The Mandal Commission Report, submitted to President Neelam Sanjiva Reddy on 31 December 1980, is the most argued-about government document in the history of the Indian republic, and it is argued about almost entirely through a single number: the 27 percent reservation it recommended for the Other Backward Classes in central government employment and public sector undertakings. The number is real, and its consequences after August 1990 were enormous. But the number was the output of a machine, and the machine is the Report. Between January 1979 and December 1980 the Second Backward Classes Commission, chaired by Bindeshwari Prasad Mandal, constructed a method for identifying backwardness in a country that had stopped counting caste in its census half a century earlier: eleven indicators grouped under three heads, a three-tier scheme of weights producing a twenty-two point scale, a classification threshold set at half the available points, an enumeration of 3,743 castes and communities, a population estimate of 52 percent built by subtraction from the census of 1931, and a reservation quantum derived as the residue left beneath a judicial ceiling. The claim defended in this article is that the methodological substance of the December 1980 Report is the foundational analytical content of the Mandal moment, and that the political trajectory after 1990, the street protests, the nine-judge verdict in Indra Sawhney, the educational reservation of 2006, operated through that method’s architecture even where governments and courts amended its parts.

The standard way of telling the Mandal story begins in 1990, with the announcement of 7 August in which Prime Minister V.P. Singh told Parliament that his National Front government would implement the Report, and it ends in the courtroom on 16 November 1992, when the Supreme Court upheld the 27 percent while fencing it about with conditions. That telling is not false, but it treats the Report as a detonator rather than as a text: a decade-old document whose political charge happened to be touched off by a weak coalition government under pressure. Read in that frame, the interesting questions are all about the detonator, why V.P. Singh chose that week, what the announcement did to his coalition, how the protests were organised, and the method by which 3,743 castes had been pronounced backward a decade earlier becomes a footnote. The scholarly treatments that resisted this compression, Marc Galanter’s Competing Equalities of 1984 and Christophe Jaffrelot’s India’s Silent Revolution of 2003 above all, insisted that the document be read as an argument: a claim about what backwardness is, about how it can be measured at all in a society ordered by caste, and about what the state owes to the measured result. This article follows their lead and goes further into the machinery than either of them needed to go, because the machinery is where the Report’s durability lies. Governments shelved the Report for ten years and then implemented it in fragments; courts struck one of its limbs and grafted another it had never contained; a constitutional amendment of 2019 built a rival reservation on a wholly different criterion. Through every one of those operations, the terms of trade were the terms the Commission’s method had set: caste as the unit of identification, indicators rather than impressions, a score, a threshold, a list, a ceiling. To understand the Mandal moment, the method has to be reconstructed piece by piece, from the warrant that created the Commission to the arithmetic that produced the famous figure.

A Document Built Under Warrant: The Commission of 1 January 1979

The Mandal Commission was the second attempt by the Indian state to answer a question the Constitution had posed and the first attempt had failed to settle. Article 340 empowers the President to appoint a commission to investigate the conditions of socially and educationally backward classes and to recommend the steps the Union and the states should take to remove their difficulties. The announcement came from a government with a debt to pay. The Janata Party, which had defeated the Congress in the election of March 1977 that closed the Emergency, rested heavily on the backward-caste electorate of the northern states, and its victory belongs to the post-Janata 1978 institutional-political context in which promises made to that electorate during the campaign had to find institutional form. On 20 December 1978, Prime Minister Morarji Desai informed Parliament that the government would appoint the Second Backward Classes Commission. The presidential order constituting the Commission under Article 340 followed on 1 January 1979. The official name, the Socially and Educationally Backward Classes Commission, recorded the constitutional phrase exactly: the classes to be found were those the Constitution’s own language described as socially and educationally backward.

When was the Mandal Commission appointed?

Prime Minister Morarji Desai announced the decision in Parliament on 20 December 1978, and the Commission was formally constituted by presidential order under Article 340 on 1 January 1979, with Bindeshwari Prasad Mandal, a former chief minister of Bihar and a sitting member of the Lok Sabha, as its chairman and five members drawn largely from the backward classes themselves.

The chairman’s biography was itself an argument about the Commission’s character. Bindeshwari Prasad Mandal came from Murho village in the Madhepura district of Bihar, from the Yadav caste that formed the largest single bloc within the backward classes of the northern plain. He had entered politics in the Congress, moved through the socialist currents of Bihar politics, and in February 1968 had held the chief ministership of Bihar at the head of a coalition assembled around his Shoshit Samaj Dal, an office he kept for a matter of weeks before the coalition dissolved. He had returned to Parliament in the Janata wave of 1977. The members appointed with him were Dewan Mohan Lal, L.R. Naik, K. Subramaniam and Dina Bandhu Sahu; S.S. Gill, an officer of the Indian Administrative Service, served as the Commission’s secretary and ran its secretariat. The composition attracted comment from the start and has attracted it ever since. Shriram Maheshwari, in The Mandal Commission and Mandalisation: A Critique, published in 1991, recorded that the First Backward Classes Commission had been broadly based while the Second was shaped on partisan lines, composed of members drawn from the backward castes, four of its five members from the Other Backward Classes and the fifth, L.R. Naik, the Commission’s only member from the Scheduled Castes. Defenders of the composition answered that a commission asked to investigate the lived condition of backward India could not be staffed as though that condition were a rumour to be verified by its beneficiaries’ social superiors, and that the First Commission’s breadth had produced a report its own chairman half disowned. The dispute matters for what follows, because the strongest version of the case against the Report’s method, set out in a later section, begins precisely here: with the claim that the method’s choices were the choices of men who knew in advance what they intended to find.

The terms of reference given to the Commission were fourfold, and their wording structured everything the Report became. The Commission was to determine the criteria for defining the socially and educationally backward classes. It was to recommend the steps to be taken for the advancement of the classes so identified. It was to examine the desirability of reserving appointments and posts, and seats in educational institutions, in favour of those classes. And it was to present its report, with its recommendations, to the President. Two features of this mandate deserve emphasis because they explain the Report’s architecture. The first is that the identification exercise and the reservation question were joined in a single brief: the Commission was not asked merely to count the backward and then, separately, to consider policy, but to build a definition of backwardness that could carry the weight of a reservation decision. The second is the order of the terms. Criteria came first. A commission that had begun with a list of castes and reasoned backwards to a justification would have produced a different document; the Mandal Report, whatever else is said of it, begins with criteria and reasons forwards to lists, and that order is the core of its defence against the charge of arbitrariness. By the time the Commission began its substantive work, the politics around it had already collapsed once: the Janata government fell in July 1979, Charan Singh’s caretaker ministry followed, and in January 1980 Indira Gandhi returned to office at the head of a Congress(I) majority, so that the Report was researched under a government that had not commissioned it and delivered into the hands described in the post-1980 Indira Congress(I) governance context, a fact that shaped its first decade of life.

The Antecedent That Failed: The Kalelkar Commission of 1953 to 1955

No reading of the Mandal method is adequate without the failure it was designed against, the story told in full in the 1953-1955 Kalelkar Commission antecedent. The First Backward Classes Commission, appointed in January 1953 under the chairmanship of the Gandhian writer and parliamentarian Kaka Kalelkar, reported on 30 March 1955. It identified 2,399 castes and communities as backward, 837 of them as most backward, and it recommended reservations and a programme of ameliorative measures. The report was laid before Parliament in 1956 accompanied by a government memorandum that found its criteria for classifying backwardness unsatisfactory, and there the matter rested: the Union government declined to draw up an all-India list, told the states to proceed by their own lights, and for a quarter of a century the question of the backward classes at the centre was answered state by state, list by list, with no common method and no common measure. The reasons for the failure were partly political, a Congress leadership unwilling to open the caste question at the centre, but they were also internal to the document. Kalelkar’s own letter forwarding the report had expressed deep unease about caste as the criterion of backwardness and had leaned towards economic tests, so that the first Commission’s report arrived, in effect, with its chairman’s misgivings printed in front of it. A report whose author doubts its unit of analysis invites burial, and burial is what it received.

How many castes did the Kalelkar Commission list?

The First Backward Classes Commission under Kaka Kalelkar, reporting on 30 March 1955, listed 2,399 castes and communities as backward across India, of which it classified 837 as most backward, a larger enumeration than any national exercise had attempted before, though its recommendations were never taken up by the Union government of the day.

The Mandal Commission worked in full knowledge of this precedent, and its method can be read as a set of answers to the Kalelkar failure, point by point. Where Kalelkar’s commission had classified castes largely by their ritual rank and reputation, Mandal’s built explicit indicators and announced them in advance of their application, so that classification could be checked against stated tests rather than attributed to collective impression. Where Kalelkar had treated caste as almost the whole of the matter and then apologised for doing so, Mandal’s commission argued the caste question through, in a sustained theoretical chapter, and declined to apologise: its position, examined below, was that caste in India is not merely a ritual rank but a class in sociological fact, the unit through which advantage and deprivation are actually distributed, and that a criterion which refused the unit would measure everything except the phenomenon. Where the first report had been a single undifferentiated list, the second graded its findings, distinguishing within the backward universe a sub-stratum of the depressed backward classes. And where the Kalelkar report had arrived without an enforcement of its own logic on the quantum of reservation, the Mandal Report tied its recommendation to the operative judicial ceiling from the outset, so that its central demand was, on its face, implementable within the law as the courts had stated it. Marc Galanter’s Competing Equalities, published in 1984, remains the indispensable scholarly account of this succession: Galanter showed how the Kalelkar exercise had been undone by the politics of list-making and by the absence of a defensible public method, and his analysis supplied much of the vocabulary, compensatory discrimination above all, in which the Mandal debate would later be conducted. The Mandal Commission’s officers knew that literature’s warning. A list without a method is a patronage instrument; a method stated in public can be attacked, but it can also be defended, revised and, in the end, upheld by a court. The Report staked everything on the second possibility.

The Constitutional Frame the Report Had to Fit

The Commission did not work in a legal vacuum, and its method bears the marks of the courtroom throughout. The frame had been assembled in the three decades before 1979 and is set out as a whole in the three-layered constitutional reservation framework: Article 15(4), inserted by the First Amendment of 1951 after the Supreme Court struck down Madras’s communal quotas in State of Madras v Champakam Dorairajan, empowers the state to make special provision for the advancement of socially and educationally backward classes; Article 16(4) permits reservation of appointments for any backward class of citizens not adequately represented in the services; Article 340 supplies the commission device itself; and behind them stands the older settlement between the fundamental rights and the directive principles and fundamental rights settlement by which the Constitution’s equality code was made to carry a redistributive charge. Within that frame the courts had already fixed the load-bearing walls before Mandal sat. In M.R. Balaji v State of Mysore, decided in 1963, the Supreme Court struck down a Mysore order reserving 68 percent of seats and laid down the propositions that governed everything after: caste is a relevant circumstance in identifying a backward class but cannot be the sole or dominant criterion; backwardness must be both social and educational; and reservation, speaking broadly, should remain below 50 percent. In R. Chitralekha v State of Mysore, decided in 1964, the Court accepted that a state might classify by occupation and income without reference to caste at all, confirming that the Constitution permits more than one route to the same destination. In P. Rajendran v State of Tamil Nadu, decided in 1968, a Madras list drawn substantially by caste survived challenge. The Commission’s lawyers and members read this case law as a set of design constraints, and the method is unintelligible except as an attempt to satisfy all of them at once: caste retained as the organising unit, as Rajendran allowed; caste fenced about with measured social, educational and economic conditions, as Balaji demanded; and the total quantum of reservation held below the Balaji line, as the Report’s own derivation, examined below, was careful to do. When the Report’s defenders later said that the Commission had anticipated Indra Sawhney, this is what they meant: not prophecy, but drafting under constraint, a method built to be litigated.

One more feature of the frame deserves record, because it explains the Commission’s treatment of the states. Reservation for the backward classes was, by 1979, already an old practice in the south and west: Mysore had run backward-class quotas since the Miller Committee of 1918, Madras since the communal orders of the 1920s, and after 1956 the state governments of the successor states maintained their own lists and their own percentages, built by state commissions and state inquiries of uneven rigour. The Union, by contrast, reserved nothing for the backward classes at the centre beyond the Scheduled Caste and Scheduled Tribe quotas. The Commission’s brief was therefore not to invent a policy but to nationalise one: to produce, for the whole country, the criteria and lists that the states had produced piecemeal, and to do so in a form the Union could adopt for its own services and institutions. That is why the Report’s lists are state lists, compiled state by state against state averages, and why its method measures every caste against the mean of its own state rather than against an all-India mean: backwardness, in the Commission’s working theory, is a relation between a caste and the society that surrounds it, and the surrounding society whose judgements of rank and whose distribution of schooling and assets do the work is the society of the state.

The Evidence Base: Census, Survey and Hearings

The Commission’s first problem was that the data its task required did not exist. The census of India had enumerated caste in full for the last time in 1931; the census of 1941 was truncated by the war, and the census of 1951, taken under the shadow of Partition and in the name of a casteless citizenship, recorded caste only for the Scheduled Castes and Scheduled Tribes. A commission directed in 1979 to identify the backward classes of the entire country therefore faced a statistical landscape in which the central fact it needed, the population and condition of each caste, had been officially unmeasured for forty-eight years. The Report’s answer was to build the evidence from four distinct sources and to say, in its methodological chapters, exactly what each source could and could not bear. The caste-wise population figures came from the 1931 census, extrapolated forward. The general statistical series, literacy, schooling, assets, housing, came from the published censuses and from the state governments’ records. The state lists of backward classes, maintained by state governments for their own reservations, supplied a starting enumeration for each state. And the Commission generated a primary source of its own: a socio-educational field survey, launched in February 1980, in which two villages and one urban block were selected in every district and every household in them was canvassed; the survey covered 405 of the country’s 406 districts, its field operations were entrusted to the state statistical organisations, and its returns were tabulated by computer with the assistance of the National Informatics Centre. A questionnaire went out to members of Parliament, to state legislators, to voluntary organisations and to the public; the Commission toured the states, held public sittings, and took evidence from caste associations, from officials and from scholars, testimony preserved in the hearing transcripts whose runs are held in the archives of the Ministry of Social Justice and Empowerment and which remain the least consulted stratum of the Commission’s record, a point to which the discussion of the method’s critics will return.

The machinery that ran this evidence base deserves a paragraph of its own, because the Report’s critics and defenders argue about the machinery as much as about the findings. The Commission constituted a technical apparatus unusual for an Indian commission of its period: a panel of experts, including statisticians of the Central Statistical Organisation and the National Sample Survey Organisation and directors of the state bureaux of economics and statistics, advised on method; a research planning team of sociologists framed the inquiry’s parameters; and a sub-committee of experts helped convert the returns into the indicators by which castes would be judged. The field survey itself was not run by the Commission’s own staff but entrusted to the statistical organisations of the states and Union territories, whose enumerators were trained for the purpose and whose schedules were then fed into a computer, the tabulation executed with the National Informatics Centre’s assistance, a strikingly early use of electronic processing for an exercise of this social depth. The design intent is legible in the arrangements. Enumeration was decentralised to the machinery that knew the districts; method was centralised in expert hands; and the conversion of raw returns into classification criteria was assigned to a named technical process, set out at paragraph 11.23 of the Report, rather than left to the impressions of the touring members. That the execution departed in places from the design, as the critics documented, is a charge examined in the adjudication below; but the existence of the design, a survey with a published sampling frame covering two villages and one urban block in 405 of 406 districts and a household canvass within them, separates the Mandal exercise from the reputation-based classifications of its predecessor as sharply as any single feature of the Report. Kalelkar’s commission had travelled and listened; Mandal’s travelled, listened, sampled, counted and computed.

The design of this evidence base carried a theoretical commitment that should be made explicit, because it distinguishes the Report from a mere administrative compilation. The survey did not measure individuals and aggregate them into classes by income or occupation, as an economic test would have required. It measured castes: the schedules were applied to the castes found in the sampled villages and blocks, and the quantities recorded, marriage ages, school attendance, asset values, house types, were computed as caste averages against state averages. The choice followed from the Report’s central sociological claim, stated in its early chapters with a bluntness that still startles: that in India caste is not a survival about to dissolve into class but the operative structure through which class advantage is held and transmitted, so that a socially and educationally backward class, in the Constitution’s phrase, presents itself to any honest inquiry as a caste or a cluster of castes. The Report put the negative form of the claim in a sentence that its critics and defenders both quote: the substitution of caste by economic tests, it said, will amount to ignoring the genesis of social backwardness in Indian society. The merit doctrine received the same treatment. What is called merit in an elitist society, the Report argued, is an amalgam of native endowments and environmental privileges, and children born with equal intelligence into an advanced-class home and a backward-class home do not run the same race. These sentences are the philosophical core of the document, and everything in the machinery that follows, the indicators, the weights, the threshold, is an attempt to give them administrative form: to convert a theory of how disadvantage works into a procedure a state can run.

A practical note belongs here, because the Report’s afterlife has been a life of paper. The findings were published as a main report with state-wise lists and statistical appendices, and the appendices, district tables, caste lists by state, the survey tabulations, are the working parts of the document for any later commission or court that has to test what Mandal did. Those appendices circulate in government libraries and in scanned sets, many of them as image files and spreadsheets rather than searchable text; a reader working through them on a machine without the originating software can open the tabulated files through an office file viewer and read the district tables directly. The point is not antiquarian. Every later revision of the backward-class lists, from the expert committee the Supreme Court ordered in 1992 to the standing National Commission for Backward Classes, has had to reason from Mandal’s tables as the baseline, because for hundreds of castes the Mandal survey remains the last occasion on which the Indian state counted them at all.

The Mandal Commission Report’s Eleven Indicators

The heart of the method is the set of eleven indicators the Commission evolved for determining social and educational backwardness, set out in the Report at paragraph 11.23 and applied, state by state, to every caste within the survey’s reach. They fall under three heads. The first head gathers the tests of social standing, four in number. A caste counted as backward by others, that is, placed low in the estimation of the surrounding castes rather than merely poor in its own account, met the first test. A caste mainly dependent on manual labour for its livelihood met the second. A caste in which marriage below the age of seventeen ran ahead of the state average, for females by at least 25 percent in rural areas with a parallel male test at 10 percent, and with lower urban thresholds of 10 percent for females and 5 percent for males, met the third. A caste in which women’s participation in work stood at least 25 percent above the state average met the fourth, the Report reasoning that heavy female labour in a caste is a mark of economic compulsion and low standing, not of emancipation. The second head gathers the tests of educational condition, three in number: a caste whose children between five and fifteen who had never attended school exceeded the state average by at least 25 percent; a caste whose dropout rate in the same age band exceeded the state average by at least 25 percent; and a caste whose proportion of matriculates fell at least 25 percent below the state average. The third head gathers the tests of economic condition, four in number: a caste whose average value of family assets fell at least 25 percent below the state average; a caste whose families living in kuchcha houses exceeded the state average by at least 25 percent; a caste in which the source of drinking water lay beyond half a kilometre for more than half the households; and a caste whose households that had taken consumption loans exceeded the state average by at least 25 percent.

Three properties of this list repay attention, because they are where the method’s intelligence, and most of the honest dispute about it, actually live. The first is that every test is relational. No indicator fixes an absolute line, a rupee figure, a literacy percentage, below which a caste is backward in the abstract. Each test compares a caste with the average of its own state, and the recurring margin, 25 percent above or below the state mean, converts backwardness into a position within a distribution. The design has a consequence its authors understood: a caste middling in a poor state and a caste middling in a rich state are measured against different means, and the method accepts that result deliberately, on the theory that the disadvantages the Constitution cares about are disadvantages within the society a person actually competes in. The second property is that the indicators are drawn from the census-taker’s world, not the philosopher’s. Each test is a quantity a field survey can in principle count: marriages, schoolrooms, matriculation registers, house walls, wells, loan books. The Report sacrificed depth for countability knowingly; it says, in effect, that a criterion which cannot be administered by a district staff is not a criterion but a sentiment. The third property is the one the critics seized on and the defenders answer for: the tests are uneven in kind. Being considered backward by others is a judgement of reputation; dependence on manual labour is a fact of occupation; early marriage is a demographic rate; the drinking-water test is a measure of public infrastructure as much as of caste condition. A method that adds such unlike quantities into a single score is exposed to the charge of adding apples to oranges, and the charge was duly made. The Commission’s answer is implicit in the structure it chose: backwardness in the Indian case is itself a compound phenomenon, ritual standing, schooling and material condition braided together, and a test battery that refused to compound them would falsify the thing measured. Whether the particular compound was the right one is the question the weights decided, and the weights are where the argument truly begins.

The recurring margin in the tests, 25 percent above or below the state average, is itself a design decision of the same character and should not pass unexamined. A uniform margin across unlike quantities, marriages, matriculates, mud walls, loan books, cannot be read off the phenomena; it is a convention, chosen so that a district staff applying the battery to a caste would face the same form of question eleven times, and so that the classification of a caste in one state would be produced by the same instrument as the classification of a caste in another. The Commission’s method throughout prefers a stated convention, uniformly applied and open to attack as a convention, to a series of expert judgements varying invisibly from caste to caste. That preference is the deepest habit of the document, and it recurs at every level: in the threshold set at half the scale, in the ceiling arithmetic of the quantum, in the roster and carry-forward machinery borrowed entire from the Scheduled Caste template. The Report behaves, in this respect, less like a work of social philosophy than like a piece of constitutional engineering, and its authors appear to have believed that in India a remedial idea survives in proportion as it can be reduced to a procedure that a clerk can run and a court can check.

The Three-Tier Weight and the Twenty-Two Point Scale

Indicators alone do not classify anything; a rule for combining them does. The Commission’s combination rule is the most contested artefact in the entire Report, and it is simple enough to state in a breath. The three heads of indicators were judged to be of unequal importance for the purpose in hand, and separate weightage was given to each head. Every social test carried three points. Every educational test carried two points. Every economic test carried one point. A caste therefore faced a maximum score of twenty-two: twelve points available under the social head, six under the educational, four under the economic. A caste that accumulated half the available points or more, eleven points on the twenty-two point scale, was listed as socially and educationally backward; a caste below the line was treated as advanced. The Report explained the ordering in a single doctrinal move: economic backwardness, it held, flows directly from social and educational backwardness, so that the economic tests deserved the least weight because they measure the consequence of the condition rather than the condition itself, while social standing, the caste’s place in the estimation of others and in the division of labour, is the condition’s root and deserved the most. The methodological map of the Mandal Commission Report (Table 1) sets out the whole architecture in one view.

Component of the method	Content	Weight or figure
Social indicators	Four tests: standing in the estimation of other castes, dependence on manual labour, early marriage, women’s participation in work	3 points each; 12 points in all
Educational indicators	Three tests: children aged five to fifteen who never attended school, dropout rate in the same band, share of matriculates	2 points each; 6 points in all
Economic indicators	Four tests: average value of family assets, kuchcha housing, distance of drinking water, households with consumption loans	1 point each; 4 points in all
Composite scale and threshold	The eleven tests summed for each caste, measured against the averages of its own state	22 points in all; backward at 11 points or more
Enumeration	Castes and communities classified as Other Backward Classes	3,743 in all, of whom 2,108 listed as depressed backward classes
Population estimate	Share of the national population found in the listed castes, computed from the census of 1931	52 percent
Reservation quantum	The residue beneath the 50 percent ceiling after 22.5 percent for the Scheduled Castes and Scheduled Tribes	27 percent; 49.5 percent in all

The weighting decision is where the Report’s theory of Indian society is encoded in arithmetic, and it should be read as theory, not as bookkeeping. To weight social standing at three times the value of material condition is to assert that the Indian distribution of disadvantage is produced by the caste order first and by poverty second, so that a caste can be comparatively prosperous and still backward, and a caste can be poor without being backward in the constitutional sense at all. That assertion did real classificatory work. Prosperous peasant castes with low ritual standing, thin schooling and a history of exclusion from office scored heavily on the social and educational heads and entered the lists despite passing some economic tests; poor families of high-standing castes scored on the economic head alone, could accumulate at most four points from it, and stayed outside. The creamy-layer controversy of the following decade, and the entire later jurisprudence of the advanced sections within the backward classes, is latent in this design decision of 1980: a method that weights standing over wealth will necessarily carry some wealthy households into a remedial category, and will be attacked for doing so by everyone who thinks the category should track wealth. The Commission’s answer, that the attack mistakes the purpose of the exercise, is the answer its whole theory required. Reservation under Articles 15(4) and 16(4) was not, in the Report’s reading, a poverty programme; it was a remedy for a structure of graded inequality, and the structure is made of standing before it is made of money.

The threshold deserves the same scrutiny as the weights. Eleven points, half the scale, is a convention, and the Report does not pretend otherwise; no property of Indian society dictates that backwardness begins at exactly half of twenty-two. What the convention purchased was administrability and a rough symmetry with the population result: applied to the survey returns and the state lists, the eleven-point line sorted the castes of each state into a backward set that covered, in the aggregate, about half the country, which is what the Commission’s theory of the caste order predicted the backward universe should look like. Critics have treated the fit between threshold and predicted population as evidence that the line was drawn to reach a predetermined total, and defenders have answered that a threshold which had produced a backward population of 15 percent or of 85 percent would have falsified the theory on its own terms. Both readings are available on the record; what cannot be said is that the threshold was hidden. It is printed in the Report with the scores’ logic open beside it, which is more than the Kalelkar exercise, classifying by reputation without a published metric, ever offered its readers. Within the backward set, the Report drew one further line of great consequence for later politics: its state tables separately identified 2,108 of the 3,743 castes as depressed backward classes, the most underprivileged stratum within the backward universe, a grading the majority recorded in the lists while declining, against a dissent examined below, to build separate safeguards upon it.

One absence in the published record must be registered at this point, because the indictment of the method returns to it repeatedly. The Report published its indicators, its weights and its threshold, and it published the lists those instruments produced; it did not publish, caste by caste, the score each caste had obtained under the eleven tests. The omission had administrative explanations: the scoring had been executed state by state from survey and census materials of uneven completeness, and for many castes the classification rested on a braid of survey returns, state list entries and recorded local enquiry rather than on a single clean computation. But the omission’s consequence was structural. A critic could verify the rule and could inspect the output, yet could not re-perform the operation for any individual caste and check the sum, and the distance between a published rule and an uncheckable application is precisely the distance across which suspicion travels. The point was made in the litigation decade and has been made since: a method that claims objectivity must furnish the minimum evidence by which its objectivity can be tested, and score sheets are that minimum. The defence available to the Commission is partial and should be stated exactly. The classification’s materials, the census tables, the survey returns, the state lists, were public or archival; the lists were issued state by state and were open to challenge entry by entry, and were in fact challenged, corrected and maintained by the standing machinery that followed; and the Supreme Court in 1992, with the omission before it, chose the remedy of periodic expert review rather than the remedy of annulment, implying a judicial judgement that the exercise was revisable rather than void. The seam is real, and the revised architecture of review and maintenance built after 1992 is best understood as the system repairing, institutionally, the one verification gap the Report had left open.

Counting the Backward: 3,743 Castes and the 52 Percent

Applied across the states, the method produced the enumeration on which everything else rests: 3,743 castes and communities classified as Other Backward Classes. The lists are state lists, and their unevenness is instructive. Karnataka’s list ran to 333 castes, Tamil Nadu’s to 288, Maharashtra’s to 272, Madhya Pradesh’s to 279, while Nagaland returned none and Lakshadweep none, a distribution that tracks the real geography of the caste order rather than any quota of entries per state. The population arithmetic built on the lists proceeded, as the Report states, from the census of 1931, the last enumeration to record caste, and it proceeded by deduction. Within the Hindu population, the shares of the Scheduled Castes and Scheduled Tribes and of the forward castes were known; deducting them left the backward share of the Hindu population at approximately 44 percent. Non-Hindus formed about 16 percent of the population, and the Commission, reasoning that the internal stratification of the other religious communities mirrored the Hindu pattern in this respect, carried roughly half of that share, some 8 percent of the national population, into the backward column, counting convert communities by their pre-conversion standing and occupational communities, the dhobis, lohars, nais and telis of every faith, by the standing of their craft. The two parts summed to 52 percent: approximately 52 percent of the people of India belonged, on the Commission’s count, to the Other Backward Classes, against the 22.5 percent covered by the Scheduled Caste and Scheduled Tribe quotas and a forward-caste population of about 17.5 percent.

The state-wise shape of the lists carries an argument of its own, and it is worth pausing on the figures the Report’s tables supply. Odisha returned 224 castes, Kerala 208, West Bengal 177, Bihar 168, Uttar Pradesh 116, Gujarat 105 and Punjab 83, a spread that no central template imposed and that reflects the Commission’s decision to compile each state’s list from that state’s own materials against that state’s own averages. Two consequences followed. The first is that the category Other Backward Classes named a genuinely federal object: a caste could be listed in one state and absent in the next, since its standing, schooling and assets were measured against different surrounding societies, and the Report accepted the resulting patchwork as the honest output of its relational tests rather than ironing it into a false uniformity. The second is that the central list the Union would eventually maintain inherited this patchwork entire, so that the politics of inclusion and exclusion after 1993 was conducted caste by caste and state by state, exactly as the method’s design implied it would be. The depressed backward classes were distributed with the same unevenness: of the 2,108 castes in that sub-stratum, the state tables show the largest concentrations in the same populous states that dominate the main lists, a geography that mattered when the sub-categorisation argument, Naik’s argument, returned to national politics in later decades.

The coverage of the enumeration had one boundary the Report acknowledged and its critics underlined. The indicator scores were computed, in the first instance, for the castes encountered in the survey’s sampled villages and urban blocks and in the census and state-list materials; a caste thinly spread across a state, or concentrated in districts the sample touched lightly, reached the classification table through a braid of sources, survey returns where they existed, the state list where it spoke, and recorded local enquiry where neither sufficed. The Commission did not present this as a defect to be apologised for but as the condition of any enumeration of a universe numbering thousands of communities across 405 districts: the alternative to a braided classification was not a perfect one but none at all, the Kalelkar outcome by another route. The answer has the same shape as the Report’s answer on the 1931 census, and it identifies the document’s consistent epistemology: proceed with the best instrument that exists, publish the instrument, and let the classification be corrected in use rather than postponed until measurement is beyond criticism.

The Commission bound its enumeration to a finding about representation that supplied the moral pressure behind the recommendation. Tabulating the composition of the central services, the Report recorded that the classes it had identified as backward, more than half the population, held 4.83 percent of Class I posts and 10.63 percent of Class II posts in the central government. The figures have been quarrelled with like every other number in the document, but their function in the argument is worth isolating: they convert the abstract claim of exclusion into a ratio. A community set comprising half the nation and a twentieth of its senior administration is, on the Report’s theory, not an underperforming population but a locked door, and the lock is the graded structure of standing and schooling the indicators measure. This is also the point at which the Report’s arithmetic meets its severest technical criticism, which belongs to the adjudication below but should be flagged at the scene: the 52 percent is a residual, not a count. No enumerator in 1980 counted 52 Indians in every hundred and found them backward; the figure is what remains when every other category has been deducted from a fifty-year-old census, and its accuracy is hostage to the accuracy of the deductions. Yogendra Yadav, a defender of reservation and no friend of the Report’s enemies, put the indictment in its most quoted form: there is no empirical basis to the Mandal figure, he wrote; it is a mythical construct arrived at by reducing the numbers of the Scheduled Castes and Tribes, of Muslims and of others, and treating what is left as the backward population. The National Sample Survey’s 1999-2000 round, estimating the Other Backward Classes at about 36 percent of the population, and the National Family Health Survey’s lower figures still, gave the criticism an empirical sequel the Commission could not have answered from its 1931 base. What the defenders reply, then and since, is that the rival estimates measure a different object: surveys that ask households to name their own category, decades after the category acquired a political charge, do not recover the caste-by-caste enumeration the Commission performed, and the direction of the discrepancy between a deduction from caste tables and a self-report in a sample survey is exactly what the sociology of the subject predicts. The dispute is genuine and it is not settled by assertion in either direction; it is recorded here at the point where the figure enters the argument, because the reservation quantum derived in the next section inherited both the figure’s authority and its fragility.

Deriving Twenty-Seven: The Residual Quantum

The Report’s path from 52 percent of the population to 27 percent of the posts is the clearest window into its character as a legal-political instrument, and the recommendations chapter states the reasoning without embarrassment. The Commission records that it first argued the proportional case to itself: a set of classes comprising 52 percent of the population should, in logic, receive reservation in that proportion. It then records why logic had to yield. The Supreme Court, in the line of authority running from Balaji, had laid down that the total of reservations must remain below 50 percent; the Scheduled Castes and Scheduled Tribes already held 22.5 percent between them, 15 percent and 7.5 percent respectively; the residue available beneath the ceiling was therefore 27.5 percent at the very most, and the Commission, choosing to keep the aggregate visibly below the line rather than flush against it, recommended 27 percent, for a grand total of 49.5 percent. The number that would detonate a decade of politics was thus not a demand but a remainder: the largest figure the Commission believed a court would sustain, fixed by subtraction from a judicial ceiling, and presented in the Report with the subtraction shown. This is the residual-quantum architecture, and it explains a feature of the subsequent history that purely political accounts miss. When the Supreme Court in Indra Sawhney finally fixed the ceiling at 50 percent as a binding rule and upheld the 27 percent, it was not imposing an alien discipline on the Report; it was ratifying the discipline the Report had imposed on itself in 1980, a full judicial generation earlier. The Commission had drafted for the courtroom because the Kalelkar report had died for want of a court-proof design, and the drafting held.

The half-point of headroom in that arithmetic, the difference between the residue of 27.5 and the recommendation of 27, is a small detail with a large meaning. Balaji had spoken of reservation remaining, broadly, below 50 percent, and the adverb was an invitation to argue that a total of exactly half might pass. The Commission declined the invitation. By fixing the grand total at 49.5 percent it made the Report’s central demand defensible without a preliminary fight over the meaning of broadly, and it signalled to the courts that the scheme had been built inside their doctrine rather than against it. The asymmetry the arithmetic created should also be faced squarely, since critics have made much of it: the Scheduled Caste and Scheduled Tribe quotas stood in rough proportion to those populations, while the backward classes, counted at 52 percent, received a fraction of their proportion, so that proportionality governed the older quotas and was expressly abandoned for the new one. The Report does not hide the asymmetry; it presents the ceiling as a legal fact of the same order as the demographic facts, and treats the space between 52 and 27 as the price of a recommendation that could be implemented on the day it was accepted rather than after a generation of constitutional litigation. Whether that price was prudence or timidity was argued in 1990 and is argued still; that it was calculated, rather than stumbled into, is what the derivation shows.

The same section of the recommendations quietly performs a second operation that deserves notice. The 27 percent was recommended for the classes identified as a whole, one undifferentiated beneficiary set, even though the Report’s own tables graded that set and marked 2,108 of its castes as depressed backward classes. The majority’s reasoning, reconstructed from the Report and from the chairman’s covering letter, was that apportionment within the quota was a refinement the first implementation could not carry: a single quota, filled from a single list, was administrable by the machinery that existed, and internal division would multiply litigation and invite the very list-politics that had destroyed the first Commission. The cost of the simplification was named at the time, from inside the Commission, by the one member whose dissent the Report carries, and the argument between the majority’s administrability and L.R. Naik’s fear of capture by the stronger backward castes is the argument that the sub-categorisation politics of later decades inherited whole. It is taken up after the recommendations themselves have been set out, since the recommendations are where the majority’s choices are most fully visible.

The Recommendations Taken in Order

The Report’s closing portion gathers its recommendations, and they repay reading as a sequence rather than as the single headline figure the public debate extracted from them. The first group fixes the quota itself. Twenty-seven percent of appointments in the services of the Union and in public sector undertakings were to be reserved for the socially and educationally backward classes. The same proportion was to run through higher education: 27 percent of seats in universities, colleges and the scientific, technical and professional institutions, including privately managed institutions that received aid from the state, were to be reserved likewise. The reach of the quota was drawn deliberately wide. It was to extend beyond the Union’s own offices to the public sector banks, to undertakings in the private sector that received financial assistance from government in any form, and, by the Commission’s recommendation to the states, to state services and state institutions on parallel lines, with the Union and the states asked to make the legal provisions the scheme required. The second group of recommendations supplies the machinery of a working quota, borrowed consciously from the Scheduled Caste and Scheduled Tribe template the administration already knew how to run: a roster system for the reserved appointments on that pattern; age relaxation for backward-class candidates to the same extent as for the Scheduled Castes and Scheduled Tribes; and a carry-forward rule under which a reserved vacancy that could not be filled in its year was to be carried forward for three years before it could be treated as dereserved. The third group is the machinery of the lists themselves: the criteria and the state lists were to be treated as living instruments, open to periodic revision as conditions changed and as errors of inclusion and exclusion came to light, rather than as a settlement carved in stone at the moment of first compilation.

A further rule in the same group governed how the quota was to be counted, and it has had a longer administrative life than any other line in the recommendations. A candidate of the listed classes who obtained appointment in open competition on merit was not to be adjusted against the reserved share: the 27 percent was to operate over and above such selections, so that the quota measured the floor of backward-class representation rather than its ceiling. The rule mattered because without it the reservation would have taxed precisely the candidates it existed to encourage, counting every open-merit success of a backward-class candidate as a discharge of the state’s obligation. With it, the reserved share and the open share remained distinct channels, and the representation the Report sought could accumulate through both. The principle survived every later amendment of the scheme, a small piece of 1980 draftsmanship that no court and no government has seen reason to disturb.

Did the Mandal Commission recommend reservation in promotions?

Yes. The recommendations carried the 27 percent through the whole of a service career: reservation at initial appointment and, in the Report’s words, for promotion at all levels as well. The promotions limb was the one major element the Supreme Court struck down for the Other Backward Classes in 1992.

The fourth group of recommendations is the one the subsequent debate discarded almost entire, and its loss is the largest single distortion in the Report’s public memory. The Commission stated plainly that reservation alone could not transform the condition of the backward classes, and it attached to the quota a programme of structural measures. It recommended land reforms of a thoroughgoing kind, the distribution of ceiling-surplus land and the reorganisation of production relations in the countryside, since the backward classes’ weakness was anchored in their dependence within the village economy. It recommended minor irrigation works and the supply of institutional credit on liberal terms, with a separate financial institution proposed to serve the credit and development needs of the backward classes, and assistance to the traditional artisans, the weavers, oil-pressers, potters and smiths whose crafts the industrial economy was squeezing, to reorganise their production and marketing. It recommended an intensive programme of adult education, special coaching for backward-class students, and residential schools, reasoning that a quota at the point of appointment arrives too late for a population the school system has already filtered out. Read whole, the recommendations describe a theory of remedy in three storeys: change the village economy at the base, change the school pipeline in the middle, and reserve the posts at the top, with the quota as the most visible and least sufficient element. The implementation history inverted the design. The Union in 1990 took the top storey and left the other two largely untouched, a selection Yogendra Yadav has repeatedly remarked upon: reservation was only one of the many recommendations of the Mandal Commission, and the remainder stayed on the page. The methodological reading restores the proportion. The Report’s analytical foundation is not the 27 percent; it is the diagnosis that backwardness is produced at several levels of a structure and must be answered at each, of which the quota is the answer at exactly one.

The Patchwork the Report Nationalised: The States Before Mandal

The Commission’s method did not descend on an empty field, and the recommendations included a direction to the states that is unintelligible without the field’s prior occupation. By 1979 the reservation of posts and seats for backward classes was already the settled practice of the southern and western states, built by their own inquiries over six decades. Mysore had acted on the Miller Committee’s report of 1918 to open its services to the non-Brahmin communities; the Madras Presidency had distributed public employment by communal orders from the 1920s; and the successor states maintained and periodically rebuilt the structure, Karnataka on the report of the Backward Classes Commission chaired by L.G. Havanur in 1975, Tamil Nadu on the report of the Sattanathan Commission of 1971, Andhra Pradesh on the Anantharaman Commission’s work of 1970, Bihar, in a more contested register, on the Mungeri Lal Commission whose recommendations the Karpoori Thakur government carried into a state formula in November 1978, weeks before Desai’s announcement in Parliament. The percentages these states ran were their own, in several cases far above anything the Union contemplated, and their lists were their own, compiled by state criteria of varying explicitness. The Mandal Commission treated this inheritance in the only way its brief allowed: the state lists and state inquiries were received as evidence and as starting enumerations for its own state-wise compilation, its eleven indicators were applied to discipline and revise them, and its recommendations to the states urged the parallel adoption of the 27 percent within state services while leaving each state free, as the Constitution leaves it free, to maintain its own list for its own purposes.

The consequence, visible across the decades since, is that India operates two backward-class regimes that the Report bound together without merging. The central list, descended from Mandal’s compilation and maintained after 1993 by the National Commission for Backward Classes, governs Union employment and central institutions; the state lists, descended from the state commissions and in cases like Tamil Nadu’s protected by devices of their own, govern state employment and state institutions, at percentages that in Tamil Nadu reached 69 percent under a state Act of 1993 placed in the Ninth Schedule the following year. Litigation and politics move constantly between the two regimes, and the boundary disputes they generate, whether a caste listed by a state but absent from the central list may claim a Union post, whether a state’s percentage can stand against the 50 percent rule, are disputes the Report’s architecture predicted by building a national method on state materials. The Commission’s design decision to measure every caste against its own state’s averages, attacked as a source of inconsistency, is what made the dual regime workable: a genuinely all-India metric, computed against national means, would have collided with the state lists at every point and forced a constitutional confrontation the Report was built to avoid. The patchwork was not a failure of nerve. It was the federal price of a national method, and the Commission paid it knowingly.

The Dissent Inside the Document

The Report was not unanimous, and its majority knew the value of carrying its dissent visibly rather than suppressing it. L.R. Naik, the Commission’s only member from the Scheduled Castes, a man whose tour diary the chairman had reason to know, declined to sign the recommendations and recorded a minute of dissent whose argument has outlasted most of the Report’s prose. Naik’s contention, as the chairman’s letter forwarding the Report to the President summarised it, was that the state-wise lists of Other Backward Classes should be split into two parts: one part for the Intermediate Backward Classes and one for the Depressed Backward Classes, the latter grouping those castes which, in Naik’s words, constitute the most deprived and under-privileged sections of the backward classes. The castes in the common list, he wrote, are not at the same degree or level of social and educational backwardness, and he feared that the safeguards recommended for their advancement would not percolate to the less fortunate sections among them. He grounded the fear in observation from the tours: a tendency developing among the Intermediate Backward Classes, he wrote, to repeat against the Depressed Backward Classes the treatment they themselves had received from the upper castes. The majority’s response, embedded in the lists themselves, was partial concession without structural change: the state tables did mark out the depressed stratum, 2,108 castes of the 3,743, but the quota remained one, and the safeguards were not apportioned between the strata.

What did the chairman’s covering letter say about L.R. Naik?

Bindeshwari Prasad Mandal’s letter forwarding the Report to the President carried an unusual tribute: L.R. Naik, the member who had refused to sign its recommendations, was praised as the most hard-working member of the Commission, unfatigued on the long tours when others tired, a courtesy that recorded the dissent without softening it.

The Naik dissent matters to the methodological reading because it identifies, from within, the design decision on which the Report is most vulnerable, and it does so on grounds the majority’s own theory should have honoured. If backwardness is graded, as the twenty-two point scale itself insists, then a remedy that treats the graded universe as flat at the point of distribution has a seam in it, and the seam runs exactly where Naik put his finger: between the advanced backward castes, the Yadavs, Kurmis and Jats of his intermediate column, possessed of land, numbers and political organisation, and the depressed castes of artisans, service castes and the very poor, possessed of none of the three. The majority’s defence was administrative and strategic, that a divided quota in 1980 would have dissolved in litigation and state-level capture before a single appointment was made, and that a unified backward-class politics was itself one of the Report’s objects. History has kept both readings alive. The politics of the following decades delivered the unified constituency the majority hoped for, and it also delivered, state by state, the internal correction Naik demanded: Bihar’s distinction between Backward Classes and Extremely Backward Classes, the sub-quotas carved out in several states, and the national argument over sub-categorisation within the central list are all continuations of the minute of dissent by other means. A method that prints its dissent, and whose dissent turns out to be the seed of its own revision, is a stronger instrument than its critics’ portrait of a partisan fix suggests, and the Report is best read with Naik’s pages bound in, as the Commission itself bound them.

The Method on Trial: The Indictment and the Defence

The named disagreement this article must adjudicate is between the implementation-history framing, which treats the Report as a political event that happened to have a text attached, and the methodological-substantive reading, which treats it as a text that happened to detonate. The indictment of the method, made by scholars and lawyers of standing rather than by the street agitation of 1990, has four counts, and each deserves its strongest form. The first count is composition. Shriram Maheshwari’s critique of 1991 argued that a commission staffed from the beneficiary classes, in pointed contrast to the broadly based Kalelkar Commission, carried its conclusion in its personnel: the method, on this account, is the rationalisation of a verdict reached before the evidence was called. The second count is the weights. V. Gauri Shankar, writing in the Hindustan Times of 24 October 1990, gave the contemporary indictment its enduring phrase: the weightage given to the indicators was arbitrary and illogical, three points for standing, two for schooling, one for wealth, with no derivation offered for the ratios and no published scores by which a reader could verify how any single caste had actually fared under them. The third count is the survey. The Commission’s expert panel, its Research Planning Team of sociologists, had designed the field inquiry with caste and occupation as twin starting points and with safeguards intended to stop the caste factor swallowing the inquiry whole; the anthropologist B.K. Roy Burman, among others, documented how the procedure laid down by the panel was substantially modified in execution, so that caste predominated and the occupational cross-checks withered, and the Report’s own candour that its findings rested in part on the personal knowledge of the members and on the evidence of interested witnesses gave the count its courtroom form: personal knowledge, let loose, is not a method. The fourth count is the figure. The 52 percent, as the counting section recorded, is a computed residue from the census of 1931, and Yogendra Yadav’s verdict stands as the count’s most damaging form precisely because it comes from the reservation camp: no empirical basis, a mythical construct, a number produced by subtraction and defended by repetition. Behind all four counts stands the theoretical objection André Béteille pressed in the essays of 1990 and 1991 that were collected in The Backward Classes in Contemporary India: that to enumerate disadvantage caste by caste is to freeze the very identities a remedy should dissolve, to convert castes into vote banks with a state-printed certificate of backwardness, and to distribute public goods by ascription at the expense of the disadvantaged individual in whatever caste he or she is found. A. Ramaiah’s examination of the identification exercise in the Economic and Political Weekly of 6 June 1992 carried the objection into the lists themselves, questioning how the state-wise compilations had in fact been assembled from their heterogeneous materials.

The defence answers count by count, and its strength is that it rarely needs to deny the facts alleged. To the composition count, the defence replies that the charge proves too much: the Kalelkar Commission’s breadth produced a report its chairman disavowed, and the presence of members who knew the backward-class world from inside is a defect only on a theory of knowledge in which the examined must always be examined by others; S.S. Gill, the Commission’s secretary, answered the partisanship charge in print by pointing to the documentary record, the survey returns, the published criteria, the printed dissent, as the acts of a body that expected to be audited. To the weights count, the defence concedes the ratios are judgements and denies that they are arbitrary: the ordering, standing above schooling above wealth, is derived in the Report from its theory of how backwardness is generated, and a weighting that announced its theory is the opposite of an arbitrary one, since it tells the critic exactly which premise to attack. To the survey count, the defence distinguishes the design from the execution and the execution from the use: the indicator scores were computed from census and survey material by a published rule, and where members’ knowledge supplemented the record, it supplemented a record that existed independently and remains available for re-examination, as the state hearing transcripts and district tables attest. To the figure, the defence gives the answer recorded earlier: the residue method was the only method a caste-blind statistical system left open, its steps are printed, and its rivals, sample self-reports taken generations later, measure a different quantity. Christophe Jaffrelot’s India’s Silent Revolution, the most consequential scholarly treatment of the whole subject, in effect delivers the defence’s summation: Jaffrelot treats the Commission’s sociological premise, that caste is the operative unit of disadvantage in India and must be the unit of remedy, as the Report’s serious and substantially correct core, and the long lower-caste mobilisation he chronicles as the political world that premise described and then remade. Marc Galanter’s Competing Equalities supplies the defence’s jurisprudential half: Galanter had shown, before the Mandal debate opened, that every alternative criterion proposed in the Indian debate, income, occupation, education taken singly, had failed in administration or in court, and that a workable system of compensatory discrimination would have to be a compound of tests anchored in the social structure, which is what the eleven indicators are.

The scholarly argument had a courtroom sequel that belongs to the adjudication, because the Supreme Court’s nine judges in Indra Sawhney were asked to rule on the method itself and did so. The petitioners’ case, as the lead petitioner framed it, ran on three rails: that the extension of reservation violated the constitutional guarantee of equality of opportunity; that caste was not a reliable indicator of backwardness; and that the efficiency of public institutions was at risk. The first and third rails attacked reservation in general and were answered in general terms. The second rail attacked Mandal’s method in particular, and the majority’s answer to it is the closest any institution has come to a judicial review of the eleven indicators. The bench accepted that a caste is a class of citizens for the purposes of the equality code, that caste may serve as the starting point of identification, and that the identification must then be cross-checked and periodically revised, a sequence that ratifies the Report’s design decision to build a compound instrument rather than rest on caste name alone. The dissenting judges would have required more of the cross-checking and less of the caste. Between the majority and the dissent, the method’s centre held: no judge proposed that the identification be abandoned, and none proposed a rival instrument to put in its place. For a document whose enemies had predicted annulment, the judgment’s most telling feature is the absence of an alternative method anywhere in its hundreds of pages.

The complication the honest adjudicator must then face is the one the brief for this article insists upon: the method’s choices were also political choices. The chairman and the majority were not disembodied analysts. Weighting standing above wealth served the claim of the organised peasant castes who formed the backward movement’s spine and who stood to gain from a test that counted their ritual rank and their thin schooling more heavily than their land. Setting the threshold at half the scale yielded a backward universe amounting to a majority of the nation, and a majority is a coalition as well as a statistic. Choosing caste as the unit consolidated a constituency that the Janata coalition, and after it every backward-class party, could address as one. None of this is denied here; it is, on the evidence, true. The methodological-substantive reading survives it, for a reason that goes to what a method is. A method is not falsified by having been chosen for reasons; it is falsified by failing to do its work in the open. Mandal’s choices were published with their reasons, their rule was applied through an auditable procedure, and their output was then subjected to the most hostile examination the Indian system affords: a decade of parliamentary deferral, a nationwide agitation, a nine-judge bench, and forty years of revision by successor commissions. What emerged from that examination was not the method’s corpse but its continuation under amendment, the creamy layer grafted on, the lists sent for periodic review, the ceiling it had anticipated converted into law. An artefact of pure strategy does not behave this way; strategies expire with their strategists, and Bindeshwari Prasad Mandal died in April 1982 with his Report unimplemented and his party in ruins. The method outlived the politics that produced it because it had become, in the strong sense, an analytical foundation: the architecture within which every later actor, including every later critic, has been obliged to work. The implementation-history framing cannot explain that survival, since on its account the Report was a pretext whose hour struck in 1990 and passed. The methodological reading explains it directly. That is the adjudication this article defends.

The Decade of Deferral and the Detonation of 1990

The Report’s first decade is the strongest evidence for the distinction the adjudication has just drawn, since a document can be ignored only if it exists in a form that survives ignoring. The Report was discussed in the Lok Sabha and the Rajya Sabha during 1982 and then referred to a committee of secretaries, after which the governments of Indira Gandhi and of Rajiv Gandhi let it rest, unwilling, as the record of their inaction shows, to spend political capital on its contentious arithmetic. The deferral was itself a tribute to the method: a mere list could have been quietly amended into harmlessness, but a argued document with a printed derivation could only be implemented or refused, and refusal, renewed annually for ten years, was the course two Congress governments chose. What ended the deferral was the arrival of a government whose leader’s position made refusal costlier than action, the story of the V.P. Singh government’s 1990 Mandal implementation: on 7 August 1990, V.P. Singh announced in Parliament that the recommendations would be implemented, the office memorandum of 13 August 1990 gave the announcement administrative form by reserving 27 percent of vacancies in the civil posts and services of the Union and in public sector undertakings for the socially and educationally backward classes, and the Prime Minister confirmed the decision in his Independence Day address of 15 August. The response is the best-remembered part of the story: protests across the universities of the north, a wave of self-immolations by students who saw the queue for government employment reordered against them, among them Rajiv Goswami of Deshbandhu College in Delhi, who set himself alight on 19 September 1990 and survived to become the agitation’s emblem, and, within weeks, a constitutional challenge carried to the Supreme Court, where a five-judge bench stayed the operation of the memorandum while a nine-judge bench was constituted to hear the matter in full. The Mandal moment, in the sense the pillar’s hub gives that phrase in the three-reading adjudication framework, belongs to those months. This article’s subject is the document the months were about, and the point of rehearsing the chronology here is only to fix the interval: ten years in which the Report changed nothing and was changed by nothing, followed by a season in which everything was suddenly referred to it. Documents that are mere pretexts do not impose that rhythm. A method does, because a method waits.

One figure from the deferral decade fixes its character. Bindeshwari Prasad Mandal died in April 1982, in the same season in which Parliament was debating his Report and declining to act on it, sixteen months after he had signed it and eight years before its implementation made his name a household word across the northern states. The chairman’s death removed the one advocate whose authority over the document was personal, and the Report’s survival of its author is the cleanest test the decade affords between the two readings of Mandal this article has adjudicated. A political artefact, a coalition’s promissory note, would have been renegotiated, traded or forgotten with its author’s party in ruins and its author in his grave; the Report instead sat in the record unaltered, its arithmetic unrevised and its demand unwithdrawn, available entire to the government that finally lifted it down in 1990. The committee of secretaries to which the 1982 discussion consigned it worked in the file’s own time; the houses of Parliament returned to the question in session after session without resolving it; and each return was conducted in the Report’s vocabulary, its percentages, its lists, its categories, because no rival vocabulary had been constructed in the interval. Deferral, in the event, was not the document’s defeat but its long examination, and it passed into implementation with its method intact precisely because a decade of opponents had attacked its politics and left its machinery standing.

What the Court Kept: Indra Sawhney and the Amended Architecture

The judgment of 16 November 1992 in Indra Sawhney v Union of India is conventionally read as the Report’s vindication, and it is more accurately read as its second drafting. The nine-judge bench, by a majority of six to three, upheld the 27 percent reservation for the Other Backward Classes in Union employment. It accepted the Commission’s foundational premise in the form the Commission would have chosen: a caste is a class of citizens within the meaning of the equality code, and caste may serve as the starting point for identifying a backward class, though it cannot be the sole criterion and must be cross-checked against other indicia of backwardness, the very structure the eleven indicators embodied. It converted the Balaji guideline on quantum into a rule: reservation shall not exceed 50 percent, save in extraordinary circumstances, thereby ratifying the ceiling arithmetic from which the 27 percent had been derived as a residue twelve years earlier. And it amended the architecture in three places where the Report’s design had left openings. It excluded the creamy layer, the socially advanced persons and sections within the listed castes, from the benefit of the quota, grafting onto a caste-unit scheme the household-level filter the Report had deliberately omitted. It confined reservation to initial appointment and struck down the promotions limb, so that the one recommendation in the service set that reached beyond entry died in the courtroom. And it directed that the lists be reviewed periodically, with a permanent expert body to examine complaints of over-inclusion and under-inclusion, a direction that produced the National Commission for Backward Classes Act of 1993 and converted the Report’s static lists into a maintained register. The implementing office memorandum of 8 September 1993, issued with the creamy-layer criteria framed by the expert committee under Ram Nandan Prasad, put the amended scheme into force. The verdict’s deepest significance for the methodological reading is the identity of what survived scrutiny. The Court did not sustain a political bargain; it sustained a method, its unit, its indicators, its ceiling, while disciplining three of its applications. Every subsequent episode in the reservation story has been fought on ground the judgment and the Report together staked out.

How many judges upheld OBC reservation in Indra Sawhney?

Six of the nine judges. The majority in Indra Sawhney v Union of India, in the judgment delivered on 16 November 1992, upheld the 27 percent reservation for the Other Backward Classes subject to the creamy layer’s exclusion and the 50 percent ceiling; three judges dissented on the central questions.

The Report’s remaining limb, the educational reservation, waited longest of all, and its route back into force displays the architecture’s persistence from another angle. The 27 percent in admissions slept from 1980 until the Constitution (Ninety-third Amendment) Act of 2005 inserted Article 15(5), expressly empowering the state to make provision for the advancement of the socially and educationally backward classes in aided and unaided educational institutions, and the Central Educational Institutions (Reservation in Admission) Act of 2006 carried the quota into the Indian Institutes of Technology, the Indian Institutes of Management, the All India Institute of Medical Sciences and the other central institutions, to be phased in over three years against an expansion of general seats. The protests of 2006 replayed those of 1990 in a milder key, and on 10 April 2008 the Supreme Court in Ashoka Kumar Thakur v Union of India upheld the amendment and the Act, applying the creamy-layer exclusion to admissions as the 1992 judgment had applied it to appointments. The methodological continuity across the twenty-eight years from Report to final educational implementation is the point to hold: the beneficiary set, the percentage, the exclusion and the ceiling that entered the central institutions in 2008 were all quantities fixed by the interaction of the 1980 method with the 1992 judgment, not fresh political inventions of 2006. The contrast that proves the rule arrived in 2019, when Parliament enacted a reservation of ten percent for the economically weaker sections outside the existing categories. That scheme, treated in its own place later in this series, works by an entirely different method: an individual and household test of income and assets, no caste unit, no indicators of standing or schooling, no enumeration of communities. To build a reservation unlike Mandal, the state had to build a method unlike Mandal, indicator by indicator and premise by premise. Nothing could demonstrate more cleanly that the substance of the 1980 Report was its method, since replacing the politics required replacing the machinery whole.

The Method Is the Substance of Mandal

The claim with which this article opened can be restated in the light of the evidence assembled. The December 1980 Mandal Report is misread when it is read as a 27 percent demand with a long document attached. The document is the demand’s meaning: a criteria-first identification of backwardness built under the discipline of Balaji, a three-tier weighting that encoded a theory of how Indian disadvantage is produced, an enumeration of 3,743 castes executed against a fifty-year-old statistical base and stated with its derivation open, a population figure of 52 percent arrived at by deduction and carrying the marks of its method, a reservation quantum computed as the residue beneath a judicial ceiling, a set of recommendations that ranged from village production relations to service rosters, and a printed dissent that identified the design’s weakest seam a decade before the politics of sub-categorisation found it. The implementation-history framing, which dominates the textbooks and the anniversary journalism, tells a true story about 1990 and a false one about causation, since it cannot say why the politics took the shapes it did, why the Court’s amendments fell where they fell, why the education limb could sleep for twenty-six years and wake unchanged, or why a rival reservation in 2019 had to be built from different parts. The methodological-substantive reading answers each of those questions from the same premise: the post-1990 trajectory operates through the methodology’s architecture, amending its applications while reproducing its categories, because the Commission of 1979 to 1980 did the one thing the Kalelkar Commission had not done. It built an instrument. Instruments outlast the hands that commission them, and this one, audited by its enemies for forty years, is still the instrument in use.

Frequently Asked Questions

Q: What is the Mandal Commission Report?

The Mandal Commission Report is the report of the Second Backward Classes Commission, chaired by Bindeshwari Prasad Mandal and submitted to the President of India on 31 December 1980. Appointed under Article 340 of the Constitution to identify the country’s socially and educationally backward classes and to recommend measures for their advancement, the Commission devised eleven indicators of backwardness grouped under social, educational and economic heads, weighted them three, two and one, and applied them state by state. It identified 3,743 castes and communities, covering about 52 percent of the population, as Other Backward Classes, and recommended 27 percent reservation for them in central government employment, public sector undertakings and educational institutions, alongside land reform, credit and educational measures. Shelved for a decade, the Report was implemented in part from 1990 and upheld, with significant amendments, by the Supreme Court in 1992.

Q: Who was B.P. Mandal?

Bindeshwari Prasad Mandal, born in 1918 in Murho village of the Madhepura district of Bihar, was a backward-class political leader of the Yadav caste who chaired the Second Backward Classes Commission of 1979 to 1980. He began in the Congress, moved through Bihar’s socialist politics, and in February 1968 became chief minister of Bihar at the head of a coalition built around his Shoshit Samaj Dal, an office he held only for a matter of weeks before the coalition dissolved. Elected to the Lok Sabha from Madhepura in the Janata victory of 1977, he was chosen by the Morarji Desai government to head the commission whose report took his name. He signed the Report on 31 December 1980 and died in April 1982, eight years before its implementation made him a posthumous figurehead of the backward-class movement across northern India.

Q: What are the eleven Mandal indicators?

The eleven indicators are the tests the Mandal Commission evolved to determine social and educational backwardness, set out at paragraph 11.23 of its Report. Four are social: a caste considered socially backward by others; a caste mainly dependent on manual labour; a caste whose marriages below the age of seventeen exceed the state average by set margins, 25 percent for females in rural areas with parallel male and urban thresholds; and a caste whose women’s participation in work stands at least 25 percent above the state average. Three are educational: children aged five to fifteen who never attended school, the dropout rate in that age band, and the share of matriculates, each measured against the state average by a margin of 25 percent. Four are economic: average family assets, families in kuchcha houses, households whose drinking water lies beyond half a kilometre, and households carrying consumption loans, again measured against the state averages.

Q: How was 27 percent reservation calculated?

The 27 percent was derived as a residue beneath a judicial ceiling, and the Report prints the derivation. The Commission first reasoned that classes comprising 52 percent of the population should in logic receive reservation in that proportion. It then applied the rule the Supreme Court had laid down in the Balaji case of 1963, that total reservation should remain below 50 percent. The Scheduled Castes already held 15 percent and the Scheduled Tribes 7.5 percent, together 22.5 percent, so the space remaining beneath the ceiling was 27.5 percent at most. The Commission recommended 27 percent rather than the full residue, keeping the grand total at 49.5 percent and visibly below the line the Court had drawn. The figure was therefore a calculated remainder, fixed by subtraction from a legal limit, not a proportion of the population counted and not a figure bargained out between parties.

Q: Why are 3,743 castes OBC?

The figure of 3,743 is the aggregate output of the Commission’s method applied state by state, not a number chosen in advance. For each state, the Commission assembled the castes found in the state lists, the census record and its own socio-educational survey, and scored every caste under the eleven indicators, which carried three points for each social test, two for each educational test and one for each economic test, to a maximum of twenty-two. A caste scoring eleven points or more, half the scale or above, was classified as socially and educationally backward and entered in the list of Other Backward Classes; a caste below the line was treated as advanced. Summed across the states, the castes above the line came to 3,743 castes and communities. Within that total the Report’s state tables separately marked 2,108 castes as depressed backward classes, the most underprivileged stratum of the backward universe.

Q: What is the creamy layer?

The creamy layer is the advanced section within the Other Backward Classes, the socially and economically forward persons and families inside the listed castes, whom the Supreme Court excluded from the benefit of reservation in Indra Sawhney v Union of India on 16 November 1992. The Mandal Commission’s Report itself contained no such exclusion: its unit of benefit was the caste, and a listed caste’s prosperous households stood inside the scheme with its poor ones. The Court, upholding the 27 percent, grafted the exclusion on as a condition, holding that the quota must not be captured by the advanced segments of the beneficiary classes. The Union government operationalised the concept through the office memorandum of 8 September 1993, applying criteria framed by an expert committee under Ram Nandan Prasad, which combined parental office and occupation with an income ceiling that governments have revised from time to time. The exclusion was later applied to admissions in central educational institutions as well.

Q: When was the Mandal Report submitted?

The Mandal Report was submitted on 31 December 1980, when the Commission, having completed a little under two years of work since its constitution on 1 January 1979, presented its findings and recommendations to the President of India, Neelam Sanjiva Reddy. The government of the day was Indira Gandhi’s Congress(I) administration, which had returned to office that January and had not commissioned the inquiry placed in its hands. The date closed the Commission’s formal life but opened the Report’s long deferral: the document was debated in both houses of Parliament during 1982, referred to a committee of secretaries, and then left unimplemented by successive governments for the remainder of the decade. Its chairman did not live to see its fortunes change; Bindeshwari Prasad Mandal died in April 1982, and the Report waited until 7 August 1990 for a government, V.P. Singh’s National Front, to announce that its recommendations would be carried into effect.

Q: Who were the members of the Mandal Commission?

The Commission was chaired by Bindeshwari Prasad Mandal, a sitting member of the Lok Sabha and a former chief minister of Bihar. Its members were Dewan Mohan Lal, L.R. Naik, K. Subramaniam and Dina Bandhu Sahu, and its work was serviced by a secretariat under S.S. Gill, an officer of the Indian Administrative Service who served as the Commission’s secretary. The composition was remarked upon from the beginning and has been debated since: of the Commission’s five members including its chairman, four came from the Other Backward Classes and the fifth, L.R. Naik, was its only member from the Scheduled Castes, a distribution the historian of administration Shriram Maheshwari described in his 1991 critique as shaped on partisan lines in contrast to the broadly based Kalelkar Commission. Defenders replied that a body charged with investigating the condition of backward India was entitled to members who knew that condition from inside. Naik’s was the Commission’s single dissenting voice, recorded in a minute appended to the Report.

Q: What were the Mandal Commission’s terms of reference?

The terms of reference were set by the presidential order of 1 January 1979 that constituted the Commission under Article 340 of the Constitution, and they were fourfold. The Commission was to determine the criteria for defining the socially and educationally backward classes of India. It was to recommend the steps that should be taken for the advancement of the classes identified under those criteria. It was to examine the desirability of reserving appointments and posts in the services, and seats in educational institutions, in favour of those classes, a term that joined the identification exercise to the reservation question from the outset. And it was to present its report, with its recommendations, to the President. The order of the terms shaped the document the Commission produced: criteria came first, lists were derived from criteria, and the reservation recommendation was built on the lists, the sequence the Report’s defenders cite against the charge that its conclusions were reached before its evidence was heard.

Q: How did the Mandal Commission gather its evidence?

The Commission built its evidence from four sources, having found that the central fact it needed, the population and condition of each caste, had not been officially measured since the census of 1931. From that census it took caste-wise population figures, extrapolated forward, supplementing them with the published statistical series and with the backward-class lists the state governments maintained for their own reservations. It issued questionnaires to members of Parliament, state legislators, voluntary organisations and the public, and it toured the states holding public sittings at which caste associations, officials and scholars gave evidence, testimony preserved in hearing transcripts held in the archives of the Ministry of Social Justice and Empowerment. Its largest single undertaking was a socio-educational field survey launched in February 1980, in which enumerators of the state statistical organisations canvassed every household in two villages and one urban block in 405 of the country’s 406 districts, the returns tabulated by computer with the National Informatics Centre’s assistance. A panel of experts and a research planning team of sociologists designed the instruments through which this material was converted into findings.

Q: How did the twenty-two point scale classify a caste as backward?

The scale converted the eleven indicators into a single composite score for each caste. Every social indicator a caste satisfied added three points, every educational indicator two points and every economic indicator one point, so that the four social tests could contribute twelve points at most, the three educational tests six and the four economic tests four, for a maximum score of twenty-two. Each test was assessed against the average of the caste’s own state, the recurring margin being 25 percent above or below that average. A caste accumulating eleven points or more, half the available total, was listed as socially and educationally backward; a caste below eleven was treated as advanced. The weights encoded the Commission’s doctrine that economic deprivation flows from social and educational backwardness rather than the reverse, which is why standing in the estimation of others counted three times as heavily as the value of a family’s assets. The Report published the rule and the resulting lists, though not the individual score obtained by each caste.

Q: Why did the Mandal Commission treat caste rather than income as the key to backwardness?

The Report’s reasoning was theoretical before it was administrative. It held that caste in India is not a ritual survival laid over an economy of individuals but the structure through which occupation, standing and opportunity are in fact distributed, so that the socially and educationally backward classes of the Constitution’s language present themselves to inquiry as castes. Economic condition, in this account, is largely a consequence of position in that structure, and a test of income alone would classify the poor of every caste together while missing the disabilities, exclusion from office, from schooling, from the estimation of others, that attach to caste standing independently of a household’s means in a given year. The Report stated the point directly: the substitution of caste by economic tests, it wrote, will amount to ignoring the genesis of social backwardness in Indian society. It did not discard economic evidence; it demoted it, giving the economic indicators the lowest weight in its scale, one point each, as measures of a condition whose causes it located in the social and educational heads.

Q: What did L.R. Naik’s note of dissent propose?

L.R. Naik, the Commission’s only member from the Scheduled Castes, declined to sign the recommendations and proposed that the state-wise lists of Other Backward Classes be divided into two parts: one for the Intermediate Backward Classes and one for the Depressed Backward Classes, the most deprived and under-privileged sections of the backward universe, with the safeguards apportioned so that each stratum received its own protection. His stated ground was that the castes gathered in the common list were not at the same degree or level of social and educational backwardness, and that without a division the benefits would not percolate to the less fortunate sections among them. He recorded, from the Commission’s tours, a developing tendency among the intermediate classes to repeat against the depressed classes the treatment they themselves had received from the upper castes. The majority declined the division, holding a single quota administrable where a divided one would dissolve in litigation, but printed his minute in the Report, whose state tables separately identified 2,108 castes as depressed backward classes.

Q: Which census supplied the Mandal Commission’s caste figures?

The census of 1931, the last enumeration in which the Indian census recorded caste in full across the population. The 1941 census was curtailed by the war, and the census of 1951, conducted after Independence in the name of a casteless citizenship, counted caste only for the Scheduled Castes and Scheduled Tribes, a practice every later census followed. A commission directed in 1979 to identify the backward classes of the whole country therefore had no caste-wise population data younger than forty-eight years, and the Report says so plainly: it took the 1931 caste tables as its base, extrapolated their proportions forward, and checked them against its field survey and the state lists. The antiquity of the base became the most technical of the charges against the Report, pressed by critics who noted that the celebrated figure of 52 percent rested on proportions recorded before Independence. The Commission’s answer was that the state had left it no alternative instrument, and that a stale count honestly derived was a better foundation than a fresh guess.

Q: What did the Mandal Commission recommend beyond reservation in government jobs?

The Report’s recommendations ranged well beyond the appointment quota, and its authors stated explicitly that reservation alone could not transform the condition of the backward classes. It recommended the same 27 percent in admissions to universities, colleges and professional institutions, including aided private institutions, and in public sector undertakings and banks. It recommended thoroughgoing land reforms, the distribution of ceiling-surplus land, minor irrigation works and the reorganisation of production relations in the village economy in which the backward classes were held in dependence. It proposed institutional credit on liberal terms, with a separate financial institution to serve the backward classes’ credit and development needs, and assistance to traditional artisans, weavers, oil-pressers, potters and smiths among them, to reorganise the production and marketing of their crafts. And it recommended an intensive programme of adult education, special coaching for backward-class students and residential schools, on the reasoning that a quota at the point of appointment arrives too late for a population the school system has already filtered out. Most of this programme remained unimplemented when the appointment quota was carried into force in 1990.

Q: What happened to the Mandal Report between 1980 and 1990?

For ten years, formally nothing and politically a great deal. Received by Indira Gandhi’s government on 31 December 1980, the Report was debated in the Lok Sabha and the Rajya Sabha during 1982 and then referred to a committee of secretaries, after which the governments of Indira Gandhi and Rajiv Gandhi let it rest, unwilling to spend political capital on its contentious arithmetic. Its chairman died in April 1982, in the season of the parliamentary debate, and the document survived him unaltered, its demand neither withdrawn nor revised. Opposition parties and backward-class organisations returned to it in Parliament session after session, so that the deferral was conducted in the Report’s own vocabulary of percentages and lists. The decade ended with the National Front government of V.P. Singh, whose coalition rested heavily on the backward-class electorate of the northern states: on 7 August 1990 the Prime Minister announced in Parliament that the recommendations would be implemented, and the office memorandum of 13 August 1990 gave the announcement administrative form.

Q: Did the Mandal Commission’s lists include communities outside the Hindu fold?

Yes. The Commission’s population arithmetic counted about 8 percent of the national population as backward classes drawn from the non-Hindu communities, within the total of 52 percent it assigned to the Other Backward Classes. Its working rules for that part of the enumeration were analogical. Communities that had belonged to castes treated as untouchable or backward before conversion to Islam, Christianity, Sikhism or Buddhism carried their standing across the conversion, since the disabilities the Commission was measuring attached to the community’s social position and occupation rather than to its professed faith. Occupational communities found in every religion, the dhobis who wash, the lohars who work iron, the nais who cut hair, the telis who press oil, were assessed by the standing of their craft, by the same indicators applied to their Hindu namesakes. The state lists accordingly contain Muslim, Christian, Sikh and Buddhist entries alongside the Hindu castes, a feature of the compilation that later litigation and list revision inherited entire, and that reflected the Report’s consistent premise: the unit of backwardness is a community’s place in the social order, which no change of creed by itself rearranges.

Q: What did the Mandal Commission find about OBC representation in central government services?

Tabulating the composition of the central services from the government’s own returns, the Commission found that the classes it had identified as backward, approximately 52 percent of the population, held 4.83 percent of Class I posts and 10.63 percent of Class II posts under the central government. The Report placed these ratios at the centre of its case for reservation. On its theory, a population comprising half the nation and a twentieth of its senior administration was not an underperforming group but evidence of a structure: the same graded order of standing and schooling that its eleven indicators measured in the villages reappeared at the top of the state as a near-monopoly of the forward castes over the higher administration. The figures were compiled for the services as a whole and have been quarrelled with in detail like every number in the document, but their argumentative function was precise. They converted the abstract charge of exclusion into a measurable disproportion, and they set the benchmark against which the working of the 27 percent quota after 1993 would later be audited.

Q: Did the Mandal Commission intend its lists to be permanent?

No. The Report treated its criteria and its state lists as revisable instruments, to be reviewed periodically as social and educational conditions changed and as errors of inclusion and exclusion came to light, rather than as a settlement fixed for all time at the moment of first compilation. The intention was only imperfectly executed for a decade because the Report itself lay unimplemented, but the Supreme Court wrote it into law in Indra Sawhney in 1992, directing that the lists be reviewed at intervals and that a permanent expert body examine complaints of over-inclusion and under-inclusion in them. Parliament answered with the National Commission for Backward Classes Act of 1993, and the central list has been maintained as a living register since, growing well beyond the original 3,743 entries as castes were added on the advice of that machinery and others were scrutinised. The design is consistent with the method’s logic throughout: a classification derived from measurements against moving state averages cannot honestly claim to be final, since the averages move.

Q: How did the Mandal Commission treat states that already had backward classes lists?

It received them as evidence and as starting enumerations, and then subjected them to its own tests. By 1979 several states had maintained backward-class reservations for decades on the authority of their own inquiries, Karnataka on the Havanur Commission’s report of 1975, Tamil Nadu on the Sattanathan Commission’s of 1971, Andhra Pradesh on the Anantharaman Commission’s of 1970, and the state lists compiled under those exercises were laid before the Mandal Commission as it built its state-wise lists. The Commission applied its eleven indicators to the castes those lists contained, retaining, revising and extending entries by the common standard, which is why its compilation is state-wise throughout and why a caste may appear in one state’s list and not in its neighbour’s. It recommended that the states adopt the 27 percent in their own services in parallel, while leaving each state free to maintain its own list for state purposes. The result was the dual regime India still operates: a central list for Union employment descended from Mandal’s compilation, and state lists with histories, and in some states percentages, of their own.