The most durable number in Indian social policy is twenty seven per cent. The Mandal Commission proposed it in December 1980 as the share of central government posts and central educational places to be reserved for the socially and educationally backward classes, the executive adopted it in August 1990, the Supreme Court sustained it in November 1992, and no government or court has moved it since. A reader who knows only that number, however, knows the least contested part of the subject. The argument that has run without interruption from 1953 to the present decade is not about the share. It is about the list: which communities count as backward, who has the authority to say so, by what evidence, and whether the benefits inside the category should be shared out again among its own members. Every major event in this history is an event in the life of the list. A commission drew it up and disowned it. A second commission drew it up and watched it sit unread for a decade. An office memorandum borrowed it from two sources at once. A court pruned it by inventing the creamy layer. Parliament constitutionalised the body that advises on it, was told it had handed the list to the President alone, and amended the Constitution again within the year to take the states’ share back. A third commission spent six years examining whether the list should be divided against itself, reported in July 2023, and its report has not been published.

This article is the legislative and constitutional history of that sequence. It traces how reservation for backward classes moved from a commission report of 1980 to an executive order of 1990 to a set of constitutional provisions, and it keeps the judicial doctrine at the level a history needs, because the doctrine has its own owner in this series and the history is incomplete without the distinction. The wider field within which this story sits, protective criminal law, welfare legislation and the equality code read together, is mapped in the guide to social justice legislation in India; what follows isolates one strand of that field, the affirmative strand for backward classes, and follows it from the framing of the Constitution to the unpublished report of 2023. The claim the record supports, and the claim this article defends, is the one stated above: the quantum was settled in 1990 and has not been reopened with any success, while the identification question has been reopened by a commission in 2017, a constitutional amendment in 2018, a judgment in 2021, a further amendment in 2021 and a report submitted in 2023, so that the contested question in this field has never been how much, but who.

The constitutional design: a deferred question and the machinery for answering it

The Constitution of 1950 named two disadvantaged groups with precision and left a third deliberately unnamed. The Scheduled Castes and the Scheduled Tribes were to be specified by presidential order under Articles 341 and 342, and their specification was a closed administrative act followed by parliamentary control over change. For the residue, the classes that were backward without being either scheduled group, the framers wrote enabling words rather than a list. Article 15(4), inserted by the First Amendment in 1951 after the Supreme Court in State of Madras v. Champakam Dorairajan held that the equality code as originally framed left no room for the communal quotas Madras had operated, permits the state to make special provision for the advancement of any socially and educationally backward classes of citizens. Article 16(4) permits the state to reserve appointments or posts in favour of any backward class of citizens which, in the opinion of the state, is not adequately represented in the services. Neither clause says which classes qualify. The judgment of who is backward was left to be formed later, and the Constitution supplied an instrument for forming it.

Why did the Constitution not name the backward classes itself?

The Constitution named the Scheduled Castes and the Scheduled Tribes through presidential orders because those lists were ready to hand from the scheduling practice of the colonial state. No agreed all India list of other backward classes existed in 1950, and the Constituent Assembly chose to defer the question to inquiry rather than guess at it in the text.

Article 340 therefore empowered the President to appoint a commission to investigate the conditions of socially and educationally backward classes and the difficulties under which they labour, and to recommend the steps the Union or a state should take to remove those difficulties, with the report laid before Parliament together with a memorandum of the action taken. The design made identification an evidence gathering exercise conducted periodically by a temporary body, rather than a permanent constitutional schedule, and that single design choice explains why the politics of this subject keeps returning to commissions.

Article 340 is worth reading closely, because the whole history in this article is the working out of its three beats. The President appoints the commission and fixes its reference. The commission investigates conditions and recommends. The President then causes a copy of the report to be laid before each House of Parliament with a memorandum explaining the action taken on the recommendations. Nothing in the article obliges any government to accept anything. A report under Article 340 is advice dressed in constitutional form, and its force is exactly the political force its contents can generate. The Kalelkar Commission discovered that a report can be laid and buried. The Mandal Commission discovered that a report can wait ten years and then be implemented by a government that needs it. The Rohini Commission has discovered that a report can be submitted and not published at all, which is a fourth outcome the text does not mention and does not prevent.

The enabling clauses around Article 340 complete the design. Article 46, among the Directive Principles, directs the state to promote with special care the educational and economic interests of the weaker sections of the people. Article 15(4) and Article 16(4) convert that direction into permission where the fundamental rights are concerned: a special provision for backward classes is not a breach of the equality guarantee but an exercise the Constitution contemplates inside it. The doctrinal elaboration of that point, the reading of clause (4) as an emphatic statement of what clause (1) already allows rather than an exception to it, belongs to the judicial history. For the legislative history the point is simpler. From 1951 onwards, any government that wished to reserve posts or seats for backward classes had the constitutional room to do so by executive order, without legislation, provided it could say who the backward classes were. The room was never the problem. The saying was.

Two features of the 1951 settlement deserve notice, because both were choices and both had consequences. The first is the phrase the First Amendment used. Article 15(4) speaks of socially and educationally backward classes, a compound test: a class must be backward in both registers, social and educational, and the word chosen for the unit is class, not caste. The Constituent Assembly and the amending Parliament had the word caste available, used it elsewhere in the Constitution, and declined it here, and that declension generated fifty years of argument about whether caste could be the measure of a class, an argument the courts eventually answered by treating caste as a permissible starting point and an impermissible finishing point. The second feature is the location of the machinery. Article 340 sits in Part XVI of the Constitution, among the special provisions relating to certain classes, beside the articles that schedule the castes and tribes and the articles that reserve their legislative seats. The placement states the framers’ conception: the backward classes were a problem of the same family as the scheduled groups, to be handled by specification and safeguard, but with the specification left to inquiry because the material for it did not exist in agreed form. Every government since has inherited both features, the compound test that must be administered and the inquiry machinery that must be staffed, and the history of the subject is the history of the machinery being used reluctantly, late, and under pressure.

There is also a road the design did not take, and its absence shaped the politics. Article 340 is wholly discretionary: the President may appoint a commission, and the article fixes no deadline for the first appointment, no periodicity for later ones and no consequence for a report ignored. The neighbouring Article 339, on the administration of the Scheduled Areas and the welfare of the Scheduled Tribes, does impose a timetable, requiring a commission at the expiration of ten years; the backward classes article contains nothing of the kind, and the contrast is the clearest evidence that the omission was deliberate. After the Kalelkar report’s reception, no government volunteered for the exercise again until a coalition manifesto compelled it in 1979. Identification therefore happened in bursts, separated by decades, each burst reopening every question the previous one had closed. A subject administered continuously, with a standing body revising a list entry by entry against fresh data, would have produced a different and duller politics. The Republic chose the burst pattern, and the drama of 1990, of 2018 to 2021 and of the unpublished report of 2023 is the burst pattern working as designed.

The first attempt: the Kalelkar Commission and the report that argued with itself

The first commission under Article 340 was appointed by presidential order on 29 January 1953, with the Gandhian writer and social reformer Kaka Kalelkar in the chair. Its reference was the template the article sets: determine the criteria for identifying socially and educationally backward classes, prepare lists of such classes, investigate their conditions, and recommend steps for their advancement. The commission worked for over two years and submitted its report on 30 March 1955. It prepared a list of 2,399 backward castes and communities for the whole country, of which 837 were classified as the most backward. Its criteria tied backwardness to low position in the traditional caste hierarchy, lack of general educational advancement among the major section of a community, inadequate representation in government service, and inadequate representation in trade, commerce and industry. On remedies it was generous. It proposed reservation of at least 25 per cent of Class I posts, a third of Class II posts and 40 per cent of Class III and Class IV posts, 70 per cent of seats in technical and professional institutions, caste wise enumeration in the census of 1961, and a wide programme of land reform, village economic reorganisation, rural industries, housing, public health and adult literacy.

Two of its proposals show how far the commission was prepared to go, and both repay attention because later generations rediscovered them as novelties. The first was enumeration: the commission recommended that the census of 1961 record the population caste by caste, so that the identification it had improvised from pre war material could be placed on a counted base. The recommendation was not taken, the census of 1961 asked no caste question outside the schedules, and the data question that dominates the modern debate was thus opened, and declined, at the first opportunity the Republic had to close it. The second was its treatment of sex as a ground of disadvantage: the commission proposed that women as a class be treated as backward for the purposes of its scheme, a classification that sat uneasily with its caste based list and was among the points its critics, including its own chairman in retrospect, found least defensible. Both proposals mark the report as a more radical and more confused document than its reputation for caution suggests. It proposed to count everyone, to classify by caste, to benefit women as a class and to reserve on a scale the Mandal Commission would later halve, and it could not persuade its own members, in the end, that the parts cohered.

The report then destroyed its own authority. Several members recorded dissents from its principal findings. The chairman, in the letter transmitting the report, distanced himself from the exercise he had led, expressing doubt about caste as the criterion of backwardness and about whether the remedies proposed might not prove worse than the disease they were meant to cure. A commission that could not hold its own chairman gave the government of the day, led by Prime Minister Jawaharlal Nehru, every reason it needed to let the matter rest. The report was laid before Parliament and not implemented. The official reasons given turned on the commission’s indecision and on its reliance on caste as the sole and dominant test, and those reasons were not mere pretext: a list built on one contested criterion, disowned in part by its own authors, was a fragile foundation for a national policy of preferential appointment. In 1961 the Union government settled for the opposite architecture, declining to prepare an all India list of backward classes and leaving each state to identify its own backward classes and decide its own measures. The government’s handling of the burial is itself instructive. The report was not rejected by a vote or answered by a counter report; it was received, and the Union Home Ministry’s position, communicated to the states in 1961, was that the government did not propose to draw up any all India list of backward classes, that the states were free to identify backward classes for their own purposes, and that in doing so they would do better to use economic criteria than caste. Each element of that position failed in practice while succeeding as deferral. The states did not adopt purely economic criteria, because their lists were already caste lists and their politics ran on them. No all India list was drawn up, but the absence of one merely privatised the question to twenty jurisdictions. And the economic criterion, commended in 1961 as the respectable alternative, returned in 1991 as the memorandum the Court struck down, and in 2019 as the amendment the Court upheld, suggesting that the argument between caste and income as the marker of disadvantage was never a disagreement that evidence could close; it is the permanent second question of the field, running underneath the list question and periodically breaking through it. For two decades after 1955, backward classes policy at the Centre effectively did not exist, while the states built the policy piecemeal. That division of labour, dormant Centre and active states, is the first form of the dual structure that runs through the entire subject and that the amendments of 2018 and 2021 would later fight over.

The Kalelkar episode set three patterns that recur. The first is that the commission form does not settle the identification question; it relocates it, because a commission’s list is only as strong as the government’s willingness to own it. The second is that caste as a criterion was contested from the very beginning, inside the first commission itself, and the contest was never afterwards about whether caste counted but about how much it should count against other markers of disadvantage. The third is that the non-reservation half of a backward classes report, the land, credit and education measures, attracts no constituency. Kalelkar’s welfare programme was as extensive as Mandal’s would be a generation later, and it shared Mandal’s fate: the reservation recommendation is the only part of any such report that anyone campaigns on, for or against.

The states move first: backward classes policy before Mandal

Because the Centre vacated the field, the early history of backward classes reservation is state history, and any account that begins in 1979 mistakes the subject. The princely states and provinces of the south had operated communal quotas in services long before independence, and the states of the Indian Union continued and expanded their own arrangements through the 1950s and 1960s under their own lists. Mysore’s order reserving 68 per cent of seats for backward classes and the scheduled groups together produced the Supreme Court’s first major encounter with the quantum question in M. R. Balaji v. State of Mysore, where the Court warned that a special provision should ordinarily remain below fifty per cent and struck the order down. The fifty per cent idea, which later generations would attribute to 1992, was therefore present in the case law thirty years earlier, formed in a dispute about a state list. Tamil Nadu and Karnataka maintained extensive state quotas through the decades when the Centre had none. The identification question in this period was answered administratively, state by state, list by list, with no national standard and no national controversy, because the stakes were regional.

The state practice had roots deeper than the Constitution. In 1902 the Maharaja of Kolhapur, Shahu, ordered that half the posts in his state’s services be reserved for the backward classes, an act conventionally counted the first reservation order in India. Mysore appointed a committee under Sir Leslie Miller in 1918 to examine the representation of non-Brahmin communities in its services, and followed it with communal orders apportioning posts. The Madras Presidency under the Justice Party issued communal government orders in the 1920s distributing public employment among communities by proportion. These were the instruments of princely and provincial governments confronting the same fact the Republic would inherit: the services of the state were monopolised by a small literate stratum, and the monopoly would not dissolve by open competition on any timescale a government could defend. When the Constituent Assembly wrote Articles 15(4), 16(4) and 340, it was not theorising in the abstract; it was nationalising a repertoire that the provinces and princely states had built over fifty years, while declining, as those governments had not needed to decline, to say for the whole country at once who the beneficiaries were. The divergence between the state practice, which specified, and the constitutional design, which deferred, is the original tension of the subject, and every later episode, including the intersection device of 1990 and the restoration of 2021, is a renegotiation of it.

The states also developed the intellectual apparatus the Centre would later borrow. Karnataka appointed a commission under L. G. Havanur in 1975 which classified the state’s backward classes into groups of differing backwardness and recommended that benefits be distributed among the groups rather than pooled, an early official statement of internal differentiation. Bihar appointed a backward classes commission under Mungeri Lal in 1971 whose report, submitted in 1976, arranged the state’s backward castes in annexures of graded backwardness. Tamil Nadu’s commissions addressed the exclusion of the advanced within the category. These state exercises matter to the national history for a reason that is easy to miss: when the Mandal Commission was appointed in 1979, it did not begin from a blank page but from a quarter century of state lists, state commissions and state litigation, and its own list was in large part an aggregation of theirs. The identification question had been answered many times before it was asked nationally. What had never been attempted was a single answer binding the Union.

Bihar supplied the dress rehearsal for 1990. In 1978 the state government of Chief Minister Karpoori Thakur adopted a formula reserving 26 per cent of state posts, divided among the backward classes and the most backward classes in separate annexures drawn substantially from the Mungeri Lal Commission’s grading, with shares also for women and for the poor among the forward castes. The formula mattered beyond Bihar for two reasons. It demonstrated that subdividing the backward category was administratively possible a decade before sub-categorisation became a national commission’s reference, and it did so by distinguishing the most backward from the merely backward inside the quota rather than by raising the quota itself. And Bihar was the home state of Bindheshwari Prasad Mandal, himself a former chief minister of the state, who was about to be asked to answer for the whole country the question Karpoori Thakur had answered for one state. The Janata Party, which had promised in its 1977 election manifesto to reserve between twenty five and thirty three per cent of government posts for the backward classes, won power at the Centre and owed its constituency an answer. On 1 January 1979 the President appointed the second Backward Classes Commission under Article 340, chaired by B. P. Mandal. It is conventionally named after him, and the convention is fair: the report’s method, its number and its fate are all inseparable from the politics that produced its chairman.

The Mandal Commission: eleven indicators, 3,743 castes and a number chosen to survive a court

The Socially and Educationally Backward Classes Commission, as it was formally styled, was asked to determine the criteria for identifying socially and educationally backward classes, to recommend steps for their advancement, and to examine the desirability of reservation in appointments. It travelled across the country, heard evidence in the states, and aggregated the state lists and the census material available to it, the last full caste enumeration being that of 1931; it also mounted a socio educational field survey, conducted through state machinery, covering villages and urban blocks in every district, from which the social and educational indicators were scored. On that apparatus it built its identification method out of eleven indicators grouped under three heads. Four social indicators, covering how a caste was regarded by others, its dependence on manual labour, marriage ages and women’s participation in work, carried three points each. Three educational indicators, covering school attendance, dropout and matriculation rates against the state average, carried two points each. Four economic indicators, covering family assets, housing, access to drinking water and consumption debt, carried one point each. A caste scoring more than half of the twenty two available points was classified as backward. The weighting was the commission’s answer to the Kalelkar failure: caste remained the unit of inquiry and the social indicators dominated the arithmetic, but the classification was presented as a composite measurement rather than an assertion about hierarchy, and economic condition was inside the test rather than outside it.

Why did the Mandal Commission fix its figure at twenty seven per cent?

The commission estimated that the backward classes it identified formed about fifty two per cent of the population, and then recommended reserving roughly half of that share. The reason was legal arithmetic, not political modesty: the courts had signalled that special provision should ordinarily stay below fifty per cent of the whole.

Reservation for the Scheduled Castes and the Scheduled Tribes already stood at twenty two and a half per cent in central services, and the Supreme Court in Balaji had indicated the same ceiling in 1963. A quota of twenty seven per cent for the backward classes brought the total to forty nine and a half per cent, inside the judicial ceiling with nothing to spare. The fifty two per cent figure therefore measured the constituency; the twenty seven per cent figure measured what the case law would bear, and the gap between the two numbers is the space in which the entire later argument about proportionality has been conducted.

Applying its indicators, the commission identified 3,743 castes and communities as socially and educationally backward, and estimated their population at about 52 per cent of the country’s total, drawn from the 1931 enumeration projected forward and from the state material. Its own tables of representation gave the policy its moral engine: the classes forming over half the population held, on the commission’s figures, under five per cent of Class I posts in the central services, and only a modestly larger share of Class II posts. The recommendations that followed, submitted with the report to the President on 31 December 1980, were headed by reservation but not exhausted by it. The commission recommended reservation of 27 per cent of posts in the services of the Union and the public sector for the backward classes, extending to promotions at all levels; reservation in public sector undertakings, banks, universities and colleges and private undertakings receiving government assistance; carry forward of unfilled reserved vacancies for three years before de-reservation; age relaxation on the pattern available to the Scheduled Castes and the Scheduled Tribes; and a roster system for the new quota on the established pattern. It also recommended that the whole scheme be given legal backing rather than left to executive instruction alone, a recommendation no government has ever taken up for the services quota, which rests on office memoranda to this day.

The other half of the report recommended a restructuring of the conditions that produced backwardness: radical land reform and redistribution of ceiling surplus land, a change in the relations of production in agriculture, dedicated financial and technical support for backward class artisans and small industry, credit on subsidised terms, scholarships and a chain of special educational facilities including residential schools and coaching for backward class students, and adult education in backward class areas. The report was also careful, in a passage its later champions and critics both under quote, to limit what reservation could be expected to do. Reservation was presented as one instrument among several, necessary because representation in the services and in education was itself a form of power, but insufficient against the structure that produced backwardness, which lay in land, credit and schooling; a quota without the economic programme would change the composition of the services faster than it changed the condition of the classes. The warning frames the report’s whole subsequent history. India implemented the instrument the commission called insufficient by itself, declined the programme it called indispensable, and then argued for thirty years about the predictable result, the concentration of the instrument’s benefits among those best placed to claim them, as if it were a discovery.

Member L. R. Naik recorded a dissent that reads, across four decades, like the first draft of the Rohini reference: he argued that the backward classes were not one bloc, that the relatively advanced among them would capture the new benefits, and that the category should be divided so that the most depressed classes received a protected share of their own. The majority did not accept internal division. The dissent was noted, the report was signed, and the argument Naik lost in 1980 was recommissioned in 2017 and remains open.

The arithmetic of fifty two per cent: how the estimate was built and why it was attacked

Because so much later argument treats the 52 per cent figure as a count, the report’s own method of arriving at it deserves a paragraph of plain description. The commission began from the census of 1931, the last enumeration to record caste, and sorted the population into groups. It took the Scheduled Castes and the Scheduled Tribes from the constitutional schedules. It took the forward Hindu castes from the 1931 returns. What remained of the Hindu population, after those two extractions, was treated as the backward classes among Hindus, and the residue came to 43.7 per cent of the total population. For the non-Hindu population, among whom no caste enumeration existed even in 1931, the commission assumed that the proportion of backward classes was of the same order as among Hindus, and so took 52 per cent of the non-Hindu share of 16.6 per cent, yielding 8.4 per cent. The two residues together gave the celebrated figure: 43.7 per cent plus 8.4 per cent, nearly 52 per cent of the country backward, against forward castes of about half that size.

The construction was vulnerable in ways its critics identified at once and its defenders have never fully answered. A residue is not a measurement: the Hindu figure was whatever was left after other categories were removed, so every error in the 1931 returns, every caste that had risen or sunk in half a century, and every community the forward list omitted, accumulated inside the backward total. The non-Hindu figure was an assumption of symmetry offered without enumeration of any kind. And the base year was 1931, chosen not because it was adequate but because it was the last of its kind, the census of 1941 having been curtailed by war and the censuses after independence having dropped the caste question for all but the scheduled groups. The commission was candid that it worked in a statistical twilight; its successor bodies have worked in the same twilight ever since, which is why the fifty two per cent has the peculiar status of a figure no one can verify and no one can replace. It is also why the identification debate keeps returning to enumeration: a list derived by residue and assumption invites the demand for a count, and the refusal to count protects the list from the count’s results. Both the demand for a caste census and the resistance to it are intelligible from the arithmetic of 1980 alone.

A second feature of the method carried its own consequences. The unit of identification was the caste, scored as a whole, so the list the commission produced was a list of communities and not of households, and a community’s score reflected its average condition across a state. The same caste could therefore be backward in one state and not in another, because the indicators measured its local position: the report’s own illustrations, repeated in every account since, are that Banias appeared on the Bihar list and not on others, and Jats were classified as backward in Rajasthan and not elsewhere at that stage. Regional relativity was built into the method, and it legitimised in advance the divergence between lists that the constitutional text would eventually protect. It also meant that the national category, Other Backward Classes, was from birth an aggregate of local judgments wearing a single label, a federation of lists rather than a list. Much of the later law, the intersection device of 1990, the advisory machinery of 1993, the presidential specification of 2018 and the state restoration of 2021, is an argument about who curates the federation.

A decade on the shelf: 1980 to 1990

The Janata government that appointed the commission had fallen before it reported, and the report reached a Congress government under Indira Gandhi that had not asked for it and did not want it. The report was laid before Parliament in 1982 with the memorandum of action the Constitution requires, and the memorandum’s substance was that the recommendations raised complex questions requiring examination. There the matter rested through the governments of Indira Gandhi and Rajiv Gandhi. The decade of non-implementation is misattributed by both sides of the later argument. Supporters of the report describe a decade of suppression, as if a settled policy had been vetoed; opponents describe a decade of rejection, as if the report had been considered and refused. The record shows something weaker and more common: a decade of deferral, in which no government spent the political capital either to implement the report or to close it, because the identification it contained was explosive in exact proportion to its electoral value. A list of 3,743 castes covering half the population is a coalition waiting to be assembled, and every party understood that the government that implemented Mandal would be rewarded by the listed and punished by the unlisted at the same election.

The deferral had an institutional form worth noticing, because it recurs. Article 340(3) required the report to be laid with a memorandum explaining the action taken, and the memorandum laid in 1982 performed the minimum the text demands: it recorded that the recommendations, including the identification of the classes and the quantum, required detailed examination in consultation with the state governments, whose own lists and quotas any central scheme would disturb. Consultation with the states then became the standing reason for inaction, available to every successor government without further invention. Meanwhile the politics the report described continued to develop underneath the deferral. State governments of several parties extended or defended their own backward classes quotas through the 1980s, backward class legislators and parties consolidated in Bihar and Uttar Pradesh, and the social constituency for implementation organised itself while the file moved between ministries. The report was not dormant in the country; it was dormant only in the secretariat, and the distance between those two conditions is what made its eventual activation so abrupt. When implementation came, it came not as the conclusion of the examination announced in 1982 but as a decision of coalition survival, taken in a week, on a text of a page and a half.

The states, during the same decade, supplied a preview of the national explosion. Gujarat, whose state list and quotas had grown under a governing strategy of assembling the backward classes, Muslims and the scheduled groups into a single electoral coalition, saw violent anti reservation agitations in 1981 and again in 1985, the second beginning in a dispute over reserved places in postgraduate medical admissions and widening into communal violence. The Gujarat episodes established the template of 1990 in advance: an incremental expansion of the reserved sphere, a student led movement of the unlisted, the conversion of an admissions question into a street question, and a government discovering that the constituency for the quota was larger but quieter than the constituency against it. They also demonstrated that none of this required the Mandal report. State lists, state quotas and state electorates were sufficient, which is a further reason the decade’s national deferral settled so little: the policy was advancing jurisdiction by jurisdiction while the Centre examined it.

What changed in 1989 was not the report but the arithmetic of power. The National Front, a coalition dependent on the very constituencies the report addressed, won the general election with implementation of the Mandal recommendations in its programme, and Vishwanath Pratap Singh became Prime Minister at the head of a minority government whose survival depended on holding together partners with opposite instincts on the question. By the summer of 1990 his government was in crisis, having dismissed the Deputy Prime Minister, Devi Lal, at the start of August. On 7 August 1990 the Prime Minister announced in Parliament that the government would implement the Mandal Commission’s recommendation on reservation in services. The announcement’s timing has been debated ever since, as an act of conviction, of coalition management, or of both; the legislative history does not need to choose, because the instrument that followed is a text and can be read. On 13 August 1990 the Department of Personnel and Training issued Office Memorandum No. 36012/31/90-Estt. (SCT), and the report of 1980 became the law of appointments in a page and a half.

The memorandum and the street: August 1990 and its aftermath

What did the office memorandum of 13 August 1990 actually order?

The memorandum reserved twenty seven per cent of the vacancies in civil posts and services under the Government of India, to be filled by direct recruitment, for the socially and educationally backward classes. Parallel instructions were to follow for public sector undertakings, financial institutions and public sector banks, extending the same proportion across the public employment the Centre controlled.

Its most consequential sentence was the quietest. The backward classes for the purpose of the reservation would comprise, in the first phase, the castes and communities common to both the Mandal Commission’s list and the lists maintained by the state governments. Faced with 3,743 names drawn from data half a century old, the executive declined to adopt the commission’s list outright and instead took the intersection of two lists, a bureaucratic compromise that answered the identification question by averaging its previous answers. The memorandum said nothing about the creamy layer, nothing about promotions, and nothing about education. It was a services order, and its narrowness was part of what the litigation then tested.

The words “in the first phase” carried a promise the order never kept, and the unkept promise is a small perfect specimen of the identification problem. A first phase implies a second, in which the executive would move from the intersection to the commission’s list itself, adopting the 3,743 names with whatever corrections a decade of change required. No second phase was ever announced. The interim list, assembled for speed in August 1990, was overtaken by the litigation, and when implementation resumed after the judgment it was the common list, corrected for the creamy layer and formally notified as the central list in 1993, that hardened into permanence and began its long career of entry by entry revision. The device deserves its place in the history because governments reach for it whenever the full identification is too dangerous to perform and too valuable to abandon: adopt the overlap, defer the residue, and let the deferral become the settlement. The Mandal list in its complete form was therefore never the operative list of the Indian state at any moment. What operated was always a negotiated descendant of it, filtered through the state lists in 1990, through the creamy layer in 1993, and through the advisory machinery thereafter. Polemic on both sides speaks of the Mandal list as a thing imposed; the record shows a thing proposed, intersected, pruned and administered, which is a less dramatic and more accurate description, and it is the description that explains why the identification question survived its supposed settlement: the settlement was of a list that had already been replaced by its own first phase.

The response divided the country in public. Through August and September 1990, students and job aspirants in the northern cities protested against the order in a movement that turned on the fear that open competition for scarce government posts, the one ladder the unlisted middle class trusted, had been shortened by more than a quarter at a stroke. The protests produced the defining images of the episode, including the self-immolation attempt by the Delhi University student Rajiv Goswami on 19 September 1990 and a series of similar acts by other young men, some fatal. Pro-reservation rallies mobilised on the other side with equal conviction that a promise made in 1950 and documented in 1980 was being honoured at last. The government fell in November 1990, in a confidence vote whose causes were wider than reservation but in which the Mandal decision was the central charge. By then the question had already moved to the forum that would decide it. A writ petition filed on behalf of the Supreme Court Bar Association challenged the memorandum, a bench of five judges stayed its operation on 1 October 1990 pending final disposal, and the executive order spent the next two years in suspension while a bench of nine judges heard the case that the whole country understood would settle the terms on which backward classes reservation could exist.

The politics of that autumn rearranged more than a quota. The announcement of August 1990 was followed in September by the launch of the temple movement’s Rath Yatra, and the two mobilisations, one for reservation and one for the temple, competed for the same northern electorate through the same weeks; contemporaries compressed the contest into a slogan pairing the commission’s name against the sacred pot carried in procession. When the yatra’s leader was arrested in Bihar in October 1990, the Bharatiya Janata Party withdrew the support on which the National Front government depended, and the government fell the following month. Its successor, a minority administration under Chandra Shekhar sustained by Congress support, neither withdrew the memorandum nor pressed it, leaving the stayed order in the keeping of the Court. The episode fixed the political grammar of the subject for a generation. Implementation had demonstrated that the listed half of the population could be mobilised as a constituency by a single executive act, and no party thereafter treated the backward classes as a welfare category alone; they were, permanently, the largest single object of electoral organisation in the republic, and the parties that grew from the listed communities, in Bihar, in Uttar Pradesh and elsewhere, carried the Mandal inheritance as their founding asset. The litigation, in other words, was deciding the validity of an order whose political consequences had already become irreversible before the Court spoke.

The modification of 1991: the economic criterion enters, and is removed

The Congress government of P. V. Narasimha Rao, which inherited the stayed memorandum, chose neither to withdraw it nor simply to defend it. By a further office memorandum dated 25 September 1991 it amended the order of 1990 in two respects. The first introduced an economic gradient inside the quota: within the 27 per cent reserved for the backward classes, preference was to be given to the poorer sections among them. The second stepped outside the backward classes framework altogether and reserved a further 10 per cent of vacancies for other economically backward sections of the people who were not covered by any of the existing schemes of reservation, a category defined purely by economic weakness and open in practice to the poor of the forward castes. The political logic was conciliatory: soften the order’s caste edge from inside, and answer the agitation’s central complaint, that poverty among the unlisted counted for nothing, by giving it a quota of its own. The constitutional logic was hazardous, because Article 16(4) speaks of backward classes, and a class defined by income alone was a different juridical object from the classes the Mandal exercise had identified.

Both halves of the 1991 memorandum were tested in the judgment, and they fared differently in a way that shaped the next thirty years. The Supreme Court, in the majority view delivered in Indra Sawhney, held that the distinction between poorer sections and others among the backward classes was not invalid if it was understood and operated as a classification by relative backwardness within the identified classes; an internal preference of that kind was a refinement of the list, not a departure from it. The separate 10 per cent, by contrast, was struck down as constitutionally invalid, on the ground that a reservation resting on economic criteria alone, for persons expressly defined by their exclusion from every backward class, could not be sustained under Article 16 as it then stood. The episode matters to the legislative history because of what happened to the idea later. The proposition that economic weakness alone could ground a quota lost in 1992 as an executive act under an unamended text, and returned a generation later through the route the judgment had left open, constitutional amendment, in the form examined in this series’ article on the 103rd Amendment and EWS reservation. Executive memorandum, judicial refusal, constitutional amendment: the sequence is the standard life cycle of a reservation idea in India, and 1991 is its clearest early instance.

The half of the 1991 memorandum that survived deserves its own notice, because it is the ancestor of the sub-categorisation debate inside government. The preference for the poorer sections among the backward classes was an instruction to differentiate within the quota by relative backwardness, issued by executive order, a generation before a commission was asked to design the differentiation scientifically. The Court’s treatment of it was indulgent but conditional: the preference stood if it operated as a classification by degrees of backwardness among the identified classes, and fell if it operated as a bare income test. That is, in embryo, the whole law of internal distribution in this field. Differentiation inside the category is permissible, indeed the judgment’s creamy layer is itself a differentiation at the category’s upper edge; what is not permissible is to let a single economic marker replace the composite judgment of backwardness. The Rohini reference, the Davinder Singh reasoning and the Bihar annexures all work variations on the distinction the Court drew in passing in 1992 between refining the list from inside and abandoning it for a simpler test.

The settlement of 1992, at the level a history needs

On 16 November 1992 a bench of nine judges of the Supreme Court delivered judgment in Indra Sawhney v. Union of India, by a majority of six to three, with the principal majority opinion written by B. P. Jeevan Reddy J. The doctrine of that judgment, the fifty per cent ceiling, the creamy layer, the bar on reservation in promotion and the account of how a caste can be a class, is the property of the article on Indra Sawhney and the fifty per cent ceiling, and this history states only the holdings a narrative cannot proceed without. The Court upheld the office memorandum of 13 August 1990 and the 27 per cent reservation it created, holding it valid and enforceable. It held that a backward class could be identified with caste as a starting point among other criteria, that the socially advanced sections within a backward class, the creamy layer, must be excluded from the benefit, and that reservation should ordinarily not exceed fifty per cent, departure being confined to exceptional circumstances. It struck down the 10 per cent economic quota of 1991, held that Article 16(4) does not permit reservation in promotion, and directed that a permanent body be created to examine requests for inclusion in and complaints about the backward classes lists, advice which produced the National Commission for Backward Classes Act, 1993.

For the legislative history, three features of the settlement deserve emphasis over the doctrine. The first is that the Court declined to re-try the Mandal Commission’s identification exercise. It recorded that it was not necessary to express any opinion on the correctness and adequacy of the exercise the commission had done, which meant the list survived not because it was vindicated but because it was not the question. The second is that the Court converted the identification question from a one time event into a continuing institution: the permanent body direction is the moment the list acquires a standing custodian, and every later development, the statutory commission of 1993, the constitutional commission of 2018, the sub-categorisation reference of 2017, is a variation on that direction. Several of the opinions in the case contemplated, in terms, that identification was not a settlement for all time: the lists were to be revisited periodically and revised as the conditions of the listed classes changed, entries being removed as classes advanced and added as evidence warranted, so that the category would behave as a measurement rather than as an inheritance. That expectation has been honoured unevenly. Entries have been added in number; removals of communities that have advanced have been rare to the point of absence, because an exit from the list is a political injury no government volunteers to inflict and no commission finds it prudent to recommend against a mobilised community. The asymmetry, easy entry and near impossible exit, is one of the quiet engines of the list’s growth, and it converts the periodic revision the Court imagined into a ratchet. A reader assessing any proposal for reform should ask which side of the ratchet it engages: proposals that add, subdivide or re-measure prosper in the literature; proposals that subtract wait for a constituency that has not yet appeared. The third is that implementation, stayed since October 1990, had to be rebuilt to the judgment’s specification. The government appointed an expert committee under Justice Ram Nandan Prasad to recommend the criteria for excluding the socially advanced, and by the office memorandum of 8 September 1993 adopted the creamy layer rules, status based categories covering holders of constitutional posts and senior officers, professional and property categories, and an income and wealth test set at one lakh rupees a year, with income from salary and from agricultural land left out of the computation. Appointments under the 27 per cent quota began on that corrected footing. The reservation had been announced in 1990; it operated from 1993.

The standing machinery the judgment directed into existence then took the identification question out of the courts and the manifestos and gave it a permanent address. The National Commission for Backward Classes Act, 1993 created the commission as a statutory body, and it was constituted in August 1993 with a chairperson who is or has been a judge of the Supreme Court or of a High Court, and members including a social scientist and members with special knowledge of the backward classes. Its statutory function was the list: to examine requests for inclusion of any class of citizens as a backward class in the central list, to hear complaints of over inclusion and under inclusion, and to tender advice to the Union government, which the Act expected to be ordinarily binding, departing governments being required to place their reasons before Parliament. Two quiet revolutions were folded into that dry scheme. The first is that identification became continuous. Between the bursts of 1955, 1980 and 1990 there had been no institution whose daily business was the boundary of the category; after 1993 there was, and entries were added, corrected and occasionally removed by a routine of hearings and advice rather than by commission and crisis. The second is that the boundary became litigable entry by entry. A caste association that failed before the commission could go to court; a government that rejected advice had to explain itself; the list acquired a jurisprudence of its own, of which the largest single episode, the Jat notification of 2014 and its quashing, is told below. The burst pattern of the commission era did not end in 1993, but it acquired a continuous undertow, and the later history alternates between the two registers: routine revision by the standing body, and convulsion when a question too large for the routine, sub-categorisation, or state power, or a whole community’s status, breaks the surface.

The promotions arc: four amendments on the other lists, none on this one

Why is there no reservation in promotion for the backward classes?

The Mandal Commission recommended 27 per cent reservation in promotions at all levels, and the recommendation has never been implemented, because the judgment of 1992 held that Article 16(4) authorises reservation only at the stage of appointment. Parliament amended the Constitution four times to restore promotion reservation, each time for the scheduled groups alone.

For the backward classes no government has introduced an equivalent amendment, so the promotion recommendation of 1980 remains the largest single element of the Mandal programme that was refused and never revived, a refusal that has held across governments of every composition since.

The four amendments belong in this history because they show what the system does when a judicial limit is one the political process refuses to accept, and because their absence on the backward classes side is the control case. The Constitution (Seventy-seventh Amendment) Act, 1995 inserted Article 16(4A), empowering reservation in promotion for the Scheduled Castes and the Scheduled Tribes. The Constitution (Eighty-first Amendment) Act, 2000 inserted Article 16(4B), allowing unfilled reserved vacancies to be carried forward as a separate class of backlog vacancies outside the fifty per cent ceiling of the year. The Constitution (Eighty-second Amendment) Act, 2000 added a proviso to Article 335 permitting relaxation in qualifying marks and lowering of standards of evaluation for the scheduled groups in promotion. The Constitution (Eighty-fifth Amendment) Act, 2001 wrote consequential seniority into Article 16(4A) with retrospective effect from 17 June 1995. The amendments were tested and upheld subject to conditions in M. Nagaraj v. Union of India in 2006, and the conditions were themselves adjusted in Jarnail Singh v. Lachhmi Narain Gupta in 2018. Two generations of constitutional labour were thus spent securing for the scheduled groups, in promotion, a benefit that the Mandal Commission had recommended for the backward classes in 1980 and that no amendment has ever proposed for them. The asymmetry is not an oversight. Promotion reservation for the backward classes lacks the two things the scheduled groups’ amendments had: a constituency organised around the demand, and a judicial invitation, however grudging, to legislate. The quantum on the backward classes side stayed at the appointment stage because 1992 froze it there and the amendment route was never taken.

The creamy layer in operation: the one instrument that is revised on a schedule

Of all the devices in this history, only one is revisited as a matter of routine, and it is the one the Court invented. The creamy layer exclusion, once the government adopted the expert committee’s scheme in the memorandum of 8 September 1993, became a standing instrument of administration with three moving parts. The first part is status: the children of persons holding constitutional posts, of officers of the All India, Central and State services in the senior groups, and of officers of equivalent rank in public sector undertakings, are excluded without any inquiry into income, on the ground that the household’s social capital is itself the disqualification. The second part is occupation and property: the children of doctors, lawyers, engineers, chartered accountants and comparable professionals, of persons engaged in trade and industry, and of owners of agricultural holdings above defined limits, are excluded where the family’s income crosses the prescribed line. The third part is the income and wealth test itself, which catches the residue: gross annual income from all sources other than salary and agricultural land, the two heads the memorandum deliberately leaves out of the computation. The ceiling under that test has been revised at lengthening intervals as prices moved: one lakh rupees in 1993, two and a half lakh in 2004, four and a half lakh in 2008, six lakh in 2013, and eight lakh rupees a year by the revision of September 2017.

The instrument matters to the identification question because it is identification continued by other means. The list says which communities are in; the creamy layer rules say which households within those communities are out, and the second determination is made afresh for every candidate, on certificates and income computations administered by local authorities, in a volume of individual decisions that dwarfs the periodic revision of the list itself. Its controversies are correspondingly granular. The exclusion of salary from the income test is defended as necessary to avoid penalising the salaried lower middle class whose parents’ generation the policy lifted, and attacked as a channel through which the children of senior salaried households, excluded by the status limb in the services but harder to classify outside them, retain eligibility. The long gaps between revisions are defended as stability and attacked as erosion, since a ceiling fixed in nominal rupees tightens the exclusion every year prices rise, quietly shrinking the eligible pool without any government having to decide that it should shrink. And the status categories, drawn for the services of the 1990s, fit awkwardly over an economy of private employment, professional partnerships and agricultural wealth converted into urban assets. The point for this history is structural. A policy whose contested boundary is administered household by household, on a test revised five times in thirty years, does not stand still even when its headline percentage does: the effective size of the beneficiary class has been continuously renegotiated inside the creamy layer rules while the argument about the percentage proceeded as if nothing had moved.

Reservation reaches the classroom: the amendment of 2005, the Act of 2006 and the judgment of 2008

For its first fifteen years, central backward classes reservation was a services policy only, because the 1990 memorandum was a services instrument and state level educational quotas operated under state lists. The extension to central educational institutions required a constitutional step first, since the admission of students to private institutions raised Article 19(1)(g) questions that Article 15(4) did not answer in terms. The amendment of 2005, the Constitution (Ninety-third Amendment) Act, inserted Article 15(5) with effect from 20 January 2006, empowering the state to make special provision by law for the admission of socially and educationally backward classes, the Scheduled Castes and the Scheduled Tribes to educational institutions including private institutions, aided or unaided, other than the minority institutions protected by Article 30(1). The pattern, a contested expansion of the reservation power carried by constitutional text rather than left to executive instruction, followed the route this series traces in the 86th Amendment and the right to education, where an entitlement disputed under the existing text was settled by amending the text. Parliament then used the new power for the central institutions. The Central Educational Institutions (Reservation in Admission) Act, 2006, which is Act 5 of 2007, provided in its Section 3 for reservation of fifteen per cent of seats for the Scheduled Castes, seven and a half per cent for the Scheduled Tribes and twenty seven per cent for the backward classes in the institutions established, maintained or aided by the Central Government, the universities, institutions of national importance and deemed universities inside the University Grants Commission Act, 1956 framework. Section 5 required a mandatory increase in seats, so that the number of unreserved seats would not fall in absolute terms as the quota was introduced, and the Act’s scheme contemplated admissions under the new reservation beginning with the academic session of 2008, phased with the expansion. The Act’s own limits are as instructive as its coverage. It bound only the central institutions; the far larger universe of state universities and colleges remained under state law and state lists, so the education settlement of 2006 to 2008, like the services settlement of 1990, was a Union settlement laid over divergent state practice. It did not extend to the minority institutions the Constitution protects, which Article 15(5) itself had carved out. And it left the unaided private sector, where the growth in professional seats was concentrated, to whatever the states chose to enact under the new Article, a patchwork that made the amendment’s promise in private institutions more cited than operative. The pattern is the field’s usual one: a constitutional power taken at full width, a statute enacted at half width, and the residue left to jurisdictions whose answers differ. An Oversight Committee chaired by Veerappa Moily worked out the expansion’s scale: to introduce a 27 per cent quota while holding the absolute number of unreserved seats constant, the institutions’ total intake had to rise by roughly fifty four per cent, an increase the committee recommended be phased over three years with matching investment in faculty and infrastructure. The arithmetic explains the politics of the education round. A services quota reallocates future vacancies invisibly; a seats quota in a small set of famous institutions reallocates a visible, counted good, and the seat expansion was the government’s answer to the charge that one group’s gain had been priced in another group’s places.

The announcement of educational reservation in 2006 produced protests by medical students and resident doctors that consciously echoed 1990, and the statute was challenged on the identification ground that has organised this whole history: the petitioners argued that the lists rested on the census of 1931 and that a quota built on seventy five year old enumeration could not stand. The litigation repeated the rhythm of 1990 in miniature: by an interim order of 29 March 2007 the Supreme Court stayed the operation of the 27 per cent in admissions for that academic year and observed that the 1931 census could not be a determinative factor for identification, so the first cohort the Act was meant to serve waited while the Constitution Bench heard the challenge, and the reservation operated only after the final judgment cleared it. In Ashoka Kumar Thakur v. Union of India, reported at (2008) 6 SCC 1 and decided on 10 April 2008, a Constitution Bench of five judges under K. G. Balakrishnan C.J. upheld the Ninety-third Amendment and the Act of 2006, and with them the 27 per cent. The Court held that the creamy layer must be excluded from the educational quota as it had been from the services quota, that the 1931 census could not be treated as a determinative basis for identifying the backward classes and that better data should be gathered, and members of the bench pressed for a periodic review of whether the reservation continued to be justified, a review after five or ten years being variously proposed. The judgment closed the second great litigation of the subject. After 2008, the quantum for backward classes stood judicially settled in both fields the Constitution touches, services and admissions, at the same figure the commission had proposed in 1980; and the identification question, on which the Court had again declined to substitute its own answer, continued underneath, awaiting better data that the state had not produced and a list machinery that was about to be rebuilt twice in three years.

The list in operation: the Jat notification and the limits of executive identification

The standing machinery met its defining test in 2014. The Jat community, a landowning and politically organised community of the northern states, stood in several state lists but not in the central list for most of those states, and its claim to central inclusion was examined by the National Commission for Backward Classes through public hearings. By its advice of 26 February 2014 the commission held that the community had not fulfilled the criteria: it was not socially backward, not educationally backward, and adequately represented in the armed forces, government services and educational institutions. The Union Cabinet, meeting on 2 March 2014, resolved that the advice did not adequately take into account the ground realities, declined to accept it, and directed inclusion; by a notification dated 4 March 2014 the Jats were entered in the central list for nine states and parts of Rajasthan. The sequence was the 1993 scheme running in reverse at full speed: the expert body examined and refused, the executive overruled within a week, and the entry was made by the same instrument, a notification, as any spelling correction.

On 17 March 2015, in Ram Singh v. Union of India, a bench of Ranjan Gogoi and Rohinton Fali Nariman JJ. set aside and quashed the notification. The judgment is the fullest modern statement of what identification requires, and its reasoning repays attention in a legislative history because it is addressed to governments rather than to litigants. The Court held that the commission’s view was adequately supported by good and acceptable reasons, that the cabinet’s stated ground did not answer it, and that inclusion of a politically organised class mainly on the basis that other groups who had fared better stood included could not be affirmed. It warned that backwardness had to be a manifestation of several independent circumstances, social, cultural, economic, educational or political, that new yardsticks had to be evolved moving away from a caste centric determination, and that possible wrong inclusions in the past could not be the basis for fresh inclusions, the gates being open only to the most distressed. The case completed the education of the executive that 1992 had begun. A government may reject its commission’s advice, but it must answer the advice with reasons that engage the criteria, and a notification that treats the list as an electoral instrument invites annulment. Identification, the Court was saying, is an evidentiary act performed by the state under discipline, and the discipline held even where the beneficiary community was powerful, the government of the day and its successor both defended the entry, and the political cost of refusal was plain. Episodes of this kind are the routine register of the list in operation, and they are why the larger convulsions, when they come, are fought over who holds the pen rather than over what the pen writes in any single case.

The mechanics beneath the episode deserve a sentence of their own, because the list’s physical form explains its politics. The central list is not a single schedule but a series of notifications, one resolution for each state and Union territory, in which entries are recorded with their synonyms, sub caste names and, frequently, area restrictions confining an entry to named districts where the community’s backwardness was found. A claim for inclusion therefore arrives as a dossier: the community’s case, the state government’s recommendation where it supports the claim, the commission’s public hearings in the state, its findings against the criteria, and its advice. Against that dossier, a notification issued in a week, on a cabinet minute, without the dossier’s discipline, stands out on the record exactly as it stood out to the Court in Ram Singh. The list is slow by design, and its slowness is the safeguard; every attempt to make it fast, in 1990 by intersection, in 2014 by override, has been answered by a court insisting that identification is evidence work. The commissions change, the articles change, the insistence does not. It is the nearest thing the field possesses to a settled rule of process, and it was assembled, like everything else in this history, not by design in a single instrument but by accumulation, each institution adding its layer of discipline after an attempt to proceed without it.

Who keeps the list: the amendment of 2018, the Maratha judgment and the restoration of 2021

The permanent body the judgment of 1992 had demanded was created by the National Commission for Backward Classes Act, 1993 as a statutory commission, charged with examining requests for inclusion in the central list and complaints of over inclusion and under inclusion, and advising the government, whose practice was ordinarily to accept the advice. For twenty five years the identification machinery at the Centre therefore ran on a statute, while the commissions for the Scheduled Castes and the Scheduled Tribes sat in the Constitution itself. The Constitution (One Hundred and Second Amendment) Act, 2018 closed that gap and, in doing so, redrew the map. It inserted Article 338B, giving the National Commission for Backward Classes constitutional status, composition and powers modelled on the older commissions, and it inserted Article 342A, under which the President may, after consultation with the Governor, specify the socially and educationally backward classes in relation to a state or Union territory, with inclusion in and exclusion from that central specification thereafter reserved to Parliament by law. The amendment was presented as a standardisation, putting backward classes specification on the same presidential and parliamentary footing as Articles 341 and 342. It was also capable of a larger reading, and it received one.

The redesign had a second, quieter effect that the Maratha argument largely overshadowed. The statutory commission of 1993 had been, in essence, a list authority with advisory teeth: its business was inclusion and exclusion, and its advice was ordinarily binding. The constitutional commission of 2018 was modelled instead on the commissions for the Scheduled Castes and the Scheduled Tribes, whose business is safeguards: investigating whether the protections for the class are working, inquiring into specific complaints of deprivation, monitoring the class’s socio economic development and reporting to the President, with the Union and every state government obliged to consult it on all major policy matters affecting the backward classes. Identification, meanwhile, was moved by Article 342A to the President’s specification and Parliament’s amendment, where the commission’s role is consultative rather than advisory in the old sense. Read as institutional design, 2018 thus promoted the commission and demoted the list function in the same stroke: the backward classes acquired a constitutional watchdog of the familiar kind, and lost the dedicated boundary keeper the judgment of 1992 had imagined. Whether that trade was wise is contested by the participants themselves, the commission’s defenders valuing its constitutional rank and complaint jurisdiction, its critics observing that the one function no other body performs, the continuous, evidence based revision of the category’s boundary, was the function the redesign dispersed. The restoration of 2021 repaired the states’ position; it did not revisit this one. During the amendment’s passage through Parliament, the government of the day gave assurances that the measure standardised the Centre’s own arrangements and did not disturb the power of the states to identify backward classes for their own purposes, and state governments proceeded for three years on that understanding, maintaining and amending their lists without challenge. The assurances were later placed before the Constitution Bench in the Maratha litigation, which held that the enacted text, and not the statements made in its support, determined the power’s location: a lesson, transferable across this whole field, that in reservation law the operative fact is always the instrument, the article, the memorandum, the notification, and never the speech that accompanied it.

In Jaishri Laxmanrao Patil v. Chief Minister of Maharashtra, the Maratha case, decided on 5 May 2021, a Constitution Bench held by a majority of three to two that the amendment had effected a transfer: after Article 342A, the power to identify socially and educationally backward classes lay with the President alone, and the states had lost the power to maintain lists of their own for the purposes of the Constitution. The Maharashtra thread that carried the question to the Court had its own instructive shape. The Maratha community, dominant in the state’s politics and landholding, had pressed for backward class status for years; the state’s backward classes commission under Justice M. G. Gaikwad reported in favour of treating the community as socially and educationally backward, and the Maharashtra legislature enacted the Socially and Educationally Backward Classes Act, 2018, reserving sixteen per cent of seats and posts for the new category, taking the state’s total reservation well beyond the fifty per cent line. The Bombay High Court sustained the Act with a reduced quantum. The Constitution Bench struck the Act down, holding both that the identification power had not survived the amendment of 2018 in the state’s hands and that no exceptional circumstances justified the breach of the ceiling on the material presented. The judgment also declined, unanimously on the point, to refer the fifty per cent ceiling to a larger bench. The identification holding lasted one season. Parliament answered within the year with the Constitution (One Hundred and Fifth Amendment) Act, 2021, which inserted clause (3) into Article 342A to provide that every state or Union territory may, by law, prepare and maintain for its own purposes a list of socially and educationally backward classes whose entries may differ from the central list, and made corresponding changes to Article 338B and to the definition of socially and educationally backward classes in Article 366(26C). The dual structure that had grown up by practice, Centre list and state lists diverging for the same community, was written into the constitutional text expressly, an unusual instance of an amendment whose whole purpose was to reverse a reading of the amendment before it. The sequence completes the pattern this article has traced since Kalelkar: the locus of identification is itself never settled for long, because each settlement reallocates a power that governments experience as electoral property.

One further judgment belongs to the sequence, though its list is a different one. In State of Punjab v. Davinder Singh, decided on 1 August 2024, a bench of seven judges held by six to one, Bela M. Trivedi J. dissenting, that a state may sub classify the castes within its Scheduled Castes list for the purpose of reservation, distributing the reserved share among groups within the list on empirical evidence of differing backwardness and representation, overruling the contrary position taken in E. V. Chinnaiah v. State of Andhra Pradesh. Four of the seven judges also expressed, in separate opinions, the view that the creamy layer principle should in time be applied within the Scheduled Castes and the Scheduled Tribes as well, a view stated as a direction for future policy rather than as a condition of the sub classification upheld. The judgment concerned the scheduled lists and is recorded here for its structural significance: the highest court has accepted, in the neighbouring field, the two propositions the backward classes debate had circled since L. R. Naik’s dissent, that a listed category is not internally homogeneous and that distribution inside the list is a legitimate object of policy. What the courts have accepted for one list, governments will be asked to attempt for the other, and the instrument waiting for that attempt is the report described in the next section.

Sub-categorisation: the Rohini Commission and the report that has not been published

On 2 October 2017 the President appointed a commission under Article 340, chaired by Justice G. Rohini, retired Chief Justice of the Delhi High Court, to examine the sub-categorisation of the Other Backward Classes. Its terms of reference, as they settled, were to examine the extent of inequitable distribution of the benefits of reservation among the castes and communities in the central list, to work out the mechanism, criteria, norms and parameters for sub-categorisation within the category in a scientific approach, to classify the entries in the central list into sub categories, and to study the list’s own entries and recommend corrections of repetitions, ambiguities, inconsistencies and errors of spelling or transcription. The commission was given twelve weeks. It received thirteen extensions, a fact that is itself evidence about the reference: the central list contains over 2,600 entries, many of them synonyms, sub castes and area specific entries accumulated from the intersection exercise of 1990 and three decades of additions, and the distribution of benefits among them could not be examined without caste wise data on admissions and appointments that the state’s own systems did not routinely keep. The commission’s method, so far as its public statements disclosed it, was to call for the records of central ministries, public sector undertakings, universities and recruiting bodies and to reconstruct, caste by caste, who had actually received the reserved posts and seats across a run of years. That reconstruction, entry by entry for a list of thousands, is the unglamorous core of every identification exercise since Kalelkar, and its difficulty is the practical reason the question is reopened by commissions and then left by governments: the evidentiary base does not exist in usable form, building it takes years, and its results, when built, will offend whichever communities the numbers show to have drawn ahead. A government that commissions the inquiry purchases the option of reform without committing to its price, which is why the reference of 2017, the extensions that followed and the silence after submission form a single intelligible sequence rather than three separate mysteries.

The commission submitted its report to the President on 31 July 2023. Its contents have not been made public, and no action on it has been announced. The non publication has a constitutional dimension that distinguishes it from an ordinary delayed report. Article 340(3) contemplates a sequence after submission: the report is to be laid before each House of Parliament with a memorandum explaining the action taken on its recommendations. For the Kalelkar report the sequence was completed and the action recorded was inaction defended; for the Mandal report it was completed in 1982 and the memorandum’s deferral became the policy of a decade. For the Rohini report the sequence has not been completed in public at all, so the identification question stands referred to a body whose answer exists, is official, and is unavailable to the Parliament the procedure was designed to inform, a posture without precedent in the seventy year working of the article. That is where the record ends, and any account that describes the sub-categorisation of the central list as implemented, or that summarises the Rohini recommendations, is describing a document it cannot have read. The position is worth stating flatly because the report’s invisibility is the fact: the most extensive official examination of the identification question since Mandal has been completed and withheld, so that the debate it was commissioned to inform proceeds without its evidence. The states have not waited. Bihar’s annexure system, which since 1978 has separated the most backward from the backward inside the state quota, Andhra Pradesh’s classification of its backward classes into lettered groups with distinct shares, and similar internal orderings in other states show sub-categorisation operating as state practice while the central list remains formally undifferentiated, and the Davinder Singh reasoning has lowered the constitutional risk of the experiment. There is an asymmetry here that the debate often misses. Sub-categorisation at the state level divides a state quota among communities whose local standing the state can plausibly measure; sub-categorisation of the central list must divide a Union quota among over 2,600 entries whose standing was measured, if at all, state by state and decade by decade. The Rohini exercise was therefore not the state practice enlarged but a different order of undertaking, the construction of a national internal distribution from local materials, and the years of extensions are the measure of the gap between the idea’s simplicity and its evidentiary price. The list question, having survived a commission that disowned its answer, a commission whose answer waited a decade, and two amendments that fought over who answers it, is at present lodged with a commission whose answer no one outside government has seen.

The data question: 1931, the census that did not publish, and Bihar

Underneath every stage of this history sits an embarrassment of evidence. Caste entered the Indian census with the first synchronous enumerations of the nineteenth century, and the colonial census did not merely count castes; in 1901, under H. H. Risley, it attempted to arrange them by social precedence, and the attempt detonated the politics the modern debate inherits. Communities organised themselves into caste associations, filed petitions and published pamphlets to claim a higher rank than the enumerators proposed, discovering in the process that a state classification is a prize worth organising for. Enumeration was progressively curtailed thereafter, and the census of 1931 was the last to publish a full caste table, the exercise of 1941 being truncated by the war. The Republic, for reasons its framers defended as nation building, declined to resume the count for any group outside the constitutional schedules. The Mandal Commission’s dependence on 1931 was therefore not an accident of laziness but the designed condition of post independence statistics: the state that distributed by caste had resolved, as a matter of policy, not to count by caste, and it held both positions simultaneously for seventy years. The last census to enumerate caste across India was that of 1931. The Mandal Commission’s fifty two per cent was an extrapolation from that enumeration through the state lists, and the identification of 3,743 castes rested substantially on material whose demographic base was two generations old when the report was written. The courts have noticed: the insistence on quantifiable data runs from Indra Sawhney through M. Nagaraj to the Maratha judgment, and in Ashoka Kumar Thakur the bench held expressly that the 1931 census could not be a determinative factor and urged the gathering of better data. The demand for a caste census is the political form of that judicial insistence, and it is argued as a data question before it is argued as a policy question. Its proponents contend that no rational distribution, of the quantum, within the list, or among sub categories, is possible while the state refuses to count the units it distributes among, and that enumeration is a precondition of honest policy in whichever direction the policy then moves. Its opponents contend that enumeration by the state hardens the identities it records, converts a administrative category into a permanent political claim, and invites competitive mobilisation in which every community’s rational strategy is to be counted as more backward and more numerous than its neighbour. Both arguments are about what counting does, and they are separable from the argument about reservation itself: a reader can favour enumeration and oppose quotas, or the reverse.

The state’s practice has been to gather and not release. The Socio Economic and Caste Census of 2011 collected caste returns across the country as a separate exercise from the population census, and the caste returns were not published, the government stating that the data contained errors on a scale that made release unwise. The evidentiary vacuum was filled, in the end, by a state. Bihar conducted its own caste based survey and released the findings on 2 October 2023: the Extremely Backward Classes formed 36.01 per cent of the state’s population of a little over thirteen crore, the Other Backward Classes 27.13 per cent, the two together 63 per cent, against 15.52 per cent for the unreserved categories. Bihar then raised its total reservation to 65 per cent by legislation in 2023, and the Patna High Court struck the increase down in 2024, holding that it lacked the exceptional justification the ceiling doctrine requires. The survey and its aftermath demonstrated both sides of the enumeration argument in a single state within a year: the data made the distributional debate concrete for the first time in ninety years, and it immediately became the premise of a quota increase that the courts then contained. At the Centre, the Cabinet Committee on Political Affairs decided on 30 April 2025 that caste enumeration would be included in the forthcoming census. The form of the decision mattered as much as its content. Caste is to be counted inside the population census itself, conducted under the Census Act by the Registrar General’s machinery, and not through a separate socio economic exercise of the 2011 kind run outside it; the distinction answers the fate of the earlier returns, which were gathered by a different instrument, under different auspices, and were then disowned for their errors. A census count is designed for publication, and its results will enter the record with the authority of the decennial enumeration that has carried every other distributional fact in Indian public life. What follows from the count, for the quantum, for the lists and for sub-categorisation, the decision does not say, and no rule of law converts a population share into a quota share automatically; the fifty per cent ceiling, the creamy layer and the list machinery will all have to be re-argued against numbers that, for the first time since 1931, both sides will have to accept as the count of the country. For researchers and journalists who work on this record, where a leaked list, an unpublished report or an interview with a commission member can carry legal and political consequence, the handling of the underlying material is part of the craft, and the discipline described in the account of how journalists use VaultBook to protect sources applies with equal force to anyone assembling a private research file on a subject where the official file is closed.

The federal life of the list: one category, many lists

The identification question has a geography, and the geography explains several episodes that a purely doctrinal account leaves opaque. Backwardness, on the Mandal method itself, is a local measurement: a community’s position in the hierarchy, its schooling and its representation are facts about a state, sometimes about a district, and the same community can stand high in one state’s order and low in another’s. A list is therefore naturally a state document, and the Indian system has produced lists at every level the Constitution permits: state lists for state services and institutions, a central list for Union services and central institutions, and, after 2021, an express constitutional statement that the two may differ. The working consequences are met by ordinary families rather than by courts. Eligibility does not travel with the person. A candidate whose caste is listed in the state of the family’s origin but not in the state where a post is advertised has no claim there; a certificate issued against one list answers nothing about another; and the central list itself is organised state by state, so that the same person may be entitled against a Union vacancy in relation to one state and not in relation to another. The certificate regime that administers this geography, caste certificates issued by revenue authorities, verified by scrutiny committees on the pattern the Supreme Court systematised for caste certificates in Kumari Madhuri Patil in 1994, generates its own litigation of false claims, migrated families and synonym disputes, a permanent low level jurisprudence of the list’s edges.

The extreme case of the federal pattern is Tamil Nadu, which maintains a total reservation of sixty nine per cent, divided among backward classes, most backward classes, the Scheduled Castes and the Scheduled Tribes, under a state Act of 1993 that Parliament placed in the Ninth Schedule by the Seventy-sixth Amendment in 1994. The Tamil Nadu settlement is the standing refutation of any claim that the fifty per cent ceiling describes Indian practice; it describes the doctrine the courts apply to new schemes, while the most populous southern state has operated above it for three decades under a constitutional shelter whose strength the Supreme Court qualified, without dismantling, when it held in I. R. Coelho in 2007 that laws placed in the Ninth Schedule remain open to challenge against the basic structure. The Maratha litigation can be read as the ceiling doctrine refusing Maharashtra what the Ninth Schedule route had secured for Tamil Nadu a generation earlier, and the contrast between the two states’ outcomes is the clearest demonstration in the field that the law of reservation is not one law but a set of settlements of different vintages, held together by a doctrine none of them fully obeys. For the legislative historian, that is the point. The quantum was settled nationally in 1992 only for schemes that came after; the map it governs is an archipelago of state settlements, central specifications and protected survivors, and the list, in each jurisdiction, is the local constitution of the category.

The two cases, stated as their proponents state them

The case for reservation for backward classes, at its strongest, begins from the commission’s own tables and never leaves the record. A set of communities forming roughly half the population held a small fraction of the senior posts of the state that governed them, and the disparity was not the residue of one generation’s choices but the product of a hierarchy that had assigned occupation, education and standing by birth for centuries. Formal equality of competition, on this view, does not correct such a structure; it ratifies it, because the examination hall treats as individual merit the accumulated advantages of communities that were permitted literacy and office and the accumulated disadvantages of those that were not. The constitutional text is on this side in terms: Articles 15(4) and 16(4) are not grudging exceptions carved out of equality but the framers’ own statement that equality in a caste ordered society requires the state to make special provision, and Article 340 shows the framers expecting identification to be a continuing, evidence based function of government. Proponents add that representation in the services is itself a public good independent of the individual beneficiaries: a bureaucracy drawn from the society it administers decides differently, and the visible presence of backward class officers in districts where their communities were historically excluded from authority changes what the state is, not only who it employs.

The case against, at its strongest, accepts the diagnosis and disputes the instrument’s operation after seventy years. Its first limb is concentration: within the backward classes, a small number of relatively advanced communities, and within them their better off families, have taken a disproportionate share of the reserved posts and seats, so that the policy’s benefits pool at the top of the category it created; that is the complaint the creamy layer was invented to answer and that the Rohini reference concedes requires examination. Its second limb is enforcement: the creamy layer exclusion depends on certificates, income computations and status categories administered by the same local machinery whose failures are documented across this series, and an exclusion imperfectly policed delivers the quota to precisely the families it excludes on paper. Its third limb is categorical: identification by caste, renewed by every list revision and by the turn to census enumeration, entrenches caste as the organising fact of a citizen’s relation to the state, teaching each generation to present itself through the category the Constitution hoped to make unnecessary. On this view the remedy should migrate over time toward economic and educational criteria applied to individuals, sub-categorisation at most, and a defined trajectory of withdrawal as representation corrects.

A third position, distinct from both, has grown with the data debate and should be recorded separately because it cuts across the other two. On this view, associated in public argument with the parties of the backward classes themselves and stated most plainly after the Bihar survey, the flaw in the settlement is not that the quota exists or that it concentrates, but that it is disproportionate: a category measured at fifty two per cent of the population in 1980 and at sixty three per cent of Bihar in 2023 holds twenty seven per cent of central posts, and the remedy is to raise the quantum toward the population share and to lift the fifty per cent ceiling that blocks the way. The judicial answer to proportionality has been consistent since Balaji: reservation is a device for adequate representation, not a mirror of population, and population share alone has never sufficed to justify a quota in the Court’s reasoning. The political answer has been the opposite movement, the ceiling defended as the settlement’s load bearing wall. The proportionality claim is therefore the one live argument in the field that is aimed at the quantum rather than the list, and its prospects turn less on doctrine than on whether the census numbers, when they arrive, make twenty seven per cent look like a settlement or like an anomaly. This history takes no side in that argument; it records that the argument, dormant since 1992, has been rearmed by enumeration, and that the settlement’s stability, which this article has treated as its central fact, is a stability of the list era that the count era may not preserve.

Two points of common ground are worth isolating, because they mark where the argument is actually moving. The proposition that benefits have concentrated within the category is accepted across the spectrum: it is the premise of the creamy layer doctrine, which reservation’s defenders administer; of the Rohini reference, which a government committed to the quota commissioned; and of the Davinder Singh majority’s acceptance of sub classification in the scheduled lists. And the proposition that the data base is indefensible is likewise shared: the courts demand quantifiable data, the commissions extrapolate from 1931, the one modern enumeration was gathered and withheld, and the decision to count caste in the forthcoming census was taken by a government that had previously declined to release the 2011 returns. The live dispute is therefore narrower than its rhetoric. It is not whether the backward classes exist, nor whether the share should be twenty seven per cent, on which there is no serious legislative movement in either direction. It is whether the category should be subdivided, how the advanced within it should be excluded, and who, Centre or state, holds the pen that writes the list.

Forty-three years from report to list: the timeline

The artifact below compresses the history into the sequence the argument turns on. Its final column records, at each stage, who decided who was backward, because that is the question that changes hands at every row. Forty three years separate the Mandal report from the Rohini report. The first commission predates both, and the table begins there, in 1953, with the first answer the Republic gave and declined to use.

Stage	What was decided	Who decided who was backward
First Backward Classes Commission, 1953 to 1955	2,399 castes listed, large quotas proposed, report laid before Parliament and not implemented	The Kalelkar Commission, whose chairman then qualified its own answer; in the event, no one
The Centre withdraws, 1961	No all India list to be prepared; states left to identify their own backward classes	Each state government, for its own services
Second Backward Classes Commission, 1979 to 1980	3,743 castes identified by eleven indicators; 27 per cent recommended in services and education, with land, credit and education measures	The Mandal Commission, by composite scoring on a 1931 data base
The decade of deferral, 1980 to 1990	Report laid before Parliament with a memorandum recording that it required examination; no action	No authority; the question held open by successive governments
Office memorandum, 13 August 1990	27 per cent reserved in central services; the operative list confined in the first phase to castes common to the Mandal list and the state lists	The Union executive, by intersecting two existing lists
Office memorandum, 25 September 1991	Preference to poorer sections within the quota; a separate 10 per cent reserved on economic criteria alone	The Union executive, extending identification beyond backward classes
Indra Sawhney judgment, 16 November 1992	The 27 per cent upheld; the 10 per cent struck down; creamy layer exclusion required; a permanent body for the lists directed	The Supreme Court, setting the conditions on which executive identification would be accepted
Creamy layer memorandum and the statutory commission, 1993	Exclusion criteria notified, income test set at one lakh rupees; National Commission for Backward Classes created by Act of Parliament	The Union executive on an expert committee’s advice; the statutory commission thereafter advising on the central list
Education extension, 2006 to 2008	Article 15(5) inserted in 2005; Act 5 of 2007 reserved 27 per cent in central educational institutions; upheld in Ashoka Kumar Thakur with creamy layer exclusion	Parliament, extending the settled list to admissions; the Supreme Court, confirming the conditions
Sub-categorisation reference, 2 October 2017	A commission under Article 340 appointed to examine inequitable distribution of benefits within the central list and to propose sub categories	The Rohini Commission, asked to divide the list against itself
Constitutional amendment, 2018	The commission made a constitutional body under Article 338B; specification of backward classes placed with the President under Article 342A	The President and Parliament, on the text as enacted; the states’ position left ambiguous
Maratha judgment, 5 May 2021	Article 342A read as vesting identification in the President alone; the states held to have lost the power	The Supreme Court, by three to two
Restoring amendment, 2021	Article 342A(3) inserted, expressly preserving each state’s power to prepare and maintain its own list	Parliament, returning the power to the states in terms
Rohini report submitted, 31 July 2023	The sub-categorisation examination completed after thirteen extensions; the report not published and not acted upon	The commission, in a report no reader outside government has seen; the question formally unanswered

Verdict: the quantum settled, the list never

The series to which this article belongs has repeatedly found that a policy’s stable headline conceals an unstable operative mechanism, and the backward classes settlement is the purest instance in the constitutional field. The headline is a percentage, and it has been stable since 1990 through changes of government, two great litigations, a constitutional amendment on a neighbouring category and repeated political attempts to move it in both directions. The mechanism is a list, and the list has been in continuous contention for the same thirty years: drawn by one commission and disowned, drawn by another and shelved, borrowed by an executive intersection, pruned by a court, advised on by a statutory body, constitutionalised, centralised by interpretation, decentralised by amendment, and referred for internal division to a commission whose report is unpublished. A candidate’s eligibility in any actual recruitment turns on the notified list and the certificate rules of that recruitment, and the general account given here is information about the structure of the law, not a substitute for the notification that governs an individual case.

A legislative history is also a prediction instrument, and this one’s record predicts the manner of the next settlement. Every durable change in this field has arrived by one of three routes, and the route has determined the change’s lifespan. What arrived by executive memorandum, in 1990 and 1991, was immediately litigable and was rewritten by the courts within two years. What arrived by judgment, in 1992, 2008 and 2015, held until a constitutional amendment or a larger bench moved it, and the conditions the judgments attached, data, exclusion of the advanced, discipline in identification, have structured every later proposal. What arrived by constitutional amendment, in 2005, 2018 and 2021, changed the ground on which the other two routes operate, and the one amendment that misjudged its own effect was corrected by another amendment within three years. A reader asked to predict the fate of sub-categorisation should therefore ask first by which route it is proposed to arrive: by memorandum it will be stayed and tested, by list revision through the commission it will be litigated entry by entry, and only by constitutional text or by a judgment that blesses it in advance is it likely to arrive whole. The passage record of seventy years contains no fourth route.

The sequence also yields a reading device for any future episode, which the aspirant and the journalist can apply before the commentary hardens. Every event in this field answers three questions, and confusion enters when they are collapsed into one. The first is the quantum question, how much is reserved, which the settlement of 1992 answered for the Centre and which reopens only through the proportionality route. The second is the membership question, who is in the category, which the list answers and which is reopened continuously by the commission, the courts and the census. The third is the authority question, who may answer the second question, Centre or state, commission or cabinet, court or legislature, which the amendments of 2018 and 2021 and the Maratha judgment fought over in terms. An announcement in this field should be interrogated in that order: does it move the number, does it move the names, or does it move the pen? Most announcements that are reported as revolutions in reservation move only the pen, and a few, the sub-categorisation reference among them, promise to move the names while leaving the number untouched, which the history suggests is the more consequential kind.

The open items are specific, and they are all identification items. The Rohini report awaits a decision on publication and action. The creamy layer income ceiling, last revised in 2017 to eight lakh rupees a year, awaits the periodic revision its own memorandum contemplates. The forthcoming census will produce the first national caste enumeration since 1931, and every list in the country will then be argued over again against new numbers. None of these will move the twenty seven per cent, which is no longer where the politics of this subject lives. It lives where it has lived since a chairman in 1955 lost confidence in his own list: in the question of who is backward, asked of a republic that has spent seventy years building machinery to answer it and has never yet let an answer rest.

Frequently Asked Questions

Q: What did the Mandal Commission recommend?

The commission recommended reservation of 27 per cent of posts for the socially and educationally backward classes in the services of the Union and the public sector, extending to promotions at all levels, and the same proportion in universities, colleges, public sector undertakings, banks and private undertakings receiving government assistance, with unfilled reserved vacancies carried forward for three years, age relaxation on the Scheduled Caste and Scheduled Tribe pattern, and a roster system. It also recommended land reform and redistribution, credit and technical support for backward class artisans and small industry, scholarships and special educational facilities, and legal backing for the scheme. Only the services reservation was ever implemented, a decade later and without the promotion element.

Q: When was reservation for other backward classes actually implemented?

The Union government announced implementation in Parliament on 7 August 1990 and issued the office memorandum reserving 27 per cent of central civil posts on 13 August 1990. The order was then stayed by the Supreme Court on 1 October 1990 while the challenge to it was heard, so no appointments were made under it during the litigation. After the judgment of 16 November 1992 upheld the reservation subject to exclusion of the creamy layer, the government notified the exclusion criteria by the office memorandum of 8 September 1993, and recruitment against the 27 per cent quota proceeded from that corrected framework. State level reservation for backward classes is much older and had operated under state lists for decades before 1990.

Q: What is the creamy layer and who is excluded?

The creamy layer is the socially advanced section within a backward class, which the Supreme Court in 1992 held must be excluded from reservation so that the benefit reaches the genuinely disadvantaged members of the class. The exclusion criteria were notified by the office memorandum of 8 September 1993 and operate by category rather than by income alone. They cover the children of holders of constitutional posts, of senior Group A and specified Group B officers, and of senior armed forces officers; children of professionals, traders and property owners in defined circumstances; and families above an income and wealth threshold. The income ceiling was set at one lakh rupees a year in 1993 and revised in stages to eight lakh rupees in 2017, with income from salary and agricultural land left out of the computation.

Q: How are backward classes identified?

Identification begins with the constitutional machinery of Article 340, under which commissions investigate backwardness and recommend, but the operative identification is the list. For central purposes the backward classes were first taken, in 1990, as the castes common to the Mandal Commission’s list and the state lists, and the central list has since been maintained through the advisory process of the National Commission for Backward Classes and notification by the Union government. Since the amendment of 2018, specification for central purposes follows Article 342A, with the President specifying the classes and Parliament controlling later inclusions and exclusions. Since the restoring amendment of 2021, every state also prepares and maintains its own list for its own purposes, and the two lists may differ. A caste is therefore backward in law when it stands in the list that governs the post or seat in question.

Q: How did reservation reach central educational institutions?

Reservation in services rested on executive memoranda under Article 16(4), but admission to private educational institutions raised questions the existing text did not answer, so Parliament first amended the Constitution. The Ninety-third Amendment of 2005 inserted Article 15(5) with effect from 20 January 2006, empowering special provision by law for admission to educational institutions including private aided and unaided institutions, other than minority institutions. Parliament then enacted the Central Educational Institutions (Reservation in Admission) Act, 2006, Act 5 of 2007, reserving fifteen per cent of seats for the Scheduled Castes, seven and a half per cent for the Scheduled Tribes and twenty seven per cent for the backward classes in central institutions, with a mandatory increase in seats. The Supreme Court upheld the amendment and the Act in Ashoka Kumar Thakur in 2008, subject to exclusion of the creamy layer.

Q: What is the National Commission for Backward Classes now?

The commission began as a statutory body under the National Commission for Backward Classes Act, 1993, created after the Supreme Court in 1992 directed that a permanent expert body examine the backward classes lists. The Constitution (One Hundred and Second Amendment) Act, 2018 repealed that statutory footing and made the commission a constitutional body under Article 338B, with a chairperson, vice chairperson and members appointed by the President, functions of investigating and monitoring the safeguards for the backward classes, inquiring into specific complaints, and advising on socio economic development, together with the powers of a civil court for its inquiries. Its reports go to the President and are laid before Parliament. The amendment of 2021 adjusted its consultation role in relation to the state lists that the same amendment restored.

Q: What happened to the Rohini Commission on sub-categorisation?

The commission, chaired by Justice G. Rohini, was appointed by the President under Article 340 on 2 October 2017 to examine the inequitable distribution of reservation benefits among the castes in the central list of backward classes and to work out a mechanism for sub-categorisation within the category, along with a study of the list’s entries and their errors. It was initially given twelve weeks and received thirteen extensions, largely because caste wise data on the distribution of benefits had to be assembled from records that did not routinely keep it. The commission submitted its report to the President on 31 July 2023. The report has not been made public, and no government action on it has been announced, so sub-categorisation of the central list remains a proposal under examination rather than a policy in force.

Q: Why is a caste census demanded?

The demand rests on the age of the data underneath every list. The last census to enumerate caste across India was held in 1931, the Mandal Commission’s population estimate of about fifty two per cent was built on that base, and the Supreme Court has repeatedly required quantifiable data to sustain classification and has held that the 1931 figures cannot be determinative. Supporters of enumeration argue that the quantum, the composition of the lists and any sub-categorisation all distribute benefits among units the state declines to count, and that a count is a precondition of rational policy in any direction. Opponents argue that state enumeration hardens caste identity and invites competitive claims to backwardness. The caste returns gathered by the Socio Economic and Caste Census of 2011 were not published, and the Cabinet Committee on Political Affairs decided on 30 April 2025 to include caste enumeration in the forthcoming census.

Q: How many castes and communities did the Mandal Commission identify as backward?

The commission identified 3,743 castes and communities as socially and educationally backward. It reached the figure by applying its eleven indicators, grouped as social, educational and economic tests with the social indicators carrying the greatest weight, to the state lists and the material available from the census of 1931 and its own inquiries, and a caste scoring more than half of the available points was classified as backward. The commission estimated that the classes so identified formed about fifty two per cent of the population. The list was never adopted wholesale by the executive: the office memorandum of 1990 took, in its first phase, only the castes common to the Mandal list and the lists the states already maintained, and the central list has been revised entry by entry ever since.

Q: Which of the Mandal Commission’s recommendations were not about reservation at all?

Roughly half the report concerned the conditions that produced backwardness rather than the distribution of posts. The commission recommended radical land reform, including redistribution of ceiling surplus land and a change in the relations of production in agriculture; dedicated financial institutions and subsidised credit for backward class artisans, traders and small industries, with technical support for village occupations; scholarships, hostels, coaching and a chain of special educational facilities including residential schools in backward class areas; and adult education directed at the same communities. These measures received almost no attention in 1980 and none at implementation in 1990, when the government adopted the reservation recommendation alone. The pattern repeated the fate of the Kalelkar report, whose land, credit and education programme was likewise the part nobody campaigned on.

Q: Why was the Kalelkar Commission report never implemented?

The first Backward Classes Commission, appointed in 1953 under Kaka Kalelkar, reported in March 1955 with a list of 2,399 castes and large recommended quotas, but the report arrived divided against itself. Several members dissented from the report’s main findings, and the chairman’s own letter transmitting the report expressed reservations about caste as the criterion of backwardness and about the remedies proposed, which the government of Jawaharlal Nehru treated as disabling indecision. The report was laid before Parliament and not acted upon, and in 1961 the Union government decided against preparing an all India list at all, leaving identification and measures to the states. The episode established that a commission’s list has force only when a government is willing to own it, which is the condition the Mandal report met, thirty five years later, in 1990.

Q: Did states reserve posts for backward classes before the Mandal Commission reported?

Yes, and for decades. Southern states operated communal quotas in services from the colonial period onward and maintained extensive backward classes reservation under their own lists after independence, at a time when the Centre had no such policy following the shelving of the Kalelkar report. Mysore’s order reserving 68 per cent of seats produced the Supreme Court’s warning in Balaji in 1963 that special provision should ordinarily stay below fifty per cent. Bihar under Karpoori Thakur adopted a formula in 1978 reserving 26 per cent of state posts, subdivided between backward and most backward classes in separate annexures with shares for women and the poor among the forward castes. The Mandal Commission therefore nationalised and standardised a policy the states had long practised, rather than inventing one.

Q: What did the office memorandum of 25 September 1991 change, and what became of it?

The government of P. V. Narasimha Rao amended the 1990 memorandum in two ways. It introduced an economic gradient inside the 27 per cent, directing that preference within the quota be given to the poorer sections of the backward classes, and it reserved a further 10 per cent of vacancies for other economically backward sections of the people not covered by any existing scheme of reservation, a category defined by economic weakness alone. In the judgment of 1992 the Supreme Court upheld the internal preference, read as a classification by relative backwardness among the identified classes, and struck down the separate 10 per cent as constitutionally invalid, because a reservation resting on economic criteria alone could not be sustained under Article 16 as it stood. The idea later returned through constitutional amendment in the economically weaker sections provisions of 2019.

Q: What happens to a backward classes commission report after it is submitted to the President?

Article 340 completes its own procedure. The President is required to cause a copy of the report, together with a memorandum explaining the action taken on the commission’s recommendations, to be laid before each House of Parliament. The Constitution imposes no obligation to accept any recommendation, and the memorandum of action taken is the only formal answer a government must give. The three commissions in this history show the range of outcomes the procedure permits: the Kalelkar report was laid and not implemented, the Mandal report was laid with a memorandum recording that it required examination and was implemented ten years later by a different government, and the Rohini report was submitted in July 2023 and has not been published at all. Laying informs Parliament; it does not bind the executive.

Q: What did L. R. Naik argue in his dissent from the Mandal report?

L. R. Naik, a member of the commission, dissented on the internal structure of the category the report created. He argued that the backward classes were not a single bloc of comparable disadvantage, that the relatively advanced communities within the list would capture the new reservation at the expense of the most depressed, and that the category should be divided, with the most backward classes given a protected share of their own rather than left to compete inside a common quota. The majority did not accept the division, and the 27 per cent was recommended as one block. Naik’s objection is the earliest official statement of the sub-categorisation case: the commission appointed in 2017 under Justice Rohini was asked, thirty seven years later, to work out the mechanism the dissent had demanded in principle in 1980.

Q: Why do the central list and the state lists of backward classes differ?

Because the Constitution, as finally settled in 2021, maintains two identifications for two purposes. The central list governs reservation in Union services and central institutions, and since 2018 it is specified under Article 342A by the President, with later inclusions and exclusions reserved to Parliament. Each state list governs that state’s services and institutions, and states built their lists from their own inquiries over decades, weighing local conditions in which the same community may be advanced in one state and backward in another. The amendment of 2021 wrote the divergence into the text: Article 342A(3) provides that a state list prepared for the state’s own purposes may differ from the central list. The practical consequence is that eligibility is a relation among the person, the list and the purpose, and a community can be entitled in its state’s services and not in the Union’s.

Q: What is sub-categorisation within the backward classes meant to achieve?

Sub-categorisation divides the single backward classes quota into internal shares, so that the communities within the list compete in smaller pools rather than in one pool of over 2,600 central entries. Its object is distributive: where a small number of relatively advanced communities take a disproportionate share of the reserved posts and seats, an undifferentiated quota reproduces inequality inside the category, and internal shares are meant to guarantee the less represented communities a portion they cannot win in open competition with their own category’s strongest members. The idea has a long pedigree, from L. R. Naik’s dissent in 1980 through the annexure systems several states operate, and it was the reference of the Rohini Commission appointed in 2017. Whether it corrects concentration or merely multiplies boundary disputes is the contested question the unpublished report of 2023 was meant to inform.

Q: Was caste counted in the Socio Economic and Caste Census of 2011?

Caste was counted, and the count was not published. The Socio Economic and Caste Census of 2011 was conducted as an exercise separate from the decennial population census, and its enumerators recorded the caste of each household alongside socio economic indicators. The socio economic findings were released, but the caste returns were withheld, the returns having produced caste and sub caste names in bewildering profusion, surnames, gotras, occupations and spelling variants recorded as castes, a harvest that illustrated the difference between asking a household its caste and enumerating the castes of a nation, the government stating that the data contained errors on a scale that made release unwise. The result is that India possesses a modern caste enumeration that no policymaker, court or researcher can use, and the identification of backward classes continues to rest on the census of 1931, commission estimates and state surveys conducted for particular purposes. The decision taken in April 2025 to include caste enumeration in the forthcoming population census is intended to replace that position with a count designed for publication.

Q: How did the amendment of 2021 restore the power of states to maintain their own backward classes lists?

The Constitution (One Hundred and Fifth Amendment) Act, 2021 answered the Maratha judgment of the same year, in which a Constitution Bench had read Article 342A, inserted in 2018, as vesting the identification of backward classes in the President alone and as having taken the power from the states. The amendment inserted clause (3) into Article 342A, providing in express words that every state or Union territory may, by law, prepare and maintain for its own purposes a list of socially and educationally backward classes, entries in which may differ from the central list maintained for the purposes of the Central Government. It also made corresponding changes to Article 338B, on the National Commission for Backward Classes, and to the definition in Article 366(26C). State lists that the judgment had cast into doubt were thereby restored as a matter of text.

Q: How did Bihar’s caste survey change the argument about reservation?

Bihar’s survey, whose findings were released on 2 October 2023, replaced extrapolation with enumeration for one large state. It found the Extremely Backward Classes at 36.01 per cent and the Other Backward Classes at 27.13 per cent of a population of a little over thirteen crore, the two together forming 63 per cent, while the unreserved categories formed 15.52 per cent. The figures converted the abstract data argument into a concrete distributional claim: the state government raised total reservation to 65 per cent by legislation in 2023 on the strength of the survey, and the Patna High Court struck the increase down in 2024 for want of the exceptional circumstances the fifty per cent doctrine requires. The episode showed enumeration changing what can be argued without changing, by itself, what can be enacted, which is the distinction on which the national caste census debate turns.