagraha has missed the place where the two grammars of this debate are taught, daily, to children who will spend their lives deciding what to make of them.

The textbook afterlife: how the argument entered the syllabus

The debate’s final institution is the syllabus, and its presence there explains the form in which most readers now meet the argument, as a set topic with set texts and a standard shape: the definition of caste, the classical position, the address, the replies, the verdict the examiner expects the candidate to reach and to qualify. The address’s journey from a refused speech to a prescribed text is the kind of irony its author, who distrusted all canonisation including his own, might have savoured with reservations. Canonisation preserves and flattens in a single operation. It preserves, because the student who would never open a party pamphlet or a movement journal now reads the division of labourers argument in a university anthology, in the company of Dumont and Srinivas and the field’s other monuments, and the argument’s survival into a third century is thereby made considerably more likely. It flattens, because the syllabus form converts a fighting intervention into a position to be summarised, its anger edited into a doctrine, its author into a thinker, a respectful category he spent a lifetime refusing to let anyone rest in. The examination industry that has grown around the topic, and whose needs this article’s companion guides serve, is the newest institution to reproduce the debate, and it reproduces, as its predecessors did, both the argument and the argument’s domestication.

There is nevertheless a defence of the syllabus that the debate’s own logic supplies. Every correction this article has recorded entered the tradition through an institution that stood partly outside the tradition’s command: the university against the party school, the independent journal against the party press, the movement of the oppressed against both. The syllabus is the point at which the correction, having won in those institutions, is deposited in the common inheritance, available to readers who owe none of the combatants anything and who will test the positions against a world the combatants did not live to see. What the next correction will be, no participant in the present debate can know; the mechanisms this article has catalogued suggest only where it will come from, from the testimony of those the present settlement misdescribes, and that it will sound, to the settlement’s custodians, exactly as the address sounded in 1936: like an outsider’s exaggeration of a difficulty the theory had already, in principle, solved.

Why the correction took a century: the institutions that carried the argument

A word is owed to the debate’s own sociology, because the slowness of the correction is a fact that itself requires explanation, and the explanation reflects on every tradition that prides itself on correcting its errors. The classical position was not maintained for decades by stupidity. It was maintained by the institutions that reproduced the tradition: the party school with its syllabus, the pamphlet with its print run, the study circle where the doctrine arrived as a finished inheritance and departure from it cost standing, friendship and sometimes livelihood. A theory embedded in an organisation acquires the organisation’s immune system, and the questions the address posed were, for two generations, processed by that immune system as attacks to be repelled rather than as evidence to be weighed. The correction, when it came, arrived through the institutions that stood partly outside the immune system’s reach: the universities, where a scholar could follow the village evidence past the doctrine’s boundary without expulsion; the independent journals of the Indian left, where the agrarian debates of the 1970s were fought in print with a freedom the party press rarely permitted; and the movements of the oppressed themselves, whose electoral and organisational weight eventually made their testimony impossible for any serious tradition to refuse a hearing.

The pattern generalises beyond this debate, and the series in which this article stands has recorded it in other theatres: a doctrine’s custodians are the last to audit the doctrine, not because they are corrupt but because the audit threatens the capital, institutional and personal, that their custodianship represents. What distinguishes the traditions that survive such audits is not the purity of their founders but the presence, somewhere in their ecology, of institutions where the evidence can accumulate in safety until it becomes undeniable. Indian Marxism possessed such institutions at its margins, in its universities and its independent journals and the movements it half-led, and the caste debate is, among other things, the record of the margins correcting the centre across three generations. The correction’s content is the fourth position. Its form, the slow permeation of a tradition by the testimony of those it misdescribed, is the form of every genuine correction the wider tradition has undergone, and a reader who grasps the form will recognise the next correction earlier than the custodians do.

The boundary of this article: the policy question it does not decide

One neighbouring question has been deliberately bracketed throughout, and honesty requires the bracket to be made visible. The constitutional order of the Indian republic contains, as a matter of historical fact, a system of safeguards and reservations for the communities the caste order oppressed, whose origins lie in the struggles this article has recounted, in the separate electorates controversy, the Poona Pact and the drafting work of the republic’s first law minister. Those arrangements have been debated in India, in legislatures, courts and elections, with an intensity proportionate to their consequences, and every position in the theoretical debate has been conscripted, at one time or another, into the policy argument, usually in a simplified form its authors would not have signed. This article takes no position on the policy questions, which are not settled by the theoretical verdict and are not this article’s business. What the theoretical debate does settle, and what a citizen may legitimately carry from it into the policy argument, is a discipline of description: that caste is a real and self-reproducing structure and not a prejudice awaiting enlightenment; that its material and its honorific operations are inseparable; and that any policy addressed to only one of its grammars should expect to be outflanked by the other. Beyond that discipline, the article’s grid grades theories, not measures, and the reader who respects the boundary will get more from both debates than the reader who collapses them into one.

The archive’s silence: whose testimony the record preserved

A last caution belongs to the evidence itself, because the debate was conducted over an archive with a shape, and the shape favoured some positions over others for reasons that had nothing to do with their truth. The oppressed communities of the caste order were, for most of the period surveyed, the least literate population in India, deliberately kept so, and their testimony survives disproportionately in the records of their opponents and administrators: in police reports of their disturbances, in the petitions they dictated to scribes, in the missionary’s journal and the collector’s tour diary, in the folklore collected by enthusiasts whose sympathy did not extend to surrendering the editorial pencil. The classical position could accordingly be maintained for decades in perfect scholarly good faith by men who read everything the archive offered, because the archive had been built by the order it described, and the testimony that would have refuted the position at a glance, the testimony of the subjected about what their subjection consisted in, was the testimony the archive was designed not to keep.

The recovery of that testimony is one of the scholarly achievements on which the modern debate rests, and it was accomplished by the methods the methods section catalogued: by reading the hostile records against their grain, by collecting the oral traditions and the songs in which the subjected kept their own archive, by the rise, within the universities, of scholars descended from the communities themselves, whose questions differed from their predecessors’ because their grandmothers’ experience did. The address of 1936 stands at the head of that recovery as its founding document, which is the final reason for its authority in this debate. It is not merely an argument about the oppressed written with unusual force. It is the moment the archive’s silence was broken from inside, by a witness with the equipment to make the testimony stick in the republic of letters, and every position formulated before it must be graded with the discount its absence imposed, as a map drawn before the territory’s inhabitants were allowed to describe it. The verdict this article has defended was reachable only after that discount was applied, and the application, rather than any single argument, is the debate’s deepest event.

The four positions grid

The argument of the preceding sections can be set out in a single instrument, the four positions grid, which arrays the debate’s four answers to its opening question and records, for each, the claim, the company it kept, its prediction about the effect of economic change, and the objection that history entered against it. The wider tradition within which the debate sits, its parties, movements and strongholds, is mapped in the series’ complete guide to Marxism in India, the companion to this article on the caste question. Read down the columns and the grid is a map of a century; read across any row and it is the biography of an idea, from its formulation through its evidence to the fact it could not digest.

Position	The claim	Principal holders and carriers	What it predicts about economic change	The objection that tells against it
Caste as superstructure	Caste is an ideological survival of feudal relations, produced by a superseded economy and expected to wither as that economy changes	The early communist movement in India, and the doctrine of the economic interpretation of history as it was preached in the movement’s first decades	Markets, factories, railways and wage labour dissolve hereditary rank as a by-product of their ordinary operation	Caste reproduced itself inside the mills, survived conversion to religions that deny it, and flourished as a principle of democratic mobilisation
Caste as the form of class	Caste is the specific shape class relations take in India, since birth assigns land, work and rank together, so caste struggle and class struggle are one struggle	The form-of-class theorists, and the Marxist historiography that read caste as class at an early level of the productive forces	Breaking the material arrangements, above all land relations, breaks the caste order at its foundation	The correspondence fails in marriage, ritual and honour, where escaped class position does not purchase escaped caste position
Caste as autonomous	Caste governs whole domains, marriage, ritual and network, by its own logic, whatever the organisation of production may be	The address of 1936 in its theoretical core, and the current of scholarship that treats hierarchy as a sacred order in its own right	Economic change leaves the core domains substantially untouched, since their warrant is religious and their mechanism is endogamy	Caste also allocates land, labour and credit, and its political defeats have repeatedly been inflicted by economic and legal instruments
Caste and class as mutually constituting	Each order produces and maintains the other: caste assigns and disciplines the class structure, and economic relations perpetually reshape caste practice	The synthesis generation, the movements that combined the two traditions, and the position toward which the later scholarship converges	Economic change transforms caste’s operation without abolishing it, and caste relations determine the form economic change itself takes	It forfeits the clean predictions of its rivals and must be shown mechanism by mechanism rather than asserted as a slogan

Two cautions govern the grid’s use, and both are lessons of the debate itself. The first is that the rows are positions, not parties: each has been held, at different moments, by people inside every camp, and the same movement has preached the first row in its theory while practising the fourth in its organising, a doubleness the grid is designed to expose rather than to hide. The second is that the grid grades arguments, not worth. The first row is wrong in its prediction and was held, in its day, by brave and serious people working with the evidence their moment supplied; the fourth row commands this article’s assent because it digests the most evidence, not because its holders were the debate’s heroes. A student who uses the grid to memorise which side to cheer has misunderstood it as completely as a student who ignores it, and the companion study tools exist, in part, to keep the rows in their proper employment as instruments of analysis.

Misreadings that this debate attracts

The debate’s difficulty is compounded by a set of misreadings so regular that correcting them has become a genre of its own, and five deserve explicit correction here. The first is that Marxists deny the existence of caste. No serious holder of any position in this debate has denied it; the classical position, at its most reductive, argued about caste’s causal standing and its expected lifespan, and the argument was wrong in interesting ways that a denial could not have been. The misreading survives because it converts a theoretical defeat into an accusation of blindness, sparing the accuser the labour of the actual refutation.

The second misreading is that the critique of 1936 came from outside the left, from a liberal or a communal standpoint indifferent to exploitation. Its author shared the socialist end of a society without exploitation, proposed state ownership beyond the programmes of the parties he criticised, and addressed the socialists as erring comrades in an argument about means, a fact the lecture of 1956 places beyond reasonable dispute. The third misreading is that the classical superstructure position is still the position of Indian Marxism, a survival that flatters its opponents at the cost of the record: the tradition’s own scholarship abandoned the doctrine across the later twentieth century, and the interesting questions now concern the successor positions. The fourth misreading infers from the overlap of caste and class that the two are identical, an inference the marriage and ritual evidence refutes daily; correlation between two orders, even at the strength the Indian evidence shows, is a relation between two things and not the collapse of one into the other. The fifth misreading, the mirror of the fourth, holds that because caste is not reducible to class, economic relations are irrelevant to it, and caste can be fought as a pure disorder of belief. The sanitation worker’s hereditary summons, the landless labourer’s customary wage and the dominant caste’s grip on the village’s land are not disorders of belief, and a politics that treats them as such will leave the order’s paymasters undisturbed in the name of its theology.

The sixth misreading is the sophisticated one, and it is directed at this article’s own verdict: that mutual constitution, by making everything the cause of everything, explains nothing and forbids nothing, a formula under which any evidence can be absorbed and none can embarrass the theory. The objection would be fatal if the thesis were the slogan. Stated with its mechanisms, it is neither empty nor permissive. It forbids the classical prediction, that development dissolves caste, and the autonomy prediction, that caste proceeds indifferent to the economy beneath it; it requires any account of an Indian labour market, land system or marriage market to exhibit the mechanisms of assignment, segmentation, legitimation and network, or the mechanisms of exit, accumulation, organisation and law, and it stands refuted in any domain where one order is shown to vary while the other, and all eight mechanisms, demonstrably sleep. A thesis with forbidden outcomes and named mechanisms is a research programme, not a mood, and the debate’s next century will grade it, as it graded its predecessors, by whether the mechanisms continue to be found where it says they will be.

The seventh misreading concerns the critic’s alternative, which is regularly summarised as a cultural programme, an affair of religion and manners set against the Marxists’ materialism, as if the man who demanded the annihilation of caste proposed to pray the landlords away. The documentary record refutes the summary at every point where it can be checked. The memorandum of 1947 proposed the state ownership of land and of key industry entrenched in fundamental law; the labour party of 1936 campaigned on land and nationalisation beside annihilation; the constitutional work armed the republic with prohibitions and safeguards meant to be wielded, not admired. What distinguishes the alternative from the programme it opposed is not a preference for culture over economics. It is the refusal of a sequence: the insistence that the social order be attacked in the same campaign as the economic one, by the same movement, with the same seriousness, because the two orders are one formation and a victory in a single grammar is a loan the other grammar will recall. Readers who carry that refusal, rather than the caricature of a saint of social reform, out of this article will have understood the critic as his opponents, in their best moments, understood him: as the most dangerous kind of critic a movement can acquire, the kind who shares its destination and has audited its map.

The verdict: the mutual constitution thesis

The verdict of this article can now be stated without remainder, and it is the position this series names the mutual constitution thesis. The classical position, which treated caste as a superstructural effect of economic relations, is not supported: its predictions failed in the factory, in the convert congregation and at the ballot box, and the tradition that produced it has itself withdrawn it. The identification of caste with the Indian form of class captures the strongest single fact in the evidence, the correspondence of rank with property, and fails on the second strongest, the independence of caste in marriage, ritual and honour. The autonomy position captures that independence and cannot digest the mechanisms by which caste allocates the material world. What remains standing, after each position has been given its strongest form and its hardest evidence, is the claim that caste and class in India constitute one another: that the class structure was built with caste components, through assignment, segmentation, legitimation and network, and that the caste order has been perpetually remade by economic change, through exit, accumulation, organisation and law, without at any point being dissolved by it. Neither order is the other’s shadow, and neither is the other’s master. They are, in the Indian record, a single historical formation with two grammars, and the debate’s century is the time the theory took to learn to read both grammars at once.

The thesis carries consequences for how the neighbouring questions should be approached, and the article closes its argument by drawing three of them. For theory, it retires the question that organised the debate’s first half, the question of which order determines the other in the last instance, and replaces it with a research programme: identify the mechanisms, domain by domain, by which each order produces the other’s effects, and grade every general claim by the mechanisms actually exhibited for it. For politics, it validates neither the postponement the address condemned nor the indifference to property that a purely cultural anti-caste politics risks; it instructs any movement that would abolish the hierarchy to fight on both grammars simultaneously, against the landlord’s rent and the priest’s warrant, the customary wage and the marriage rule, in the knowledge that each defends the other and that a victory confined to one grammar will be repossessed by the other in a generation. For the series of which this article forms a part, it closes the Indian statement of a pattern the reader has met before and will meet again: a master category, carried into a society it was not built for, corrected by the testimony of those the category misdescribed, and rebuilt, by the correction, into an instrument stronger than the one that arrived. That pattern is not a refutation of the tradition. It is, on the evidence of this debate, the way the tradition has always renewed itself when it renewed itself at all, and the address of 1936, written by the tradition’s critic and shelved for decades by its custodians, stands in retrospect as one of the tradition’s own founding documents in India, admitted to the canon by the only door that has ever finally mattered: the door of being right about the object.

The questions the verdict leaves open

A verdict that closes every question it touches should be distrusted on principle, and this article’s closes by naming the three it cannot settle. The first concerns the boundary of the object: the tribal communities of India, the adivasis, stand in an ambiguous relation to the caste order, ranked by it where they adjoin it, yet organised by kinship and territory rather than by varna in their own internal life, and the movements that organised them drew on Marxist vocabularies of class and on languages of indigenous dispossession that neither the caste debate nor the class debate supplies complete. Whether the mutual constitution thesis extends across that boundary, or whether a third order awaits its theorist there, is a live question in the scholarship, and the honest answer is that the debate surveyed here posed it more often than it answered it. The second question is comparative: whether mutual constitution is the general form of the relation between a master economic category and a hereditary status order, or a specifically Indian settlement forced by caste’s peculiar machinery of endogamy and scripture. The American parallel examined in this article suggests the pattern is wider than India; the machinery suggests it is not identical anywhere else; and the comparative work that would decide the matter is only beginning to be written by scholars willing to hold the two literatures in one head. The third question is the agent’s question, and it is the oldest in the debate. If the order is annihilated neither by economic change nor by preaching, but by organised attack on both its grammars at once, the identity of the attacker remains undertheorised: the oppressed communities themselves, whose organisations the record shows divided by the same gradations they oppose; the state, whose machinery the hierarchy has historically staffed; or alliances of the kind the synthesis movements improvised, whose durability their own histories question. The address’s author answered with a sequence, educate, agitate, organise, and with a lifetime’s demonstration of what the sequence costs. The debate he founded has yet to produce a better answer, and a verdict that pretended otherwise would fail the standard of candour he set it.

A fourth open question is one of measurement, and it is the question on which the thesis will ultimately be graded. Mutual constitution names mechanisms, but the debate’s hard sciences measure magnitudes: how much of an outcome, in a given village, industry or generation, flows through the caste mechanisms and how much through those of property, and how the proportions shift as law, markets and organisation press on them. The older positions were, whatever their errors, committal on this point: they predicted proportions, and the proportions refuted them. The synthesis will earn its place in the same currency or not at all, by generating accounts of particular places and decades in which the mechanisms’ relative weights are argued from evidence rather than assumed from sympathy, and revised when the evidence returns a different answer. Until that literature exists in bulk, the verdict of this article should be held the way its own method prescribes: as the best-supported reading of the record assembled so far, stated plainly, exposed to refutation, and forbidden the comforts of a slogan. The debate has buried better-armed theories for less.

For students and exam candidates

Caste and class is among the most examined topics in Indian sociology and political science, and the scripts that score are the ones that treat it as a debate with positions rather than as a fact with a definition. Learn the four positions as a sequence: the superstructure doctrine and the evidence that buried it, the form-of-class reply and the domains where it fails, the autonomy the address established, and the mutual constitution thesis that the strongest answers now defend. Quote sparingly and exactly: the division of labourers sentence and the monster sentence are worth more than a page of summary. Keep the four positions grid as a revision card in the VaultBook study notes for this series, work the question bank in the series guide to Marxism in India exam answers, and sharpen the answers against the previous year questions gathered at ReportMedic’s UPSC previous year question explorer, where the examiners’ own phrasing shows how often this debate is set and in what forms.

For teachers and seminar leaders

The four positions grid is built for the classroom: set it as the spine of a two-session sequence, with the first session reconstructing the classical position in its period strength and the second working through the critique and the replies, and require students to argue the row they disagree with before they are permitted to defend their own. The address rewards direct handling; a seminar that reads its central pages aloud, in sequence, will find the division of labourers distinction doing more work in an hour than a week of secondary summary. Handle the hierarchy itself with care in the room, since the seminar will contain students whose families stand at every level of the object under discussion, and the discipline the topic teaches, describing a system of humiliation exactly without either euphemism or relish, is itself one of the things being taught. Pair the grid with the intersectionality material for the comparative session, and let the class discover the structural parallel before it is named for them.

For researchers, historians, and journalists

The primary texts are short and decisive: the address in its published form with its prologue and the appended exchange with Gandhi, the lecture of 1956 on Marx and the Buddha, and the early numbers of the movement’s journals for the doctrine the address attacked. For the critic’s own position in full, the companion study of Ambedkar and Marx in this series carries the detailed comparison. The scholarly quarrel between Dumont’s ideological reading and the field rejoinders of Berreman and Mencher remains the fastest route into the anthropological literature, and the empirical programme associated with Thorat and his collaborators, village studies, wage decompositions and the correspondence experiments, is the fastest route into the evidence on discrimination’s mechanisms. Journalists covering the politics of caste will find the grid a serviceable check against the two standing errors of deadline writing: treating every caste mobilisation as a survival from a vanishing world, and treating every economic grievance in India as if its caste grammar were a detail. Both errors are refuted in the record this article surveys, and both will be made again, in print and on air, by writers who did not consult it.

Frequently Asked Questions

Q: Is caste a form of class?

Not exactly, and the exactness matters. Where class names a position in the relations of production, caste names a hereditary rank that assigns people to production, and the two maps coincide often enough, village by village, to have persuaded a generation of theorists that one was the other’s Indian costume. They part company where the evidence is most characteristic: at marriage, at worship and in the honour that survives both riches and ruin. The defensible summary is that caste in India is the machinery through which class positions were historically distributed and policed, which makes the systems inseparable in practice and distinct in analysis.

Q: What did the early communists say about caste?

They said, in substance, that caste was a leftover. In the doctrine the movement inherited and preached through its first decades, caste belonged to the feudal past: a division of labour sanctified by religion, sustained by landlords and priests, and already being dissolved by the factory, the railway and the wage. From that diagnosis a strategy followed without further argument. Fight landlordism and capital, and the soil caste grew in would be removed; raise caste as a separate question, and the workers would be divided by the very identities the enemy wished them to keep. The position was stated in party literature and education rather than in any single canonical text, and its grip lasted well beyond the point where the movement’s own organisers had reason to doubt it.

Q: Why did the classical position on caste fail?

It failed on prediction. A doctrine that classifies caste as an ideological survival of a dying economy commits itself to watching caste weaken as the economy modernises, and the record ran the other way in the places that mattered most: industrial workforces reproduced caste segmentation, communities carried caste across conversion into faiths that gave it no warrant, and democratic politics rewarded caste organisation with a second career. A theory may survive any one of these as a lag. It cannot survive all three while continuing to describe persistence as residue, and the classical position’s own tradition eventually withdrew it for exactly that reason.

Q: Is caste superstructural?

In the vocabulary of the old model, no, and the question shows why the model strained. Caste does not behave as the superstructure was supposed to behave: it did not arise afresh from each economy it inhabited, it did not dissolve when its supposed base was transformed, and it reaches down into the organisation of production itself, assigning labour, land and credit before any market convenes. The more careful conclusion the debate reached is that the base and superstructure picture is the wrong instrument for this object. Caste is one of the relations through which Indian production was organised, and simultaneously a sacred order with mechanisms of its own, which is why the later positions abandoned the classification rather than revising it.

Q: Does caste prevent working class unity?

It obstructs unity systematically without making it impossible, and the distinction is the practical heart of the debate. The graded order gives almost every labourer a superiority to defend and a contempt to practise, segments labour markets by community, and routes jobs, credit and help through caste networks, so solidarity must be constructed against strong contrary incentives rather than merely proclaimed. Where organisers have done that construction, in land struggles, union drives and combined movements, unity across caste has repeatedly been achieved and held under pressure. What the record refutes is the assumption that unity arrives by itself once exploitation is pointed out. The address of 1936 made exactly this the test of any Indian socialism, and the movements that passed the test are the ones that took the warning as instruction.

Q: Do caste and class overlap in practice?

Yes, heavily, and the overlap is the best-documented finding in the whole field. Land, occupation, schooling, income and political office all track ritual rank closely, across regions and across every method of inquiry the literature has used, from the village monograph to the matched-application experiment in city hiring. Two cautions complete the answer. The overlap is a correspondence, not an identity, since honour, marriage and ritual standing run on a second ledger that income does not settle. And the overlap is not a natural fact but a maintained one, renewed each generation by the mechanisms of assignment and enforcement that the debate exists to describe.

Q: What is the Dalit and Marxist synthesis?

It is the attempt, pursued since the 1970s in movements and in scholarship, to refuse the forced choice between the class vocabulary and the anti-caste vocabulary and to hold both as descriptions of one oppression. Its organisational emblem is the Dalit Panthers of Bombay, founded in 1972 by activists from the oppressed communities who named landlords and capital as enemies alongside the sacred order that sanctified subordination. Its scholarly form appears in work, associated with figures such as Gail Omvedt, that theorises caste, class and gender as interlocking systems neither of which can be derived from another. The synthesis is less a finished doctrine than a discipline: no analysis is accepted that cannot state, in the same breath, who owns the land and who is barred from the well.

Q: Can economic change dissolve caste?

The record answers that economic change transforms caste without dissolving it. Markets, migration and wage labour have broken hereditary occupations for millions, funded the organisation of the oppressed and given individuals exits from village jurisdiction that earlier centuries never offered. The same centuries show the order reconstituting itself inside each new arrangement: segmented urban labour markets, endogamous marriage among the affluent and educated, caste networks operating as the credit system of whole industries. Dissolution, on the evidence, is not something economic change does to caste as a side effect. Where caste has retreated, it has retreated before organised attack on several fronts at once, economic, legal, educational and religious, which is a conclusion about strategy rather than about the economy’s powers.

Q: How does this debate compare with intersectionality?

The two debates share a structure and differ in almost everything else. Intersectionality, named by Kimberlé Crenshaw for the crossing of race and gender in American law and life, criticises single-axis analysis and argues that systems of oppression constitute one another rather than adding up. The Indian debate arrived at the same architecture by its own road: class first and caste as residue gave way, under pressure from those the single axis misdescribed, to the claim that caste and class make each other. The materials are not interchangeable, since caste’s hereditary closure and religious warrant have no exact American counterpart, and the two literatures developed in mutual ignorance. Their agreement is therefore the more instructive: the same mistake, corrected in the same way, on opposite sides of the world.

Q: What is the strongest current position on caste and class?

The position with the most evidence behind it is the one this article calls the mutual constitution thesis: caste and class in India produce and maintain one another, and neither is the other’s effect or disguise. It absorbs the form-of-class insight, since the correspondence of rank with property is real and is explained by caste’s assignment of persons to economic places. It absorbs the autonomy insight, since marriage, ritual and honour obey caste’s own logic and survive every economic transformation so far recorded. And it explains the fact that defeated both simpler views, caste’s combination of extreme durability with perpetual change, by locating the durability in the mechanisms of reproduction and the change in the economy that keeps re-equipping them.

Q: How does the mutual constitution thesis grade the older positions?

It grades them as partial truths promoted beyond their evidence. The superstructure doctrine is marked down most severely, since its central prediction failed and its own tradition withdrew it, though it is credited with noticing that caste had an economic history. The form-of-class position receives the highest partial credit, because the correspondence it rests on is the strongest single finding in the field, and loses marks only where correspondence fails, in marriage, ritual and honour. The autonomy position is credited with establishing caste’s independent mechanisms and marked down for neglecting the material instruments by which the hierarchy is enforced and paid for. The grading is the thesis’s method in miniature: each older view survives as the description of one mechanism, and fails as the description of the whole.

Q: What should a student separate between caste as superstructure and caste as the form of class?

Separate the location each position assigns to caste, because the locations generate opposite predictions and opposite strategies. The superstructure position places caste above the economy, in the realm of ideas and customs, and therefore predicts its dissolution by economic change and prescribes patience: transform production and the ideology follows. The form-of-class position places caste inside the economy, as the very shape production relations take in India, and therefore predicts that caste stands or falls with the material arrangements that carry it, and prescribes land and labour struggles as anti-caste struggles in themselves. An answer that can state both locations, the evidence each explains and the evidence each cannot, has mastered the distinction examiners are testing when they set this topic.

Q: Why did the classical position fail on its own evidence?

Because the movement’s own experience contained the refutation before the theorists admitted it. Organisers in the mills met caste in the recruitment line and the department roster; organisers among agricultural labourers met it in the customary wage and the barred temple path; the movement’s ranks, drawn from the oppressed communities in its strongholds, lived the distinction between what money buys and what birth forbids. The doctrine nevertheless continued to classify caste as a fading ideology, and organisers worked around their own theory for decades, fighting caste in practice while denying it in principle. A position that its holders must circumvent in order to work has failed on its own evidence in the strictest sense available: the evidence was in its possession, and it declined to read it.

Q: How does the four positions grid order the debate?

The grid orders the debate by the relation each position asserts between caste and class, which is the debate’s true axis. It begins with subordination in one direction, caste as an effect of economic relations in the superstructure doctrine, passes to identity, caste as the form class takes in the form-of-class reply, moves to independence, caste as an autonomous order governing its own domains, and ends with reciprocity, the two orders constituting each other. Arranged this way, the sequence is also chronological, since each position arose as the correction of the one’s failures before it, and a student who can walk the grid in either direction, naming for each row its claim, its carriers, its prediction and its fatal objection, holds the whole debate in a form that fits an examination answer or a seminar hour.

Q: What separates a mutually constituting relation from an additive one?

An additive account keeps the systems separate and sums their effects: a person is disadvantaged by caste and, additionally, by class, as weights are added to a scale, and each weight can in principle be lifted while the other remains. A constitutive account denies that the weights exist separately. Caste decides who arrives at the labour market with land, schooling and credit, and class position decides what resources a community can spend on pressing or defending its rank, so each system is among the causes of the other’s concrete shape. The practical test is subtraction. Where removing one system in thought leaves the other unchanged, the relation was additive; where the subtraction dissolves the phenomenon being explained, as the removal of caste would dissolve the actual structure of Indian land, labour and honour, the relation was constitutive all along.