Reservation jurisprudence has no theory

A constitutional doctrine, on the standard account, is a controlling theory plus the cases that have applied it. Take the dormant Commerce Clause in American constitutional law, or proportionality review in European public law, or — to stay close to home — the basic-structure doctrine in India. Each has a theory: a stated proposition about what the doctrine is for, in what kind of case it applies, what it requires the Court to do. Indian reservation jurisprudence has no such theory. It has cases, in abundance — the seven-judge bench in Champakam Dorairajan, the five-judge bench in MR Balaji, the seven-judge bench in NM Thomas, the nine-judge bench in Indra Sawhney, the five-judge bench in M Nagaraj, the five-judge bench in Ashoka Kumar Thakur, the five-judge bench in Jarnail Singh, the five-judge bench in Janhit Abhiyan, the seven-judge bench in State of Punjab v Davinder Singh. What it does not have is a controlling theory. Each of these benches reached for a different account of what equality requires, what reservation is for, and how the constitutional clauses fit together. None of those accounts has been adopted as the field's anchor. The doctrinal incoherence is not a temporary state. It is the structure of the field.

This essay attempts to map the landscape. Three theories of reservation are visible in the case law, each invoked at different moments by different benches. The first is corrective: reservation is a remedy for past discrimination, owed to particular groups because the State or society inflicted historic disabilities on them. The second is descriptive: reservation is representation in public institutions, required because the institutions of a self-governing republic should mirror the demographic composition of the people they govern. The third is transformative or substantive-equality: reservation is the constitutional vocabulary for treating unequals as unequals, the corollary of the proposition that formal equality without redistributive correction is not equality at all. The Court has used all three. It has never picked one. The argument of this essay is descriptive, not prescriptive: the field is the way it is because the Court has had no occasion to choose between these theories, and successive constitutional amendments have made the choice progressively harder to face.

The opening positions: Champakam and the First Amendment

The story starts where the Indian Constitution starts to mean things. In 1951, the Madras Government's 'Communal G.O.' reserved seats in State medical and engineering colleges by caste and community. The Supreme Court, a seven-judge bench in State of Madras v Champakam Dorairajan, struck it down.² The reasoning was elementary. Article 15(1) of the Constitution prohibited the State from discriminating against any citizen on grounds only of religion, race, caste, sex, or place of birth. Article 29(2) prohibited the denial of admission to any educational institution maintained by State funds on those same grounds. The Madras G.O. classified candidates on caste-and-community lines and admitted them on caste-and-community quotas. The relevant fundamental rights were violated. Article 46 — a Directive Principle on the educational and economic interests of the weaker sections — could not override the fundamental-rights position because Directive Principles are not enforceable and cannot defeat the rights in Part III. The Court was not, in Champakam, choosing between theories of equality. It was applying the simplest possible theory: equality means non-classification on prohibited grounds.

Parliament responded within the year by inserting Article 15(4) into the Constitution through the First Amendment, permitting the State to make 'any special provision for the advancement of any socially and educationally backward classes of citizens or for the Scheduled Castes and the Scheduled Tribes'. The amendment did not say what 'special provision' meant, what theory of equality it embodied, or how it sat against the rest of Article 15. The framers' silence on these questions is the silence the subsequent jurisprudence has been arguing about, in fragments, ever since. The First Amendment is the moment the Indian Constitution became a document with two competing visions of equality on the same page, with no instruction manual for resolving the tension between them.¹

For the first decade after Champakam, the Court resolved the tension by treating Article 15(4) as an exception to Article 15(1). This is the position taken in MR Balaji v State of Mysore in 1962-63.³ A five-judge bench struck down a Karnataka Government order classifying 90 per cent of the State's population as 'backward' on caste lines and reserving 68 per cent of medical and engineering seats for them. Gajendragadkar J's opinion did three things that have shaped the field ever since. It held that caste 'could not be the sole or dominant criterion' in identifying backwardness; that 'social backwardness is in the ultimate analysis the result of poverty to a very large extent'; and that reservations could not, as a matter of constitutional principle, exceed 50 per cent. The 50 per cent ceiling — which would be confirmed, undone, reconfirmed, and now sits in a contested state after Janhit Abhiyan — was first announced in Balaji. It was announced as a consequence of treating Article 15(4) as an exception: if the exception swallows more than half the field, it ceases to be an exception. The theory of equality was corrective-formal: equality means equal treatment, the exception is for past wrongs, and the exception cannot grow large enough to displace the rule.

The flip: NM Thomas and substantive equality

The doctrinal switch came in 1976, with a seven-judge bench in State of Kerala v NM Thomas.⁴ The case turned on a State rule that gave temporary lower-division clerks belonging to Scheduled Castes and Scheduled Tribes additional time to clear a departmental examination required for confirmation as permanent staff. The rule was, on a formal-equality reading, classification on the ground of caste in matters of public employment — a prima facie violation of Article 16(1). The Court upheld the rule, and in doing so it announced a different theory of equality from the one in Balaji.

Ray CJ's majority opinion held that Article 16(4) is not an exception to Article 16(1) but an illustration of it. The right to equal opportunity in public employment is not breached when a State recognises that 'unequals cannot be treated equally'. If the equality clause does not require, but in fact forbids, the equal treatment of those whom history has placed in different positions, then the special provisions in Articles 15(4), 16(4), and 46 are not deviations from the equality principle; they are applications of it. The theory of equality is now substantive: equality means recognising and correcting the substantive disadvantages that formal equal treatment would perpetuate. This is a different theory from the one in Balaji, and the Court in NM Thomas did not acknowledge it as different. It treated the new position as if the old one had been a confusion, not a different theory.

If NM Thomas were the last word, the field would have a controlling theory: substantive equality, with reservation as one of its instruments, and Article 16(4) as a textual confirmation rather than a textual exception. But NM Thomas was not the last word, and the substantive-equality vocabulary it introduced has had to coexist, awkwardly, with the corrective-formal vocabulary that Balaji had introduced and that subsequent benches were unwilling to discard. The seventeen years between NM Thomas and Indra Sawhney are a period of doctrinal drift, in which the Court alternates between the two theories without ever explaining why.

Indra Sawhney and the legislated compromise

In 1992, a nine-judge bench handed down what is, by every measure, the central reservation case in Indian constitutional law: Indra Sawhney v Union of India.⁵ The case had been listed to test the V P Singh Government's 1990 decision to implement the Mandal Commission's recommendation of 27 per cent reservation in central public employment for the Other Backward Classes. The nine-judge bench delivered six separate majority opinions and three dissents. Jeevan Reddy J's opinion is generally treated as the leading judgment. It is also, on close reading, an attempt to broker a compromise between the corrective-formal theory of Balaji and the substantive-equality theory of NM Thomas — without committing to either.

What did Indra Sawhney hold? Several things at once. The 50 per cent ceiling from Balaji was retained, but on grounds that combine the formal and the prudential: a reservation in excess of 50 per cent would, the Court said, defeat the equality of opportunity guaranteed by Article 16(1). The exception-vs-illustration debate from NM Thomas was largely sidestepped; Jeevan Reddy J treats Article 16(4) as both, depending on which provision is being considered. Caste-based identification of backwardness was upheld, but with the qualification that caste could not be the 'sole' criterion — a return, in effect, to the formulation in Balaji. The creamy-layer carve-out was announced for the first time as a constitutional requirement for OBC reservations: individuals within the OBC list who had achieved a certain economic and social status would no longer be eligible for the reservation benefits.¹³ Reservation in promotion was held to be unconstitutional under Article 16(4) as it then stood — a holding Parliament reversed within three years through the Seventy-seventh Amendment, inserting Article 16(4A).

The theory of equality in Indra Sawhney is, on inspection, all three of the theories I named at the start of this essay. The creamy-layer carve-out is corrective: it sorts those who have, individually, overcome the disability the reservation is meant to remedy. The 50 per cent ceiling is formal: it preserves the structural priority of equal opportunity for the unreserved category. The acceptance of caste-based identification of OBCs is substantive-transformative: it recognises that disadvantage is, in India, distributed along caste lines, and that the constitutional measure of equality must engage with that distribution. Indra Sawhney does not choose between these three theories. It picks elements from each and assembles them into a doctrinal compromise that has the appearance of a controlling decision but does not, in fact, control the field.

What makes Indra Sawhney a compromise rather than a controlling judgment is the way its holdings have been progressively unbundled by subsequent amendments and benches. The reservation-in-promotion holding was reversed by Parliament. The creamy-layer holding was applied to OBCs in Indra Sawhney itself, then declined for SCs and STs in Ashoka Kumar Thakur in 2008, then applied to SCs and STs in Jarnail Singh in 2018 — a five-judge bench reversing a five-judge bench on a basic-structure question without convening a larger bench, because the doctrinal vocabulary of Indra Sawhney permits both readings.⁷⁸ The 50 per cent ceiling, once described as a basic-structure feature, was crossed by the EWS reservation upheld in Janhit Abhiyan.⁹ A compromise that has been unbundled in this many places is not a controlling decision. It is the doctrinal placeholder a divided Court left behind in 1992 because it could not agree on more.

Nagaraj and the data trap

In 2006, a five-judge bench in M Nagaraj v Union of India upheld the constitutional amendments that had, between 1995 and 2001, restored reservation in promotion and added consequential seniority for SCs and STs.⁶ The Court held that the amendments were within the constituent power and did not violate the basic structure. But it imposed three procedural conditions on the operation of any specific promotion-reservation policy: the State must have 'quantifiable data' showing the backwardness of the class, 'quantifiable data' showing inadequacy of representation in the cadre concerned, and an assessment of the impact on 'overall administrative efficiency' under Article 335.

The theoretical move in Nagaraj is interesting. It is not the substantive-equality theory of NM Thomas, because Nagaraj does not say that promotion-reservation is an application of the equality principle; it says it is constitutionally permissible if the State can demonstrate three specific empirical facts. The data requirement looks descriptive — reservation is for representation in the cadre, and representation is a question of cadre composition that requires data. But the backwardness requirement looks corrective — reservation is for a class whose past disabilities have a present footprint, and the present footprint requires data. The efficiency requirement looks formal — equality is constrained by administrative competence, and where the constraint binds, the reservation must yield. Nagaraj blends all three theories, in different proportions for different conditions, and presents the blend as if it were a single coherent test. It is not. It is a test that asks the State to satisfy three different equality theories at once, which is one reason why Nagaraj-compliance has produced a generation of reservation litigation in which the Court has rarely been satisfied that the data is adequate.

Jarnail Singh in 2018 partially relaxed the Nagaraj data requirement, holding that the State need not collect quantifiable data showing the backwardness of the SC/ST class because the Constitution had already identified those classes as backward.⁸ This is a return to a more straightforwardly corrective theory: the constitutional identification of SCs and STs is itself the substantive proof that the corrective work is owed. Whether this reasoning is consistent with Indra Sawhney's position that creamy-layer exclusion applies only to OBCs is a question Jarnail Singh did not address; the bench simply held that creamy-layer applied to SCs and STs in the promotion context as well. The doctrinal move was made without confronting the theoretical disagreement it embodied.

Janhit Abhiyan: economic-only reservation and what it implies

The most recent realignment came in Janhit Abhiyan v Union of India in 2022, decided by a 3-2 majority of a five-judge bench.⁹ The Hundred-and-third Amendment to the Constitution had inserted Article 15(6) and Article 16(6), authorising 10 per cent reservation for the 'Economically Weaker Sections' of citizens 'other than the classes mentioned in clauses (4) and (5)' of Article 15 — that is, other than the SCs, STs, and OBCs. The amendment was challenged on three grounds. First, reservation on a purely economic criterion was said to violate the equality principle as developed in Indra Sawhney, which had held that economic criterion alone was not a permissible ground. Second, exclusion of SCs, STs, and OBCs from EWS eligibility was said to be itself a violation of Article 14, since a poor person from those communities is no less economically weak than a poor person from a forward caste. Third, the 10 per cent reservation took the total reservation in central institutions across 50 per cent, breaching the Indra Sawhney ceiling.

The majority — Maheshwari J, Trivedi J, and Pardiwala J — upheld the amendment on all three grounds. Economic backwardness was held to be a permissible standalone ground for special provision, on the reasoning that the universe of disadvantage extends beyond caste and social-and-educational backwardness. The exclusion of SCs, STs, and OBCs was justified on the basis that those classes already had separate reservations and adding them to EWS would amount to double benefit. The 50 per cent ceiling was held not to be a basic-structure rule but a 'rule of prudence' that could be set aside where the constitutional amendment so required. Lalit CJ and Bhat J dissented on the second and third grounds, holding that the exclusion of SCs, STs, and OBCs from EWS eligibility violated Article 14 and that the breach of the 50 per cent ceiling rendered the amendment constitutionally infirm.

What theory of equality is doing the work in Janhit Abhiyan? The majority opinion is, on inspection, drawing on a fourth theory — one that the previous cases had not announced. Reservation is, on this view, a permissible legislative device for the alleviation of any structurally identifiable disadvantage, with the Constitution permitting different streams for different forms of disadvantage. SC/ST reservation is for caste-based historical disability. OBC reservation is for social and educational backwardness, however defined. EWS reservation is for economic weakness in groups not otherwise covered. The streams need not be analytically comparable; they need only be constitutionally authorised. This is a pluralist theory of reservation — one that abandons the search for a single equality principle and instead treats each reservation stream as a separate constitutional instrument with its own conditions.

The dissenters were arguing on a different theory altogether. For them, equality under Article 14 requires that the State's classifications be tested against a uniform standard, that the standard is substantive equality understood as the correction of present disadvantage, and that any reservation scheme must justify itself within that single framework. The exclusion of SCs, STs, and OBCs from EWS, on this account, fails because economic disadvantage among those communities is structurally identical to economic disadvantage among forward-caste communities, and the differentiation has no equality-related justification.

Janhit Abhiyan is the clearest contemporary illustration of the field's doctrinal incoherence. A five-judge bench, with three judges on each side of the central theoretical question, split 3-2 on which theory of equality controls reservation review. The 3-2 split is not on a matter of doctrinal application; it is on what doctrine applies. That kind of split is the signature of a field that has not chosen its controlling theory.

Sub-classification: Davinder Singh and the limits of EV Chinnaiah

The most recent major intervention is State of Punjab v Davinder Singh in 2024, a seven-judge bench that expressly overruled the five-judge bench in EV Chinnaiah v State of Andhra Pradesh from 2005.¹⁰ The question was whether the States may sub-classify Scheduled Castes — that is, divide the SC list into more and less disadvantaged sub-groups and allocate reservation benefits differentially within the SC quota. Chinnaiah had held that the Constitution treats SCs as a single homogeneous class, identified by the President's notification under Article 341, and that no further classification within that class is constitutionally permissible. Davinder Singh overruled that holding, holding that sub-classification is permissible if the sub-classification rests on intelligible differentia and on a demonstrated inadequacy of representation of the sub-group within the SC quota.

The theoretical move in Davinder Singh is striking. The seven-judge bench reasoned that the SC category is a constitutional aggregation of historically disadvantaged groups that are not, sociologically, uniformly disadvantaged. To treat them as uniformly disadvantaged is to perpetuate, within the reservation system itself, the inequalities the system was meant to correct. The Court therefore drew on substantive-equality reasoning — the very reasoning NM Thomas had introduced and that Indra Sawhney had only partially adopted — to permit sub-classification. But the Court did not say it was overruling Chinnaiah because Chinnaiah had been decided on the wrong theory; it said it was overruling Chinnaiah because Chinnaiah had misread Articles 341 and 342. The doctrinal move was theoretical; the doctrinal vocabulary was textual.

This pattern — substantive moves dressed in textual or precedential clothing — is characteristic of the reservation jurisprudence. The Court has rarely been willing to say, in so many words, that an earlier bench had the equality theory wrong. It has preferred to distinguish, to refine, to add procedural conditions, to overrule on subsidiary grounds. The reason is partly institutional: the Court that decides a reservation case in 2024 is composed of judges who came to the Bench through a long process in which earlier reservation cases were authoritatively taught as part of constitutional doctrine, and judges are not generally in the business of repudiating their predecessors' theoretical commitments. But the consequence is a body of jurisprudence that has, in net effect, moved through three different equality theories without ever settling on any of them as controlling.

The three theories, laid out

Let me state the three theories more precisely, and locate the cases that have used each.

The corrective theory holds that reservation is a remedy for historical wrongs. The State, or society with State acquiescence, inflicted disabilities on certain groups; reservation is the constitutional instrument by which those disabilities are addressed. On this view, the constitutional identification of beneficiary groups should track the structure of historical disability — caste-based for SCs, ethnic for STs, hereditary or status-based for OBCs. The duration of the remedy is, in principle, time-limited: the remedy ends when the disability is undone. The creamy-layer doctrine is at home in this theory, because it asks whether the individual within the reserved class has, in fact, escaped the disability. So is the data-requirement of Nagaraj, because the empirical demonstration of present disadvantage is what justifies continued correction.

The descriptive or representational theory holds that reservation is about the composition of public institutions. A self-governing republic owes its citizens the assurance that the institutions of public power — legislatures, civil services, public universities — will reflect the demographic composition of the citizenry. On this view, the constitutional question is not whether a group has been historically wronged but whether the group is presently under-represented in the institution at issue. The 'inadequacy of representation' condition in Article 16(4) and in Nagaraj's test is at home in this theory. So is reserved-constituency election under Articles 330 and 332, where the constitutional purpose is plainly representational rather than corrective. The descriptive theory has no inherent time limit; institutions need representational mirroring on an ongoing basis, not just until past wrongs are remedied.

The substantive-equality or transformative theory holds that equality is a positive constitutional value requiring the State to address systemic group disadvantage by whatever instruments are effective. The theory does not distinguish between corrective and descriptive purposes; it treats both as instances of the State's broader obligation to dismantle structures of subordination. On this view, reservation is one of several constitutional instruments — alongside non-discrimination protections, social welfare schemes, and procedural due-process requirements — through which the equality value is operationalised. The theory is at home with NM Thomas's 'unequals cannot be treated equally' formulation, with the broader reading of Articles 15(4) and 16(4) as illustrations rather than exceptions, and with the seven-judge bench's reasoning in Davinder Singh.¹²

The three theories are not always inconsistent. There are cases where any of the three would produce the same outcome — the protection of an SC or ST individual against a discriminatory rule, for instance, is over-determined under each theory. But there are also cases where the three theories diverge, and the field's failure to choose one of them produces the divergences that constitute the doctrinal incoherence. The 50 per cent ceiling makes sense on a corrective theory (the remedy must not swallow the rule) and on a formal-equality theory; it has no obvious justification on a substantive-equality theory, which does not see reservation as a deviation from equality at all. The exclusion of forward-caste poor from SC/ST quotas makes sense on a corrective theory (forward-caste poverty is not the disability the reservation addresses) and on a representational theory (forward castes are not under-represented in the same way), but has trouble on a substantive-equality theory under which forward-caste poverty is itself a structurally produced disadvantage.

The Indian Court has, in different cases, decided each of these questions in ways that draw on different theories. The 50 per cent ceiling was preserved through Indra Sawhney on a corrective-formal mix; it was breached through Janhit Abhiyan on a pluralist theory that treats different reservation streams as separate constitutional instruments. The creamy-layer doctrine was extended to SCs and STs in Jarnail Singh on a corrective theory, in tension with Ashoka Kumar Thakur's earlier refusal on what was effectively a descriptive theory. The sub-classification of SCs was permitted in Davinder Singh on a substantive-equality theory, against Chinnaiah's earlier reading of Articles 341 and 342 in a textual mode that obscured its theoretical commitment.¹⁰ Each of these moves was, in its own case, defensible on the theory the bench was operating on. None of them is consistent with the others as elements of a single theory of reservation.

Why the field has stayed this way

The doctrinal incoherence of reservation jurisprudence is not a temporary state. It has been the structure of the field for at least sixty years, and there is no reason to expect it to resolve. The reason is partly institutional. The Indian Court has decided reservation cases through five-judge, seven-judge, and nine-judge benches, with each generation of judges adding its own contribution to the case law without being required to overrule earlier decisions on theoretical grounds. The reason is partly political. Reservation is, in the description of Vinay Sitapati's chapter on the subject, an area in which the political branches have been more decisive than the Court, with parliamentary majorities overriding judicial holdings on promotion, on the carry-forward rule, on the 50 per cent ceiling, on the EWS exception.¹ A field in which the constitutional amendments do much of the operative work is a field in which the Court has limited room to develop a controlling theory of its own.

And the reason is partly constitutional. The text of the relevant provisions — Articles 14, 15(1), 15(3), 15(4), 15(5), 15(6), 16(1), 16(2), 16(4), 16(4A), 16(4B), 16(6), 17, 29(2), 46, 335, 338, 341, 342 — does not announce a controlling theory of reservation. Different provisions use different vocabularies. Article 15(4) speaks of 'socially and educationally backward classes'; Article 16(4) speaks of 'Backward Class of citizens'; Article 46 speaks of 'weaker sections', particularly Scheduled Castes and Scheduled Tribes; Article 335 speaks of 'maintenance of efficiency of administration'.¹⁴ The text contains the materials for each of the three theories I have identified and does not require the Court to pick one. The Court has therefore picked all three, at different moments, in different cases, with the consequence that the field's doctrine is the assembled artefact of those choices.

Marc Galanter, writing in the 1980s about the body of reservation case law that existed before Indra Sawhney, observed that there is in Indian reservation jurisprudence not a problem of questionable doctrine but a problem of 'the absence of doctrine'.¹¹ Galanter wrote that line forty years ago, on a thinner case law than the one we now have. Adding Indra Sawhney, Nagaraj, Ashoka Kumar Thakur, Jarnail Singh, Janhit Abhiyan, and Davinder Singh has not changed the position. The body of cases has grown; the underlying doctrinal absence has remained. The field has more material now, in 2026, than it had in 1984, but it does not have a controlling theory now any more than it did then.

This is where the survey ends. The reservation jurisprudence is, on the descriptive account this essay has tried to offer, a body of case law that has accumulated through accretion rather than through theory-building. The three equality theories I have identified are visible in the cases. None has been adopted as controlling. The Court has not had to choose, because the constitutional amendments have, at every critical juncture, foreclosed the choice. The body of doctrine that has resulted is doctrinally incoherent, not because the Court has reasoned badly, but because the institutional and constitutional structure of the field has permitted the Court not to choose. Whether the field can ever acquire a controlling theory is a question this essay does not answer. It is a question that depends on whether a future bench is willing to do what the past benches have, with strategic intelligence, declined to do — name the theoretical commitment, justify it on its own terms, and let the earlier cases stand or fall by the choice. Until that happens, reservation jurisprudence will continue to be what it has been for sixty years: a sequence of decisions in search of the theory that would make them, retrospectively, into a single doctrinal field.

Get the week's biggest decisions, distilled.

A weekly digest for lawyers, students and the legally curious. Every Friday morning. Free.