India’s reservation framework is usually met one article at a time, and anyone who searches the Indian constitution reservation articles in sequence meets a scatter: a non-discrimination clause in Part III, a services clause two pages later, a directive in Part IV that no court can enforce, and a block of seat reservations and commissions in Part XVI that most summaries treat as an appendix. The claim defended in this article is that the scatter is an artefact of the reading, not of the design. The Constitution of India distributes its reservation provisions across three layers, each layer carrying a distinct logic: a rights-protective layer (Articles 15, 17 and 46) that forbids discrimination and abolishes untouchability, a services-affirmative layer (Articles 16 and 335) that authorises and disciplines preferential appointment to public posts, and a representational layer (Articles 330, 332, 334, 338 and 339) that reserves legislative seats and builds the oversight machinery. Read as layers, the framework is coherent. Read as a list, it never is.

Indian Constitution Reservation Articles: One Architecture, Three Layers

Two standard treatments organise almost everything written on this subject, and each is organised so as to miss the architecture. The first is the article-by-article enumeration, perfected in the textbook tradition that descends from D.D. Basu’s Introduction to the Constitution of India, first published in 1960: each provision is glossed in turn, its clauses parsed, its case law appended, and the reader is left with thirteen accurate commentaries and no account of why the provisions sit where they sit. The second is the case-law-driven survey, which narrates the framework as a sequence of Supreme Court confrontations, from the State of Madras v. Champakam Dorairajan judgment of 9 April 1951 through Indra Sawhney v. Union of India of 16 November 1992 to Janhit Abhiyan v. Union of India of 7 November 2022. That narrative is dramatic and true in its parts, but it makes the Constitution look like a passive surface on which courts write, when the documentary record shows a drafted structure that the courts were interpreting, sometimes against its grain and sometimes along it.

Granville Austin’s The Indian Constitution: Cornerstone of a Nation, published in 1966, remains the strongest version of the founding-centred account, and Austin saw the founding choices whole: he treated caste uplift as one strand of the social revolution the Assembly believed itself to be conducting alongside the national and democratic revolutions. What Austin did not do, because his organising unit was the founding intention rather than the constitutional text’s internal arrangement, was map the provisions as a working system with separable parts. Marc Galanter’s Competing Equalities: Law and the Backward Classes in India, published in 1984, came closer to a system view: Galanter named the whole field compensatory discrimination and traced its programmes with a lawyer’s precision. Yet Galanter’s units of analysis were programmes and beneficiary categories, not the textual layers that authorise them. The three-layered reading defended here takes the text’s own arrangement as the evidence: the Assembly put its anti-discrimination guarantees, its employment preferences and its legislative reservations in three different Parts of the Constitution, with three different enforcement logics, and that placement is the framework’s most under-read fact.

The layers differ first in what kind of norm each one is. The rights-protective provisions are prohibitions and abolitions addressed to the State and, in the case of Article 17, to everyone: they say what must not be done, and Articles 15 and 17 are enforceable in court under Article 32 and Article 226. The services-affirmative provisions are enabling clauses drafted as exceptions-shaped permissions inside equality guarantees: they say what the State may do, and they confer no individual right to a reserved post, a point the Supreme Court has repeated from the 1960s onward. The representational provisions are neither prohibitions nor permissions but direct constitutional commands: seats stand reserved because Article 330 says so, a commission exists because Article 338 says so, and the Union’s duty toward Scheduled Areas runs because Article 339 says so. A prohibition, a permission and a command are three different instruments. A framework that holds all three, each aimed at the same underlying condition of caste hierarchy, is not a pile of concessions extracted clause by clause. It is a design with a division of labour, and the sections that follow take the design apart layer by layer before testing it against the seventy years of amendment and litigation that have worked on it.

Where in the Constitution does reservation actually live?

Reservation lives in three places at once. Articles 15 and 16 carry the rights and services clauses inside Part III, Article 46 carries the protective directive inside Part IV, and Articles 330 to 342 carry seats, commissions and lists inside Part XVI, the part headed Special Provisions Relating to Certain Classes. The text never gathers them under one heading.

The distribution was not an accident of drafting sequence. The fundamental rights clauses were debated in November 1948, the directive in the same month, and the Part XVI block in June, August and September 1949, as the Assembly worked through distinct questions at distinct sittings: what the State must never do, what the State should aim at, and what institutional machinery the settlement required. The Assembly’s calendar, preserved in the Constituent Assembly Debates, is itself the first piece of evidence for the layered reading. The framers did not write one reservation clause and scatter copies of it. They answered three different questions in three different sittings, and the answers became three layers.

The Founding Draft: Three Decisions Taken in Three Sittings

The under-cited primary source for this framework is the Constituent Assembly Debates, Volume VII, which records the November and December 1948 sittings at which the equality code was drafted. On 29 November 1948 the Assembly debated Draft Article 9, the text that became Article 15, at pages 650 to 664 of that volume. The draft prohibited discrimination against any citizen on grounds only of religion, race, caste, sex, place of birth or any of them, and its second clause opened shops, public restaurants, hotels, places of public entertainment, wells, tanks, bathing ghats, roads and places of public resort to every citizen. What Draft Article 9 did not contain was any enabling clause for special provisions in favour of backward classes. Members speaking for the Depressed Classes pressed the omission. The Drafting Committee’s answer, delivered by B.R. Ambedkar as its chairman, was that the services clause, Draft Article 10, carried the enabling power for appointments, and that the directive principles carried the broader charge of advancement. The Assembly accepted that distribution on 29 November 1948, and it is why Article 15 stood without clause (4) until Parliament inserted it in 1951.

Later the same day, at pages 664 to 669, the Assembly debated Draft Article 11, which became Article 17. Naziruddin Ahmad moved to substitute a text that defined the wrong by its ground, proposing that no one be treated as an untouchable on account of religion or caste. The amendment was negatived. The Assembly instead kept the word untouchability in quotation marks, undefined, and abolished its practice in any form, leaving the content of the term to be filled by the historical practice it named. Muniswamy Pillai, speaking from the Madras Presidency’s long experience of temple-entry and civic-rights agitation, supported the abolition in the draft’s own terms. The decision to leave untouchability undefined was a deliberate drafting choice, and it mattered for the layered architecture: the rights-protective layer’s hardest clause was written as a prohibition of a practice rather than a definition of a status, which is why its enforcement could later be carried by ordinary penal legislation, the Untouchability (Offences) Act of 1955, without any return to the constitutional text.

On 30 November 1948, at pages 672 to 704 of Volume VII, the Assembly debated Draft Article 10, which became Article 16. The contested words were those of clause (4): the State’s power to reserve appointments or posts for any backward class of citizens not adequately represented in the services. H.J. Khandekar and other members pressed for the Scheduled Castes to be named in the clause itself. Ambedkar resisted a closed list, defending the general expression backward class of citizens on the ground that a constitutional text had to cover communities the Assembly could not enumerate in advance. The general wording survived, and it is the hinge on which the whole later history turns, because Article 16(4)’s open texture is what allowed the Mandal Commission to argue, thirty years later, that caste could serve as the starting point for identifying a backward class. A week earlier, on 23 November 1948, at pages 552 to 553, the Assembly had taken up Draft Article 37, which became Article 46, and placed the special-care directive for the educational and economic interests of the weaker sections, the Scheduled Castes and the Scheduled Tribes named expressly, among the non-justiciable principles. The rights layer and the directive layer were thus drafted within days of each other, by the same hands, as instruments of deliberately different force.

The representational layer was drafted later and separately. Draft Article 301, which became Article 340, was debated on 16 June 1949 at pages 943 to 948 of Volume VIII, giving the President power to appoint a commission to investigate the conditions of backward classes. Draft Article 292, which became Article 330, was debated on 23 and 24 August 1949, Draft Article 294, which became Article 332, on 24 August 1949, and Draft Article 295A, which became Article 334, on 24 and 25 August 1949, the record running from page 632 of Volume VIII into pages 637 to 698 of Volume IX. The inheritance those August sittings carried was the Poona Pact of 24 September 1932, examined in this series as the 1932 Poona Pact’s architectural inheritance: Ramsay MacDonald’s Communal Award of 16 August 1932 had allotted the Depressed Classes 71 provincial seats through separate electorates, Gandhi’s fast at Yerwada prison from 20 September 1932 had forced the renegotiation, and the Pact’s settlement, 148 reserved seats inside joint electorates, became the template Article 330 later nationalised. Ambedkar, Gandhi and Patel worked these questions as a triangle of mutual constraint, the geometry set out in the founding triangulation, and the layered framework is unintelligible without that geometry, because each layer answers to a different corner of it: abolition answered Gandhi’s moral demand, services reservation answered Ambedkar’s institutional demand, and legislative reservation answered the electoral arithmetic both men accepted at Poona.

The representational device itself had already been test-run under colonial law. The Government of India Act of 1935 wrote the Poona Pact settlement into its schedules, reserving seats for the Scheduled Castes in the provincial legislatures to be filled through joint electorates with a preliminary panel election among Depressed Class voters, and the provincial elections of 1937 were conducted on that design. The Assembly’s August 1949 sittings were therefore not inventing a mechanism but constitutionalising a working one, transferring it from an imperial statute, revocable in London, into the fundamental law of a republic, and raising its level from the provinces to the Union’s own legislature. The transfer matters to the layered reading because it explains the representational layer’s finished, almost untroubled drafting: alone among the three layers, Layer Three arrived with its design already settled by seventeen years of operation, and the debates of August 1949 could concentrate on duration and proportion rather than on principle.

Draft Articles 300A and 300B, which became Articles 341 and 342, the presidential list-making powers for Scheduled Castes and Scheduled Tribes, were debated on 17 September 1949 at pages 1638 to 1642 of Volume IX, and Draft Articles 296 and 299, which became Articles 335 and 338, returned for revision on 14 October 1949 in Volume X. The Assembly, in other words, revisited the services safeguard and the oversight office after the main structure stood, adjusting the machinery rather than the principle. The founding record therefore supports a precise claim about authorial intent, anchored in the rights-development founding tension that the Assembly never resolved and never tried to resolve: the Assembly built a framework in which protection from discrimination, preference in services and presence in legislatures were separate instruments, separately justified, and separately revisable. The October 1949 return to Draft Articles 296 and 299 rewards a closer look, because it shows the Assembly strengthening the framework’s machinery after its principles were beyond argument. When the two clauses came back on 14 October 1949, in the sittings recorded across pages 229 to 265 of Volume X, the Assembly settled the claims clause in its final form, the efficiency condition expressed as a consistency requirement rather than a ceiling, and it rebuilt the Special Officer provision into a reporting institution: the officer’s reports were to be laid before Parliament, and copies concerning a state forwarded to that state, so that the oversight of Layer Three would generate a public, legislative record rather than private advice to the executive. The first holder of the office, L.M. Shrikant, used exactly that channel, and his annual reports of the 1950s, documenting unfilled quotas, evasive state returns and the persistence of prohibited practices district by district, became the framework’s first audit trail. The episode carries a general lesson for reading the 1949 text: where the founders wanted a provision to sleep, they wrote a permission; where they wanted one to work, they attached a reporting duty, a deadline or a list. Layer Three is dense with such attachments, and their density is the signature of intent.

The fullest statement of the rights environment inside which the first of those instruments operates is the constitutional rights framework, the drafting history of the Part III code itself.

The raw material for those sittings had been laid two years earlier. The Advisory Committee on Fundamental Rights, Minorities and Tribal and Excluded Areas, chaired by Vallabhbhai Patel, received in 1947 a memorandum from Ambedkar, States and Minorities, which demanded for the Scheduled Castes a complete settlement in the future constitution: reserved seats in legislatures, a fixed share in the public services, and safeguards administered as constitutional rights rather than governmental favours. The memorandum’s architecture is recognisably the one the Assembly eventually built, protection, services and seats treated as three distinct instruments requiring three distinct clauses, and its presence in the record rebuts the notion that the layers were an accident of committee sequencing. They were argued for, as layers, before the drafting began. The same committee process also fixed the framework’s outer boundary: in May 1949 the Assembly, acting on the Advisory Committee’s recommendation, withdrew the reserved seats it had provisionally accepted for religious minorities in 1947, confining the representational layer to the Scheduled Castes and Scheduled Tribes and their Anglo-Indian exception. The framework’s beneficiary grammar was thereby set as caste and tribe rather than religion, a boundary whose consequences the closing sections of this article record.

Why does Article 17 stand among the fundamental rights instead of the directive principles?

Article 17 stands in Part III because the Assembly treated untouchability as a practice to be forbidden outright, not a goal to be pursued by policy. Placed among enforceable rights, its abolition bound private actors as well as the State, and its breach became an offence punishable by law rather than a failure of governance.

The drafting asymmetry is the point. Article 46 shows that the Assembly knew how to write a protective aspiration when it wanted one: the special care for weaker sections was made expressly non-enforceable, a charge on policy rather than a right in court. Article 17 shows the opposite choice taken on the same subject matter in the same month. The Assembly’s members did not lack the vocabulary of gradualism; they used it for education and economic interest, and withheld it from untouchability. That contrast, between Draft Article 37 and Draft Article 11 debated six days apart, is the cleanest evidence in Volume VII that the layers were distinguished on purpose. A constitution that meant its caste provisions to be one undifferentiated gesture would not have split them between a court-enforceable abolition and a non-enforceable directive. The split is the architecture in miniature.

Layer One: Rights-Protective Provisions and the Logic of Prevention

The first layer opens with the equality code’s commanding sentence. Article 15(1) provides that the State shall not discriminate against any citizen on grounds only of religion, race, caste, sex, place of birth or any of them. The word only carries the doctrine: the prohibition strikes at discrimination whose sole ground is a listed attribute, and the framers’ inclusion of caste among the grounds placed the institution they most intended to disable inside the Constitution’s first equality sentence. Article 15(2) then reaches past the State into civil life, providing that no citizen shall, on the same grounds, face any disability or restriction in access to shops, public restaurants, hotels and places of public entertainment, or in the use of wells, tanks, bathing ghats, roads and places of public resort maintained from State funds or dedicated to public use. The wells and bathing ghats are not quaint detail. They name the exact sites at which caste exclusion had been enforced for centuries, and their presence in a fundamental right made the everyday geography of humiliation a constitutional subject.

Article 17 is the layer’s centre of gravity, and its text deserves quotation in full: ‘Untouchability’ is abolished and its practice in any form is forbidden. The enforcement of any disability arising out of ‘Untouchability’ shall be an offence punishable in accordance with law. Three features of that sentence repay attention. It abolishes a practice rather than regulating a status, which is why no definition is supplied. It binds the whole world, not only the State, making it one of the very few fundamental rights that runs horizontally against private persons. And it commands penal enforcement, which Parliament executed through the Untouchability (Offences) Act of 1955, in force from 1 June 1955, and then strengthened through the 1976 amendment that renamed the statute the Protection of Civil Rights Act and sharpened its punishments, in force from 19 November 1976. The Scheduled Castes and the Scheduled Tribes (Prevention of Atrocities) Act of 1989, in force from 30 January 1990, added a further penal storey addressed to atrocities against members of the two constitutionally listed communities. The enforcement chain from constitutional abolition to penal statute to atrocities legislation is Layer One working exactly as drafted: the Constitution forbids, and ordinary law punishes.

Layer One’s clauses had been rehearsed in the two decades before the Assembly met, and the rehearsal explains the specificity of Article 15(2)’s list of wells, tanks, bathing ghats and roads. The Vaikom Satyagraha of 1924 and 1925, in the princely state of Travancore, was fought over exactly such a road: the public roads skirting the Vaikom Mahadeva temple, from which avarna Hindus were barred. The Guruvayur Satyagraha of 1931 and 1932 pressed the adjoining claim, entry into the temple itself. The breakthrough came from princely legislation rather than agitation alone: on 12 November 1936 the Maharaja of Travancore, Chithira Thirunal Balarama Varma, issued the Temple Entry Proclamation, opening every temple under the state’s control to all Hindu worshippers regardless of caste, and Madras followed with temple-entry authorisation legislation in 1947, on the initiative of the Congress ministry of T. Prakasam and his successor O.P. Ramaswamy Reddiar, the campaign in both provinces carried by Congress and Self-Respect activists over two decades. When the Drafting Committee wrote wells, tanks, bathing ghats and roads into a fundamental right in 1948, it was constitutionalising the satyagraha generation’s list of contested sites, converting victories won province by province into a uniform national prohibition. The episode is Layer One’s prehistory, and it confirms the layer’s character from before the founding: prevention first, site by site, against named exclusions, with distribution left to other instruments entirely.

Article 46 completes the layer from the directive side. Its text directs that the State shall promote with special care the educational and economic interests of the weaker sections of the people, and, in particular, of the Scheduled Castes and the Scheduled Tribes, and shall protect them from social injustice and all forms of exploitation. Because Article 37 makes the directive principles non-enforceable in court, Article 46 cannot be sued upon. That weakness is structural, not accidental, and the layered reading explains it: the directive states the protective purpose of the entire framework at the level of policy, while the enforceable instruments of that purpose are located elsewhere, in Article 15’s prohibitions, Article 17’s abolition, and the affirmative and representational machinery of the other two layers. Where a single-clause reading sees Article 46 as a pious wish trailing two strong rights, the layered reading sees a purpose clause deliberately placed outside the courts so that the framework’s aims could steer legislation and administration without freezing policy into litigation.

The layer’s logic can be stated precisely. Layer One is preventive and negative. It removes legal and civic disabilities: it stops the State from discriminating, opens public space to every citizen, criminalises the enforcement of untouchability, and charges policy with protecting the weakest from injustice and exploitation. What it does not do, anywhere, is distribute a single post, seat or admission. Distribution belongs to the second and third layers, and the founders’ discipline in keeping prevention and distribution in different provisions, with different remedies, is the framework’s first architectural fact. Confusion on this point has real consequences, because a litigant who asks Layer One to deliver a quota is asking a prohibition to do a permission’s work, and the Supreme Court’s earliest reservation judgment turned on exactly that confusion.

Article 15 itself records the accretive method in miniature, and the miniature is worth pausing on because it shows the framework’s clauses being added one permission at a time. As enacted in 1950 the article carried clause (3), an original provision that nothing in the article prevents the State from making any special provision for women and children, a protective permission the Assembly placed in the text from the start. Clause (4) arrived only in 1951, after Champakam exposed its absence, and was confined to backward classes, Scheduled Castes and Scheduled Tribes in the field of advancement, including education. Clause (5) arrived in 2006, after Inamdar, to carry the same beneficiaries into private educational institutions. Clause (6) arrived in 2019, carrying a new beneficiary defined by economic weakness. One article, four enabling clauses, four separate constitutional moments spread across sixty-nine years, each answering a distinct judicial or political pressure and none disturbing the prohibition in clause (1) with which the article opens. A reader who doubts that the Constitution’s equality provisions form a worked structure rather than a found one can be answered from this single article’s sedimentation: the deposits are dated, and each stratum is legible.

The First Test: Champakam Dorairajan and the First Amendment

The test arrived within fifteen months of the Constitution’s commencement. Madras Presidency’s communal government order, a distributive scheme descending from the Justice Party era, apportioned admissions to the state’s medical and engineering colleges by community: of every fourteen seats, six went to Non-Brahmin Hindus, two to Backward Hindus, two to Brahmins, two to Harijans, one to Anglo-Indians and Indian Christians, and one to Muslims. Champakam Dorairajan, a Brahmin applicant denied admission under that arithmetic, challenged the order in the Madras High Court and then in the Supreme Court under Articles 15(1) and 29(2). On 9 April 1951, in State of Madras v. Srimathi Champakam Dorairajan, AIR 1951 SC 226, a seven-judge bench struck the order down. The Court held that the communal allocation discriminated on grounds of caste and religion, that Article 46’s directive could not override the fundamental rights because the directive principles were subordinate to them, and that the Constitution as drafted contained no clause authorising the State to make special provision for backward classes in education. The judgment was, on the text as it then stood, correct, and that is the point the layered reading preserves: the Court read Layer One exactly as the Assembly had enacted it, as prevention without a distributive valve, and found the valve missing.

Parliament’s response is the strongest single piece of evidence that the layers are revisable separately. Jawaharlal Nehru’s government carried the Constitution (First Amendment) Act through the Provisional Parliament in May and June 1951, and the amendment received presidential assent on 18 June 1951. Among its changes, the one that matters here inserted clause (4) into Article 15, providing that nothing in Article 15 or in Article 29(2) shall prevent the State from making 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 repair was surgical. Parliament did not touch Article 16’s services clause, which already carried its enabling power, and did not touch Part XVI, whose seat reservations stood on their own text. It amended exactly the layer that the judgment had exposed, inserting a distributive permission into the rights-protective layer’s education flank while leaving the layer’s prohibitions intact. The episode, treated in full in the First Amendment as founding-era repair, establishes the pattern the whole later history repeats: a judicial reading reveals a gap at one layer, and a constitutional amendment closes that gap at that layer alone.

The judgment also fixed the hierarchy that the layered reading must respect. By holding the directive principles subordinate to the fundamental rights, Champakam confirmed that Article 46’s protective charge could inspire legislation but could not itself authorise what Article 15 forbade. The relationship between the two Parts is the subject of the directive-principles-versus-fundamental-rights hierarchy, and the settlement reached there governed until the Minerva Mills judgment of 31 July 1980 recast the two Parts as complementary halves of one commitment. For the reservation framework, the durable lesson of 1951 is simpler. The architecture has load-bearing distinctions between what is forbidden, what is permitted and what is directed, and a court or a Parliament that ignores the distinctions produces either a struck-down order or a constitutional amendment. The framework’s designers, and its first repairers, understood it as a structure. Its readers should do the same.

Layer Two: Services-Affirmative Provisions and the Logic of Permission

The second layer sits inside the Constitution’s second equality sentence. Article 16(1) guarantees equality of opportunity for all citizens in matters relating to employment or appointment to any office under the State. Article 16(4) then provides that nothing in Article 16 shall prevent the State from making any provision for the reservation of 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 under the State. Every phrase of that clause has been litigated for seventy years. The clause is enabling, not mandatory: it empowers governments without commanding them. Its beneficiary unit is a backward class of citizens, a formulation the Assembly chose on 30 November 1948 over demands to name the Scheduled Castes expressly. Its trigger is the State’s opinion on adequacy of representation, which made representation, rather than population share alone, the clause’s working test. And its field is services under the State, the vast apparatus of public employment that the founding generation regarded, correctly, as the principal escalator of status and security available to communities excluded from land, capital and the professions.

Article 335 is the layer’s governor, and its late drafting history shows the Assembly calibrating it with care. Draft Article 296 was debated on 26 August 1949 and returned on 14 October 1949, in Volume X of the Debates, where its wording was settled: the claims of the members of the Scheduled Castes and the Scheduled Tribes shall be taken into consideration, consistently with the maintenance of efficiency of administration, in the making of appointments to services and posts under the Union or a State. The efficiency phrase is often misread as a grudging concession pasted onto a welfare clause. In the layered reading it is a design element of Layer Two itself: the services-affirmative power was drafted from the start as a balanced instrument, in which preferential appointment and administrative efficiency are constitutionally co-present, each limiting the other’s absolutist reading. Ambedkar, defending the clause in the Assembly, treated the safeguard as the price of making the preference durable, a formulation that anticipated by four decades the Supreme Court’s insistence in Indra Sawhney that Article 335’s balance travels with Article 16(4) wherever the latter goes.

The early case law worked out the layer’s tolerances. In M.R. Balaji v. State of Mysore, AIR 1963 SC 649, a five-judge bench considered Mysore’s order reserving 68 percent of seats in medical and engineering colleges and struck it down, holding that caste could be a relevant factor in identifying a backward class but could not be the sole criterion, that the interests of the community had to be balanced against the general interest, and that reservations should broadly remain below 50 percent. In T. Devadasan v. Union of India, AIR 1964 SC 179, the Court struck down the carry-forward rule under which unfilled reserved vacancies accumulated across years until, in the year in question, 64.4 percent of the filled posts went to reserved candidates; the majority treated the accumulated figure as a fraud on the 50 percent discipline. Then, in State of Kerala v. N.M. Thomas, AIR 1976 SC 490, a seven-judge bench changed the layer’s interpretive centre of gravity. The majority upheld a Kerala rule giving Scheduled Caste and Scheduled Tribe employees extra time to pass promotion tests, and the reasoning of judges including Ray CJ, Mathew J and Krishna Iyer J recast Article 16(4) not as an exception to equality but as an emphatic statement of what equality of opportunity already contained: a substantive, result-sensitive equality that permits classification in favour of the historically disadvantaged. The layered reading absorbs N.M. Thomas naturally. The Court was not discovering a new provision; it was re-describing the relationship between Layer Two’s permission and the equality code around it, holding that the permission is internal to the code rather than a breach in it.

The layer’s operating machinery deserves one further specification, because adequacy of representation is a test that must be administered post by post. Indra Sawhney had already held that adequacy is not proportionality: a class need not be represented in exact population ratio, and reservation is not a proportional-representation device. In R.K. Sabharwal v. State of Punjab, (1995) 2 SCC 745, decided on 10 February 1995, a five-judge bench settled the machinery, holding that reservation operates on posts in a cadre rather than on vacancies as they arise, so that the prescribed percentage is measured against the cadre’s total strength, and once the roster has delivered the percentage, reserved appointment pauses until a vacancy in the reserved share occurs. The Department of Personnel and Training’s office memorandum of 2 July 1997 replaced the older vacancy-based roster with the post-based roster across the Union services, and the 200-point roster became the layer’s everyday instrument, the grid through which Article 16(4)’s abstract permission is converted into a sequence of individual appointments. Administrative detail of this order rarely enters constitutional commentary, but it belongs in an architectural account: Layer Two is the only layer whose permissions must be re-executed at every single appointment, and the roster is where the architecture meets the file.

Why does the Constitution say backward class where every listener understood caste?

The Constitution says backward class because Ambedkar defended the general phrase on 30 November 1948 against amendments that would have named the Scheduled Castes alone, arguing that a constitutional clause had to reach every community whose backwardness the future might register. The open wording is what later allowed caste to be argued, rather than assumed, as evidence of class.

That argumentative openness has structured the layer ever since. Because the text says class, every exercise of Article 16(4) has had to present its beneficiary as a class and defend the identification with evidence, which is why the commissions appointed under Article 340 became evidentiary machines, why Indra Sawhney could hold both that caste is a permissible starting point and that a caste is not automatically a class, and why the creamy-layer doctrine could be framed as an internal correction of the class inquiry rather than a political trimming of its results. The Assembly’s lexical choice in November 1948, one word in one clause, generated the framework’s entire later jurisprudence of identification. Few single words in the Constitution have done more work.

The Mandal Moment Inside Layer Two

The layer’s largest peacetime expansion came through the Constitution’s own investigative route. Article 340, debated on 16 June 1949, empowers the President to appoint a commission to investigate the conditions of socially and educationally backward classes and recommend steps for their improvement. The first such body, the Backward Classes Commission chaired by Kaka Kalelkar, appointed on 29 January 1953, reported on 30 March 1955 and was shelved amid its chairman’s own doubts. The second, the Backward Classes Commission chaired by B.P. Mandal, was appointed on 1 January 1979 by the Janata government of Morarji Desai. Its report, submitted on 31 December 1980, identified 3,743 castes as socially and educationally backward, estimated them at 52 percent of the population, and recommended 27 percent reservation in central services, the figure chosen because the existing 22.5 percent for Scheduled Castes and Scheduled Tribes, added to 27 percent, stopped just below the 50 percent ceiling the courts had announced. The Mandal report is best understood, in the vocabulary of this article, as a Layer Two specification exercise: it supplied the evidentiary content that Article 16(4)’s phrase backward class of citizens had been waiting for at the level of the Union.

A decade separated the report from its execution. On 7 August 1990 Prime Minister V.P. Singh announced in Parliament that the report would be implemented, and the Office Memorandum of 13 August 1990 reserved 27 percent of vacancies in civil posts and services under the Government of India, and in public sector undertakings, for the socially and educationally backward classes. The political convulsion that followed, and the adjudication among the rival readings of what Mandal meant, belongs to the pillar hub’s treatment of the three-reading adjudication framework, which this article takes as given. The constitutional question travelled to a nine-judge bench, and on 16 November 1992, in Indra Sawhney v. Union of India, AIR 1993 SC 477, the Supreme Court upheld the 27 percent reservation by a majority of six to three, reading the layered architecture whole. The holdings can be stated by layer. For Layer Two, the Court held that Article 16(4) permits reservation for Other Backward Classes, that caste can be the starting point for identifying a backward class, that the more advanced sections of a backward class, the creamy layer, must be excluded, that reservation should ordinarily not exceed 50 percent, that the 10 percent quota for economically backward sections of the unreserved classes announced in 1991 was invalid, and that Article 16(4) does not permit reservation in promotions. For the framework as a whole, the Court treated Article 335’s efficiency condition as a standing qualification on preferential appointment, and it located the entire reservation enterprise inside the equality code rather than outside it, completing the interpretive turn that N.M. Thomas had begun in 1976.

Why did the framework’s working limits come from courtrooms rather than from Parliament?

The working limits came from courtrooms because the constitutional text fixed the permissions and left their tolerances open: Article 16 states no ceiling, no exclusion for a beneficiary class’s advanced sections, and no rule for promotions. Into that silence the Supreme Court wrote the 50 percent ceiling, the creamy-layer exclusion and the promotions bar between 1963 and 1992.

Parliament’s part in the exchange should not be understated, because it is the layered architecture that made the exchange orderly. When the judiciary fixed a tolerance that the political branches wished to alter, the response was again an amendment aimed at the precise layer in dispute: the First Amendment at Layer One in 1951, and, after Indra Sawhney barred reservation in promotions, a sequence of amendments at Layer Two between 1995 and 2001. Courts proposed tolerances; amendments disposed. The pattern is not judicial supremacy or parliamentary supremacy but a structured conversation between the two, conducted clause by clause, and it is visible only when the provisions are read as the layers they form. An article-by-article account can narrate each exchange. Only the layered account explains why every exchange stayed inside its own layer, leaving the untouched layers to continue operating exactly as before.

One administrative sequel of the judgment belongs in this account because it shows Layer Two generating its own registry. Indra Sawhney had directed that the creamy layer be identified by expert examination, and the Union government appointed a committee under Justice Ram Nandan Prasad, a retired judge of the Patna High Court, in February 1993 to frame the criteria. The committee’s work issued in the Department of Personnel and Training’s office memorandum of 8 September 1993, which defined the creamy layer through occupational and income tests and published the first central list of Other Backward Classes for Union services and admissions. The income ceiling in that memorandum, set first at one lakh rupees a year, was revised upward in 2004, 2008, 2013 and 2017, reaching eight lakh rupees, each revision a small Layer Two maintenance operation conducted entirely by executive instruction under the judgment’s authority. The National Commission for Backward Classes Act of 1993 gave the beneficiary-identification process a statutory adviser in the same year, two decades before Article 338B gave the commission constitutional form. Identification, exclusion and revision, the unglamorous third of the framework’s life, has run since 1993 through this machinery, and it runs on Layer Two’s logic throughout: the class is defined by evidence, the advanced sections are filtered out, and the list is republished.

Layer Two Extended: The Amendment Sequence of 1995 to 2001

Indra Sawhney’s holding that Article 16(4) does not authorise reservation in promotions lasted thirty-one months as operative law. The Constitution (Seventy-seventh Amendment) Act of 1995, in force from 17 June 1995, inserted clause (4A) into Article 16, providing for reservation in matters of promotion, with consequential seniority, to posts in the services under the State in favour of the Scheduled Castes and the Scheduled Tribes which, in the State’s opinion, are not adequately represented. The amendment is a model of the framework’s layer discipline: it restored for two named communities, at the promotion stage only, a power the judgment had denied to the enabling clause generally, and it left the judgment’s other tolerances, the ceiling and the creamy layer, entirely undisturbed. Four further amendments completed the sequence. The Eighty-first Amendment of 2000, in force from 9 June 2000, inserted clause (4B), permitting unfilled reserved vacancies of a year to be carried forward as a separate class of vacancies, outside that year’s 50 percent computation, thereby reversing the practical effect of Devadasan within a stated limit. The Eighty-second Amendment of 2000, in force from 8 September 2000, added a proviso to Article 335 allowing relaxation in qualifying marks and lowering of standards of evaluation for Scheduled Caste and Scheduled Tribe candidates in promotion. The Eighty-fifth Amendment of 2001 then gave the consequential-seniority element of clause (4A) retrospective effect from 17 June 1995, settling the seniority contests the 1995 amendment had generated.

The Supreme Court audited the whole sequence in M. Nagaraj v. Union of India, (2006) 8 SCC 212, decided on 19 October 2006. A five-judge bench upheld all four amendments, but conditioned their operation: the State must show, with quantifiable data, the backwardness of the class, its inadequate representation in the services, and the maintenance of administrative efficiency under Article 335, and the creamy-layer exclusion applies to Scheduled Castes and Scheduled Tribes in promotion as it does to Other Backward Classes in appointment. In Jarnail Singh v. Lachhmi Narain Gupta, (2018) 10 SCC 396, decided on 26 September 2018, a five-judge bench refined Nagaraj, holding that the State need not collect quantifiable data to prove the backwardness of Scheduled Castes and Scheduled Tribes, whose listing under Articles 341 and 342 already embodies that determination, while retaining the creamy-layer condition and the representation and efficiency inquiries. Read through the layered lens, Nagaraj and Jarnail Singh are not new law piled on old; they are the judiciary specifying the operating conditions of Layer Two’s promotion machinery, just as Balaji and Indra Sawhney had specified the conditions of its appointment machinery. The layer absorbed four constitutional amendments and two Constitution Bench audits in twenty-three years without its underlying clause, Article 16(4), changing a single word. That absorptive capacity is what an architecture, as opposed to a rule, possesses.

Layer Three: Representational Provisions and the Logic of Presence

The third layer answers a question the first two cannot: who sits in the room where laws are made. Article 330 reserves seats for the Scheduled Castes and the Scheduled Tribes in the House of the People, and Article 332 does the same in the Legislative Assembly of every State, the number of reserved seats in each case bearing, as nearly as may be, the same proportion to the total seats as the community’s population bears to the population of the state or constituency bloc. Under the delimitation in force, 84 of the 543 elected seats in the Lok Sabha are reserved for Scheduled Castes and 47 for Scheduled Tribes, 131 seats whose occupants are chosen by joint electorates of all voters in reserved constituencies. The joint-electorate design is the Poona Pact’s direct bequest: the seat is guaranteed to the community, the mandate comes from the whole constituency, and the representative therefore answers to a mixed electorate, an arrangement the Assembly adopted in August 1949 in conscious descent from the 1932 settlement and in conscious rejection of the Communal Award’s separate electorates.

Article 334 gave this layer the framework’s only clock. As enacted, it provided that the reservation of seats and the special representation of the Anglo-Indian community should cease after ten years from the Constitution’s commencement, a period the Assembly debated on 24 and 25 August 1949 in the expectation, voiced by Ambedkar among others, that a decade of guaranteed presence might suffice to normalise representation. Members who doubted that ten years would suffice were answered from the Drafting Committee’s side that the period was a constitutional expectation, not a prophecy, and that a future Parliament persuaded of continued need could extend it by amendment, as the text of Draft Article 295A itself contemplated by fixing a date rather than a principle. The expectation failed, and the clock has been reset seven times: by the Eighth Amendment to 1970, the Twenty-third to 1980, the Forty-fifth to 1990, the Sixty-second to 2000, the Seventy-ninth to 2010, the Ninety-fifth to 2020, and the One Hundred and Fourth Amendment, in force from 25 January 2020, to 2030. The same One Hundred and Fourth Amendment let the Anglo-Indian nomination provisions of Articles 331 and 333 lapse, ending the special representation that Draft Article 295A had yoked to the seat reservations in a single sunset. The sunset’s career is a standing datum for the layered reading. Parliament has repeatedly extended Layer Three’s duration without reopening its design, treating the ten-year clause as a review mechanism rather than a terminal date, and each extension debate has been fought on the condition of representation, exactly the ground the layer exists to govern.

The layer’s second half is oversight machinery, and its institutional history mirrors the first layer’s growth. Article 338 as enacted created a single Special Officer for the Scheduled Castes and Scheduled Tribes, an office first held by L.M. Shrikant, whose annual reports to the President through the 1950s documented the distance between the text’s promises and district-level practice. The Sixty-fifth Amendment of 1990, in force from 12 March 1992, replaced the single officer with a multi-member National Commission for the Scheduled Castes and Scheduled Tribes, armed with the powers of a civil court for investigation. The Eighty-ninth Amendment of 2003, in force from 19 February 2004, divided the commission in two, leaving the National Commission for Scheduled Castes under Article 338 and creating the National Commission for Scheduled Tribes under the new Article 338A. The One Hundred and Second Amendment of 2018, in force from 15 August 2018, then inserted Article 338B, giving the National Commission for Backward Classes the same constitutional standing and the same investigative powers. Article 339 supplies the Union’s corresponding executive instrument: it empowers the President to appoint commissions on the administration of the Scheduled Areas and the welfare of the Scheduled Tribes, a power exercised in the commission chaired by U.N. Dhebar, appointed on 28 April 1960 and reporting in 1961, and again in the commission chaired by Dilip Singh Bhuria, appointed in 2002, and it empowers the Union to direct states on the drawing up and execution of schemes for tribal welfare. Beneath the whole representational layer sit Articles 341 and 342, debated on 17 September 1949, which assign the specification of the Scheduled Castes and Scheduled Tribes themselves to presidential order, variable thereafter only by Act of Parliament. The first orders of 1950, the Constitution (Scheduled Castes) Order of 6 September 1950 among them, thus became part of the constitutional furniture: the lists decide who the framework’s beneficiaries are, and the lists’ home in Part XVI, among the representational provisions, marks them as elements of the settlement’s machinery rather than entries in a welfare schedule.

The representational logic was later extended downward to the third tier of government, in a development that confirms the layer’s distinctness. The Seventy-third and Seventy-fourth Amendments of 1992 inserted Parts IX and IXA into the Constitution, and Articles 243D and 243T reserve seats for the Scheduled Castes and the Scheduled Tribes in panchayats and municipalities respectively, again in proportion to population, together with not less than one third of all seats for women, the reserved seats rotating among constituencies by law. The local reservation is not a borrowing from Articles 330 and 332 by analogy; it is the same constitutional device rebuilt at village and municipal scale, complete with its own proportionality rule and its own rotation machinery, and several states, Bihar first in 2006, raised the women’s share in local bodies to one half by ordinary legislation. A reader tracking the framework’s career will notice that when the republic wanted guaranteed presence in a new arena, it did not stretch Articles 15 or 16 to cover the case. It built another representational provision, because presence is Layer Three’s work and the Constitution’s grammar reserves that work to Layer Three’s instruments.

The Lists Beneath the Settlement: Specification as Constitutional Machinery

Every layer of the framework depends on a prior act of specification: someone must say which communities count as Scheduled Castes, which as Scheduled Tribes, and which classes count as backward. The Constitution assigns that act, for the two scheduled categories, to the President in the first instance and to Parliament thereafter. Article 341 empowers the President, after consultation with the Governor of a state, to specify by public notification the castes, races or tribes deemed to be Scheduled Castes in relation to that state, and reserves every later inclusion or exclusion to Parliament by law. Article 342 does the same for Scheduled Tribes. The first instruments were the Constitution (Scheduled Castes) Order and the Constitution (Scheduled Tribes) Order of 1950, and their state-by-state form explains a durable feature of the framework: the lists are territorial, so that a community scheduled in one state may be unscheduled in the next, and migration across a state line can carry a family out of its constitutional category. The lists are the framework’s registry office, and their placement in Part XVI, among the representational provisions rather than in Part III among the rights, records the founders’ judgment that specification is machinery, to be operated and revised politically, rather than a definition to be frozen into the rights text.

The Scheduled Castes Order of 6 September 1950 carried a condition with long consequences. Its third paragraph provided that no person professing a religion different from the Hindu religion shall be deemed a member of a Scheduled Caste, tying the category to the Hindu fold on the theory that untouchability was an institution of that fold. Parliament amended the condition by law in 1956 to include Sikhs and in 1990 to include Buddhists, after movements among Sikh and Buddhist converts, many of the latter following Ambedkar’s own conversion of 14 October 1956 at Nagpur, pressed the claim that the disabilities of caste survived conversion. Christians and Muslims of Dalit origin remain outside the category, and the exclusion has been challenged for decades on the argument, pressed at its strongest by converts’ organisations and examined by the National Commission for Religious and Linguistic Minorities in its report of 21 May 2007, that caste disability is a social fact that conversion does not erase; the counter-argument, also attributed to its holders in official reports, is that the category was framed for a specifically Hindu institution and that extension would dilute the settlement’s focus. The dispute is live and is recorded here as disputed. Its relevance to the layered reading is structural: the argument is conducted entirely at the level of the lists, through the machinery of Articles 341 and 342, which shows the registry layer performing exactly the revisable, political function the founders assigned it, absorbing a profound moral controversy without disturbing a single right or quota in the layers above.

For the backward classes, specification took the commission route rather than the presidential-order route, which is why Articles 340’s commissions, Kalelkar in 1953 and Mandal in 1978, carry such weight in the framework’s history, and why the Union and the states maintain parallel lists that do not coincide: a caste may stand in a state’s backward-classes list and not in the central list used for Union services and admissions, and litigation over list membership, usually turning on whether a community is genuinely the one the list names or a synonym subgroup of it, forms a steady stream of Layer Two adjudication. The Rohini Commission, appointed on 2 October 2017 under Article 340 to examine the distribution of the Other Backward Classes quota among the listed communities, belongs to the same machinery, an attempt to audit the registry from inside. Specification, in short, is not a preliminary to the framework. It is one of its working parts, and the founders’ decision to house it in Part XVI, revisable by ordinary law and order, is among the most consequential placement choices the layered reading brings into view.

The Framework in the Federal System: Two Governments, One Architecture

Because education and public employment are fields on which both the Union and the states legislate, the framework operates simultaneously at two levels of government, and the layers behave differently at each. The rights-protective layer binds every government identically: Article 15’s prohibitions and Article 17’s abolition run against the Union, the states and private persons without variation. The services-affirmative layer is exercised in parallel: the Union reserves in Union services and central institutions on the strength of the central lists, each state reserves in state services and state institutions on the strength of its own lists and its own assessment of adequacy of representation, and the two exercises can and do diverge, in beneficiary, in percentage and in doctrine, subject always to the judiciary’s tolerances. The representational layer divides by subject: Article 330 speaks for the Union’s legislature, Article 332 for the states’, and the commissions of Layer Three report to Presidents and Governors respectively. Federal divergence inside a single architecture is therefore a designed condition, and the friction it produces, most visibly when a caste is backward in one list and forward in another, is the friction of two legitimate operators working the same machine.

The percentages record the divergence faithfully. In Union services, the shares settled after the Mandal judgment stand at 15 percent for Scheduled Castes, 7.5 percent for Scheduled Tribes and 27 percent for Other Backward Classes in direct recruitment, tracking the communities’ population weights within the ceiling. The states answer to their own demography and politics: Tamil Nadu’s 69 percent, protected for a generation through the Ninth Schedule device examined earlier, is the extreme case, while Bihar’s settlement of the 1970s, built by Chief Minister Karpoori Thakur on the state’s own backward-classes inquiry, divided its quota between backward and extremely backward strata a decade before the Union adopted any OBC reservation at all. The layered reading absorbs this variation without strain, because adequacy of representation, the trigger written into Article 16(4), is a judgment the clause expressly leaves to each State, and the ceiling is a tolerance imposed on both levels alike. What the federal picture adds to the architectural claim is a test of load: a framework that can be worked by twenty-nine state governments and the Union at once, producing divergent settlements that courts nonetheless review under one set of layer tolerances, is not a metaphor imposed by scholars. It is the operating system the litigation assumes.

The state lists carry their own commission genealogies, of which Karnataka’s is the most documented. The state’s first backward-classes inquiry, the committee chaired by R. Nagan Gowda in 1961, preceded the Constitution’s own first commission report; the L.G. Havanur Commission’s report of 1975 reorganised the state’s categories and governed its quota for a decade; the T. Venkataswamy Commission’s report of 1986 was set aside amid protest; and the Justice O. Chinnappa Reddy Commission’s report of 1990 supplied the classification on which Karnataka’s settlement then rested. Four state inquiries in thirty years, each revising the registry while the state’s reservation percentages moved within the courts’ tolerances, is the registry function operating exactly as Articles 15(4) and 16(4) contemplate: the State forms an opinion on backwardness and adequacy, evidences it through a commission, and defends it in court. The Union’s two Article 340 commissions and the states’ dozens of inquiries together form a single evidentiary tradition, and its continuity is further proof that specification was designed into the architecture rather than improvised under political pressure.

The federal balance inside Layer Two was itself litigated to a settlement in the framework’s eighth decade. When Maharashtra’s Socially and Educationally Backward Classes Act of 2018 reserved up to 13 percent for the Maratha community, carrying the state’s total well past the ceiling, a five-judge bench struck the Act down in Jaishri Laxmanrao Patil v. Chief Minister, Maharashtra, on 5 May 2021, reaffirming Indra Sawhney’s 50 percent limit and finding no extraordinary circumstance to breach it. The same bench read the One Hundred and Second Amendment’s Article 342A as transferring the identification of socially and educationally backward classes for both Union and state purposes to the President and Parliament alone. Parliament answered within three months, in the single-layer repair pattern this article has tracked since 1951: the Constitution (One Hundred and Fifth Amendment) Act of 2021 restored the states’ power to prepare and maintain their own lists of backward classes, amending Articles 338B, 342A and 366 for that purpose. The episode is the layered architecture photographed mid-operation: a state overruns a tolerance, the Court enforces the tolerance and, in doing so, narrows the registry power; the amendment power is then used to restore the registry power while leaving the tolerance exactly where the Court put it.

The Education Flank: Clause (5), Private Institutions and the Minority Carve-Out

The First Amendment had repaired Article 15 for state action in education, but the expansion of private higher education reopened the flank. In P.A. Inamdar v. State of Maharashtra, (2005) 6 SCC 537, a seven-judge bench held that the State could not impose its reservation policy on unaided professional institutions, reasoning from the institutional autonomy protected by Articles 19(1)(g) and 30. Parliament replied within months, in the by then familiar pattern of a targeted amendment at the affected layer. The Constitution (Ninety-third Amendment) Act of 2005, in force from 20 January 2006, inserted clause (5) into Article 15, empowering the State to make special provision by law for the advancement of socially and educationally backward classes, Scheduled Castes and Scheduled Tribes in admission to educational institutions, including private institutions whether aided or unaided by the State, other than minority educational institutions protected by Article 30(1). The carve-out is itself a piece of layered design: the distributive permission stops at the boundary of another constitutional commitment, the minority right to establish and administer institutions, so that the amendment extends Layer One’s education flank without consuming a neighbouring right.

The operative legislation followed in the Central Educational Institutions (Reservation in Admission) Act of 2006, which reserved 15 percent of seats for Scheduled Castes, 7.5 percent for Scheduled Tribes and 27 percent for Other Backward Classes in centrally funded institutions, phased across three years with an expansion of total seats designed to protect unreserved admissions. In Ashoka Kumar Thakur v. Union of India, (2008) 6 SCC 1, decided on 10 April 2008, a five-judge bench upheld both the Ninety-third Amendment and the 2006 Act, applying the creamy-layer exclusion to admissions and directing periodic review of the backward-class lists. For the layered reading, the education flank’s forty-year path, from Champakam through the First Amendment to Inamdar, the Ninety-third Amendment and Ashoka Kumar Thakur, demonstrates the same mechanics as the services flank’s path through Indra Sawhney and the amendments of 1995 to 2001. A judicial tolerance is announced, a single clause is amended, the rest of the architecture continues untouched. Education and services are distinct policy fields, but constitutionally they are serviced by the same layered machine, and the machine’s serviceability is the strongest practical evidence that the layers are real: governments keep repairing the framework one layer at a time because that is how it was built.

Table 1: The Three-Layered-Provisions Framework

The argument of this article stands or falls on whether the three layers can be specified exactly, with their provisions, their governing logics and their amendment trajectories distinguished. Table 1, the three-layered-provisions framework, states the specification in citable form. Each row names one layer, the constitutional provisions that compose it, the distinct logic those provisions execute, and the trajectory of amendment and adjudication through which that layer, and that layer alone, has been revised since 1950. Readers who carry Table 1 into a lecture deck can open the resulting slides in an online presentation viewer without converting the file, which keeps the framework’s rows intact for citation.

Layer	Constitutional provisions	Governing logic	Amendment and adjudication trajectory
Layer One: rights-protective	Articles 15(1) and 15(2), 17, and 46, with 15(4) and 15(5) inserted into Article 15	Prevention: prohibiting discrimination, abolishing untouchability, and directing the State to protect weaker sections from injustice and exploitation	First Amendment of 1951 inserts 15(4) after Champakam Dorairajan; Ninety-third Amendment of 2005 inserts 15(5); enforcement carried by the Protection of Civil Rights Act and the 1989 Atrocities Act
Layer Two: services-affirmative	Articles 16(1) and 16(4), extended by 16(4A), 16(4B) and 16(6), governed by Article 335	Permission: authorising reservation of appointments and posts for inadequately represented backward classes, balanced against efficiency of administration	Seventy-seventh Amendment of 1995 (promotions), Eighty-first and Eighty-second of 2000 (carry-forward, Article 335 proviso), Eighty-fifth of 2001 (consequential seniority), One Hundred and Third of 2019 (economically weaker sections); read through Balaji, Indra Sawhney, Nagaraj and Jarnail Singh
Layer Three: representational	Articles 330, 332 and 334, the commissions at 338, 338A and 338B, the Union’s charge at 339, and the lists at 341 and 342	Presence: reserving legislative seats in proportion to population, time-limiting the reservation, and building commissions and lists to oversee and define the settlement	Seven extensions of the Article 334 period, the Seventh in the One Hundred and Fourth Amendment of 2019 carrying reservation to 2030 and ending Anglo-Indian nomination; Sixty-fifth Amendment of 1990, Eighty-ninth of 2003 and One Hundred and Second of 2018 building the commission line

Three features of Table 1 deserve emphasis. The first is that no provision migrates between rows: when the framework changed, the change registered inside one row, as the trajectory column shows, and the other rows’ provisions continued in their existing form. The second is that each row’s logic is grammatically distinct in the constitutional text itself, prohibition in the first, permission in the second, command in the third, a distinction drawn by the Drafting Committee’s verbs before any court interpreted them. The third is that the rows are mutually dependent without being mutually reducible. Layer Three’s reserved-seat legislators supply the political majorities that amend Layer Two; Layer One’s prohibitions supply the normative ground on which Layer Two’s permissions are defended; Layer Two’s services supply the administrative presence through which Layer One’s protections reach district offices. Dependence is not identity, and the framework’s coherence consists precisely in distinct instruments interlocking, not in a single instrument repeated.

The Layers in Interaction: Mandal, the Ceiling and the EWS Amendment

Layered systems are tested at their joints, and the framework’s three hardest cases all arose where one layer’s operation pressed on another’s tolerance. The Mandal implementation of 1990 was a Layer Two event, an exercise of Article 16(4) through the Article 340 commission route, but its defence in court drew on Layer One: counsel for the Union grounded the 27 percent reservation in the Constitution’s protective purpose, the Article 46 charge made argumentative flesh, and the majority in Indra Sawhney accepted the grounding while refusing to let purpose dissolve the services clause’s own disciplines. The ceiling the Court announced, 50 percent as the ordinary outer limit, is a tolerance of Layer Two, derived from Balaji’s 1963 balancing and stated as a rule of the services and admissions field. The same judgment’s distinction between vertical reservation, the quotas running between the large beneficiary categories, and horizontal reservation, the cuts for women, persons with disabilities and ex-servicemen that run across every vertical compartment, has since structured every reservation scheme in the country, and it is unintelligible except as a Layer Two doctrine: it governs how permissions stack inside appointments and admissions, and it has no application to Layer One’s prohibitions or Layer Three’s seats, where no comparable stacking exists.

Tamil Nadu supplied the stress test. The state’s reservation total reached 69 percent, sustained after 1993 by placing the Tamil Nadu Backward Classes, Scheduled Castes and Scheduled Tribes (Reservation of Seats in Educational Institutions and of Appointments or Posts in the Services under the State) Act of 1993 in the Ninth Schedule as Entry 257A, beyond ordinary judicial review as the schedule was then understood. In I.R. Coelho v. State of Tamil Nadu, (2007) 2 SCC 1, decided on 11 January 2007, a nine-judge bench held unanimously that laws placed in the Ninth Schedule after 24 April 1973, the date of the Kesavananda Bharati judgment, remain open to challenge where they damage the basic structure, including the equality code. Coelho matters to this article because it shows the framework’s tolerances being policed at the level of the Constitution’s own identity: a state may legislate within Layer Two’s permissions, and Parliament may schedule the legislation, but the layered equality architecture through which the legislation must pass is part of the basic structure that neither can amend away. The framework, on this showing, is not merely an organisational scheme imposed on the text by commentators. The Court itself treats the equality provisions as a structure with integrity conditions.

The Economically Weaker Sections amendment is the complication this article must answer rather than admire, because it is the strongest evidence against the layered reading. The Constitution (One Hundred and Third Amendment) Act of 2019, assented to on 12 January 2019 and in force from 14 January 2019, inserted clause (6) into both Article 15 and Article 16, authorising up to 10 percent reservation for economically weaker sections of citizens other than the classes covered by the existing clauses, in education and in appointments. The amendment’s criterion is economic, where every earlier layer had worked through caste-linked backwardness, and its exclusion of the Scheduled Castes, Scheduled Tribes and Other Backward Classes from the new benefit inverts the framework’s founding direction. Critics of the layered reading argue that an architecture which can absorb a provision contradicting its original animating principle was never an architecture at all, only a vocabulary. The amendment belongs to the post-2019 constitutional-amendment context in which Parliament used the amendment power expansively, and its challenge reached the Court as Janhit Abhiyan, addressed in the next section.

The amendment’s working criteria were supplied within days by executive instruction, in the by then standard manner of Layer Two’s administration. The Department of Personnel and Training’s office memorandum of 31 January 2019 fixed the benefit’s tests: a candidate whose family’s gross annual income from all sources falls below eight lakh rupees, whose family does not own five acres of agricultural land or more, a residential flat of one thousand square feet or more, or residential plots above the stated sizes in notified and other areas, and who does not belong to the classes covered by the existing reservations. However contested the criteria, their form is the framework’s form: an income and asset test administered through certificates, lists and verification, the registry habits of Layer Two applied to a beneficiary the founders did not name. The amendment’s defenders made exactly this point in Janhit Abhiyan, that the EWS provisions speak the architecture’s administrative language fluently, and the majority agreed.

The answer turns on a distinction the amendment itself observes: layer content against layer architecture. The One Hundred and Third Amendment changed what Layer Two contains, adding an economic criterion to the services and admissions permissions, and it correspondingly extended Layer One’s Article 15. It did not change how the framework is organised: the new benefit was drafted as a clause inside the existing articles, subjected by its defenders and its challengers alike to the existing layer’s doctrines of ceiling, classification and basic structure, and administered through the existing machinery of lists, commissions and review. A content amendment of this depth is a serious event, and the layered reading does not minimise it. What the reading denies is that new content abolishes structure. The EWS provisions entered the architecture through its doors, were argued in its vocabulary, and were judged by its tolerances. An architecture that admits a contested extension through its own procedures, and survives the argument about whether the extension belongs, is behaving as architectures behave.

What makes three separate layers a single framework?

Three features make the layers one framework: they address a single condition, caste hierarchy, through deliberately separated instruments; each supplies what the others lack, prohibition without distribution in the first, permission without presence in the second, presence without either in the third; and every amendment since 1951 has revised one layer while relying on the other two.

The Case-Law Line as the Architecture’s Interpreter

Read in sequence, the great reservation judgments form a single interpretive line, and the layered reading explains the line’s shape better than the line explains itself. The sequence opens with Champakam Dorairajan on 9 April 1951, where the Court enforced Layer One’s prohibitions as written and reported the missing permission back to Parliament. It continues with Balaji on 28 September 1962, which supplied Layer Two’s first tolerances, caste as factor rather than sole criterion, a broad ceiling of 50 percent, and the balance between the claims of backward classes and the general community. Devadasan in 1964 policed the carry-forward device; N.M. Thomas in 1976 relocated Article 16(4) inside the equality code as its emphatic application rather than its exception; Indra Sawhney on 16 November 1992 systematised the tolerances for the Mandal era and barred the promotion stage; Nagaraj on 19 October 2006 audited the amendment sequence that answered the promotions bar and attached evidentiary conditions; Jarnail Singh on 26 September 2018 corrected Nagaraj’s evidentiary demand for the two listed communities; and the line arrives at Janhit Abhiyan v. Union of India, Writ Petition (Civil) No. 55 of 2019, decided on 7 November 2022, where a five-judge bench upheld the One Hundred and Third Amendment by three to two.

Janhit Abhiyan repays statement by holding, because it is the line’s most divided moment and because both sides argued in the framework’s own terms. Justice Dinesh Maheshwari, for the majority, held that reservation is an instrument of substantive equality rather than an exception to it, that extending up to 10 percent reservation to economically weaker sections does not damage the basic structure, and that the 50 percent ceiling governs the reservations under Articles 15(4), 15(5) and 16(4), so that the new clause (6) reservations stand outside the ceiling’s field without breaching it. Justice Bela M. Trivedi held that the economically weaker sections form a distinct class for whom preferential treatment is a legitimate classification, and Justice J.B. Pardiwala held that economic criteria may ground affirmative action. Justice S. Ravindra Bhat, dissenting with Chief Justice U.U. Lalit concurring, held that the exclusion of the Scheduled Castes, Scheduled Tribes and socially and educationally backward classes from the EWS benefit discriminates against the very communities the equality code most protects, and that admitting economic criteria alone, while excluding the historically disadvantaged from the new benefit, violates the equality code at the level of basic structure. The dissent is the strongest statement of the anti-architectural position available: on Bhat J’s reasoning, the EWS amendment did not merely add content but broke the framework’s animating rule that preferential treatment tracks caste-linked disadvantage. The majority’s answer, that the ceiling and the classification doctrines are layer-internal tolerances which the new clause respects by standing in its own compartment, is the layered reading delivered from the bench. A reader who wants to know whether the three-layered architecture is a commentator’s imposition or the Constitution’s working self-understanding can start with the fact that, in the framework’s hardest case, both the majority and the dissent framed their disagreement as a disagreement about the architecture: what its compartments are, what its tolerances govern, and what would count as breaking it.

The Framework’s Silences: What the Layers Do Not Cover

An architecture is defined by its exclusions as well as its provisions, and the layered reading gains credibility from facing the three large silences in the framework’s coverage. The first silence concerns the private sector. Article 16 governs employment under the State, Article 15(5) reaches aided and unaided educational institutions, and neither touches hiring in private industry, so that the economy’s largest source of new employment stands outside Layer Two altogether. Demands to extend reservation into private employment have been pressed by Dalit and backward-class organisations since the 1990s, and were recorded with sympathy in the report of the National Commission to Review the Working of the Constitution of 31 March 2002, which recommended that the question be examined; no amendment has followed, and the Union government of the day answered the demand with voluntary affirmative-action assurances from industry instead. The silence is a design boundary, not an oversight: the services layer was built for the state’s own apparatus, and extending it would require the architecture’s first genuinely new compartment rather than another clause in an existing one.

The second silence concerns religious minorities as such. The founding settlement, after the Advisory Committee’s May 1949 turn against reserved seats for religious minorities, confined preferential machinery to caste-linked categories, and the claims of educationally backward religious communities have since been argued at the framework’s edge: the Sachar Committee’s report of November 2006 documented the Muslim community’s deficits in education and public employment, and the National Commission for Religious and Linguistic Minorities, reporting on 21 May 2007, recommended a 15 percent reservation for minorities in education and government employment, divided between Muslims and other minorities, a recommendation no government has enacted into the constitutional framework. The layered reading records the boundary without adjudicating the claim: the framework’s beneficiary grammar is backwardness verified by lists and commissions, and a religion-based compartment would change the grammar rather than extend it, which is precisely why the question remains political rather than doctrinal.

The third silence is internal to the state itself. The higher judiciary carries no reservation, Articles 124 and 217 providing for appointment without the framework’s machinery, and representation arguments inside the judicial branch are conducted through the conventions of the collegium rather than through any layer described here. Each silence marks a place where the architecture stops, and the fact that reformers must argue for extension, amendment or new compartments at exactly those places is further evidence that the structure is real: a boundary that has to be argued across is a boundary that exists.

The Adjudication: Why the Layered Reading Wins

The named disagreement this article set out to adjudicate can be decided. The article-by-article framing fails on its own strongest ground, fidelity to the text, because the text’s most conspicuous feature is the placement of its provisions across three Parts with three normative grammars, and an enumeration that reports each clause accurately while saying nothing about placement has omitted the text’s largest fact. The case-law-driven framing fails on sequence: it cannot explain why Parliament’s repairs were always surgical, one clause in one layer per amendment, across 1951, 1995 to 2001, 2005 and 2019, because it treats amendments as episodes in a political narrative rather than as maintenance operations on a structure whose joints the amenders could see. The three-layered reading explains both bodies of evidence with one hypothesis, and it does so while preserving what each rival gets right: the enumeration’s clause-level accuracy, which the layered reading needs and Table 1 incorporates, and the case-law narrative’s drama, which the layered reading absorbs as the history of the tolerances.

The scholarly record, fairly read, converges toward the layered view without having stated it. Austin’s Working a Democratic Constitution: A History of the Indian Experience, published in 1999, narrated the republic’s first fifty years as the working of an institutional system under stress, and its treatment of the Mandal years shows the amendment-and-litigation exchange operating as system maintenance rather than as constitutional breakdown. Galanter’s Competing Equalities of 1984 remains the indispensable map of the programmes, and its organising term, compensatory discrimination, captures Layer Two and the education flank with precision; the layered reading’s claim against Galanter is narrow, that a programme-centred map cannot register the representational layer as law rather than politics, Articles 330 to 342 sitting outside his frame although they are the framework’s oldest continuous element, descending from 1932. Anupama Rao’s The Caste Question: Dalits and the Politics of Modern India, published in 2009, presses from the opposite side, reading the constitutional settlement as the state’s management of caste into administrable categories, with the lists and commissions of Part XVI as the management’s instruments. Rao’s reading is the strongest available statement of the suspicion that the architecture disciplines the disadvantaged more than it empowers them, and the layered reading answers it on the record rather than by dismissal: the same machinery she names, the lists at Articles 341 and 342 and the commissions at Article 338 and its successors, has repeatedly been turned by its subjects into a forum of claim, from L.M. Shrikant’s annual reports in the 1950s to the commission hearings through which backward-class organisations placed their evidence before the state. Instruments of administration can be, and in this record have been, instruments of claim. The disagreement is marked as disputed here because the evidence runs both ways, and the layered reading does not need it resolved: whether Part XVI’s machinery is read as management or as forum, its separateness as a layer, the fact the article defends, is common ground between the two readings.

One falsifier disciplines the whole claim. If the layered reading were false, the framework’s history should show provisions migrating between functions, prohibitions doing distributive work, seat reservations justified as rights, amendments redrawing several layers at once. The record shows the opposite at every point where it can be checked: the Communal Award’s separate electorates were rejected for breaking the representational layer’s joint-electorate design, not for bad policy; the 1951 amendment repaired one clause; the 1995 amendment restored one stage of one layer; the 2019 amendment added clauses inside existing articles and was defended, successfully, on the ground that the compartments held. Convergence of this order, across seventy years and dozens of actors who agreed on little else, is what an architecture looks like when it is real.

Three Logics, One Settlement

The constitutional reservation framework, read whole, is a settlement among three necessities that could not be met by any single clause. A republic founded against caste needed to forbid discrimination and abolish untouchability with the full force of enforceable right. It needed to open its services and classrooms to communities that formal equality alone would have left at the gate for generations, and it wrote that opening as a permission guarded by an efficiency condition. It needed the dispossessed present in its legislatures from the first election, and it reserved their seats inside joint electorates under a clock that every generation since has chosen to reset. Prevention, permission and presence are not three names for one gesture. They are three logics, drafted in three sittings, amended on three separate tracks and interpreted in one continuous line of cases, and the framework’s coherence lies in their separation. India’s reservation framework operates through three layered provisions with distinct logics; the layered architecture is what makes the framework coherent.

Frequently Asked Questions

Q: What is constitutional reservation in India?

Constitutional reservation in India is the set of provisions through which the Constitution authorises and organises preferential treatment for the Scheduled Castes, the Scheduled Tribes and other backward classes in public employment, education and legislative representation. It operates through three layers: rights-protective provisions in Articles 15, 17 and 46 that forbid discrimination and abolish untouchability; services-affirmative provisions in Articles 16 and 335 that permit reservation of appointments and posts for inadequately represented backward classes; and representational provisions in Articles 330 to 342 that reserve seats in the Lok Sabha and state assemblies and create commissions and lists to administer the settlement. The provisions were adopted by the Constituent Assembly between 1948 and 1949, have been amended layer by layer since 1951, and are interpreted by the Supreme Court within stated tolerances, including the 50 percent ceiling and the creamy-layer exclusion. The framework’s beneficiaries and instruments have expanded over time, most sharply through the economically weaker sections provisions of 2019, without any change to the three-layer organisation adopted at the founding.

Q: What is Article 15 of the Indian Constitution?

Article 15 is the Constitution’s principal non-discrimination clause in Part III. Clause (1) forbids the State to discriminate against any citizen on grounds only of religion, race, caste, sex, place of birth or any of them, and clause (2) opens shops, public restaurants, hotels, places of public entertainment, wells, tanks, bathing ghats and roads to all citizens. The later clauses are permissions built into the prohibition: clause (3) allows special provisions for women and children; clause (4), inserted by the First Amendment of 1951, allows special provisions for the advancement of socially and educationally backward classes, Scheduled Castes and Scheduled Tribes; clause (5), inserted in 2006, extends such provisions to admission in private educational institutions other than minority institutions; and clause (6), inserted in 2019, allows up to 10 percent reservation for economically weaker sections. Article 29(2), the companion guarantee against denial of admission to state-maintained or state-aided institutions on grounds of religion, race, caste or language, is expressly subordinated to clause (4), which is why the First Amendment of 1951 addressed the two provisions together.

Q: What is Article 16 of the Indian Constitution?

Article 16 guarantees equality of opportunity in public employment. Clause (1) promises all citizens equality of opportunity in matters relating to employment or appointment to any office under the State, and clause (2) bars discrimination in that field on grounds of religion, race, caste, sex, descent, place of birth or residence. Clause (4) is the services-affirmative core of the reservation framework: it empowers the State to reserve appointments or posts for any backward class of citizens that, in the State’s opinion, is not adequately represented in the services. Clause (4A), added in 1995, extends reservation to promotions for Scheduled Castes and Scheduled Tribes; clause (4B), added in 2000, governs the carry-forward of unfilled reserved vacancies; and clause (6), added in 2019, permits up to 10 percent reservation for economically weaker sections. Every clause operates subject to Article 335’s efficiency condition. The trigger written into clause (4) is the State’s own opinion that a class is not adequately represented, which the Supreme Court has held does not demand proportionality between a class’s share of posts and its share of the population.

Q: How are SC and ST reservations decided?

The decision runs in three stages, each assigned by the Constitution to a different authority. Identification comes first: under Articles 341 and 342 the President specifies, by public notification for each state, the castes and tribes that count as Scheduled Castes and Scheduled Tribes, and only Parliament may alter those lists thereafter. Authorisation comes second: governments decide whether and how much to reserve, exercising the permissions of Articles 15(4) and 16(4), while seats in legislatures are reserved directly by Articles 330 and 332 in proportion to population. Review comes third: the Supreme Court polices the tolerances, including adequacy of representation under Article 16(4), the efficiency condition of Article 335, the 50 percent ceiling and the creamy-layer exclusion, and the commissions under Articles 338, 338A and 338B monitor the safeguards and report on their working. The percentages themselves are settled by the reserving government within those limits: 15 percent for the Scheduled Castes and 7.5 percent for the Scheduled Tribes in direct recruitment to Union services, tracking population weight, while state shares vary with demography and state policy.

Q: What is the EWS amendment?

The EWS amendment is the Constitution (One Hundred and Third Amendment) Act of 2019, which introduced reservation for economically weaker sections. Assented to on 12 January 2019 and in force from 14 January 2019, it inserted clause (6) into Article 15 and clause (6) into Article 16, authorising up to 10 percent reservation in education and in public employment for economically weaker sections of citizens other than the classes covered by the existing backward-class, Scheduled Caste and Scheduled Tribe provisions. The implementing criteria, set by the Department of Personnel and Training’s memorandum of 31 January 2019, include a family income ceiling of eight lakh rupees a year and asset limits. A five-judge bench upheld the amendment in Janhit Abhiyan v. Union of India on 7 November 2022 by a majority of three to two, holding that it did not damage the Constitution’s basic structure. The benefit is bounded by exclusion as well as income: persons already covered by the Scheduled Caste, Scheduled Tribe and backward-class reservations cannot claim it, a feature the dissenting judges in Janhit Abhiyan attacked as discriminatory and the majority sustained as a permissible classification.

Q: How long has reservation been extended?

Only one part of the framework carries a time limit: the reservation of seats in the Lok Sabha and state assemblies under Article 334, together with the Anglo-Indian special representation that originally shared its clause. As enacted, that reservation was to cease ten years after the Constitution commenced, in 1960. Parliament has extended the period seven times, by the Eighth Amendment to 1970, the Twenty-third to 1980, the Forty-fifth to 1990, the Sixty-second to 2000, the Seventy-ninth to 2010, the Ninety-fifth to 2020, and the One Hundred and Fourth Amendment of 2019 to 2030. The One Hundred and Fourth Amendment simultaneously ended the Anglo-Indian nominations. Reservation in services and education under Articles 15 and 16 has never carried a sunset clause, so the extensions concern legislative seats alone. Each extension was itself carried as a constitutional amendment, debated and voted, which makes the representational layer the only part of the framework whose continuation requires a fresh parliamentary decision at stated intervals.

Q: What is the National Commission for SCs?

The National Commission for Scheduled Castes is the constitutional oversight body established by Article 338. The article as enacted created a single Special Officer for the Scheduled Castes and Scheduled Tribes, whose annual reports to the President documented the working of the constitutional safeguards. The Sixty-fifth Amendment of 1990 replaced the officer with a multi-member National Commission for the Scheduled Castes and Scheduled Tribes, and the Eighty-ninth Amendment of 2003 divided that body, leaving the National Commission for Scheduled Castes under Article 338 and creating a separate commission for Scheduled Tribes under Article 338A. The Commission investigates and monitors the safeguards provided for the Scheduled Castes, inquires into complaints of deprivation of rights, advises on planning for their advancement, and reports to the President, exercising the powers of a civil court when it investigates a matter. Its reports are laid before Parliament, and copies touching a state are forwarded to that state’s government, so its findings enter the public legislative record.

Q: What does Article 17 abolish, and how is the abolition enforced?

Article 17 abolishes untouchability and forbids its practice in any form, providing that the enforcement of any disability arising out of untouchability shall be an offence punishable in accordance with law. The Constituent Assembly, debating the clause as Draft Article 11 on 29 November 1948, deliberately left the word untouchability undefined, so that the prohibition attaches to the historical practice rather than to a statutory list of acts. The right operates against private persons as well as the State. Enforcement runs through penal legislation: the Untouchability (Offences) Act of 1955 first punished the practice, the Protection of Civil Rights Act, as the statute was renamed in 1976, strengthened the punishments, and the Scheduled Castes and the Scheduled Tribes (Prevention of Atrocities) Act of 1989 added offences and special courts for atrocities against members of the two communities. Article 17 is also among the very few fundamental rights that operate against private persons, a necessary design since the practice it abolishes is maintained by society rather than by the State alone.

Q: What does Article 46 direct the State to do?

Article 46, in Part IV of the Constitution, directs the State to promote with special care the educational and economic interests of the weaker sections of the people, and in particular of the Scheduled Castes and the Scheduled Tribes, and to protect them from social injustice and all forms of exploitation. As a directive principle of state policy it is not enforceable by any court, a limitation Article 37 states expressly, but it is declared fundamental in the governance of the country and binding on the State as a guide to legislation and administration. Within the reservation framework, Article 46 states the protective purpose that the enforceable provisions execute: the Supreme Court treated it as subordinate to the fundamental rights in Champakam Dorairajan in 1951, while later benches read the directive principles and the fundamental rights together, so that Article 46 supplies the aim and Articles 15, 16 and 330 to 342 supply the instruments.

Q: What does Article 335 require when claims of Scheduled Castes and Scheduled Tribes are considered for posts?

Article 335 requires that the claims of members of the Scheduled Castes and the Scheduled Tribes be taken into consideration, consistently with the maintenance of efficiency of administration, in the making of appointments to services and posts under the Union and the states. The clause, settled in its final form when the Assembly revisited Draft Article 296 on 14 October 1949, builds the efficiency condition into the services layer of the reservation framework rather than leaving it as an external limit. The Eighty-second Amendment of 2000 added a proviso permitting relaxation in qualifying marks and lowering of standards of evaluation for Scheduled Caste and Scheduled Tribe candidates in promotion. In M. Nagaraj in 2006 the Supreme Court treated Article 335 as one of the standing conditions on reservation in promotions, requiring the State to respect administrative efficiency alongside backwardness and adequacy of representation. The condition is read as a genuine discipline rather than a recital: the balance it names must be demonstrated in the State’s material, not merely asserted in its order.

Q: How does Article 330 reserve seats for Scheduled Castes and Scheduled Tribes in the Lok Sabha?

Article 330 reserves seats in the House of the People for the Scheduled Castes and for the Scheduled Tribes, with a separate provision for the Scheduled Tribes of the autonomous districts of Assam. The number of seats reserved for each community in a state bears, as nearly as may be, the same proportion to the state’s total seats as the community’s population bears to the state’s population, so the reservation tracks demography rather than a fixed share. Reserved constituencies are filled by joint electorates, in which every voter in the constituency votes and only members of the reserved community may stand, the design inherited from the Poona Pact of 1932. Under the delimitation in force, 84 of the 543 elected Lok Sabha seats are reserved for Scheduled Castes and 47 for Scheduled Tribes. Article 332 applies the same mechanism to the legislative assembly of every state.

Q: Why was reservation of legislative seats originally limited to ten years under Article 334?

The ten-year limit expressed the founding expectation that guaranteed legislative presence was a transitional device. When the Assembly debated Draft Article 295A on 24 and 25 August 1949, the case for the sunset was that the Scheduled Castes and Scheduled Tribes needed a secured place in the first legislatures while electoral competition normalised, and that a fixed term would force the republic to judge, after a decade, whether the device remained necessary. Members who feared that ten years was too short were answered that a future Parliament could extend the period by constitutional amendment if conditions required, which is what successive Parliaments have since done, seven times, the last extension carrying the period to 2030. The limit therefore worked less as an expiry date than as a standing review mechanism, periodically obliging the legislature to re-argue the case for the representational layer on the evidence of its working. Temporary provisions were the Assembly’s standard device for carrying a contested settlement past the founding moment without pretending that the underlying contest had closed.

Q: What is the National Commission for Scheduled Tribes?

The National Commission for Scheduled Tribes is the constitutional oversight body for the Scheduled Tribes established by Article 338A. It was created when the Eighty-ninth Amendment of 2003, in force from 19 February 2004, divided the combined National Commission for the Scheduled Castes and Scheduled Tribes into two bodies, leaving the Scheduled Castes commission under Article 338 and constituting the Scheduled Tribes commission under the new Article 338A. Its functions parallel those of its sister commission: it investigates and monitors all matters relating to the safeguards provided for the Scheduled Tribes under the Constitution and other laws, inquires into specific complaints of deprivation of rights, participates in and advises on the planning process for tribal advancement, evaluates progress, and presents annual reports to the President. When investigating, it holds the powers of a civil court, including summoning witnesses and requiring the production of documents. The separation answered a practical strain on the combined commission: the administration of the Scheduled Areas and the Fifth and Sixth Schedules gave tribal safeguards a volume and specificity that a single body covering both communities had struggled to serve.

Q: What is the National Commission for Backward Classes?

The National Commission for Backward Classes is the oversight body for the socially and educationally backward classes, given constitutional status by Article 338B through the One Hundred and Second Amendment of 2018. It began as a statutory body under the National Commission for Backward Classes Act of 1993, enacted after the Supreme Court in Indra Sawhney directed that a permanent expert body examine the lists of backward classes and complaints of over-inclusion and under-inclusion. Article 338B gives the Commission the same investigative and reporting powers that Articles 338 and 338A give the commissions for the Scheduled Castes and Scheduled Tribes, including the powers of a civil court during an investigation and the duty to report to the President. The One Hundred and Fifth Amendment of 2021 confirmed that the states retain the power to prepare and maintain their own lists of backward classes alongside the central list. In its statutory years the Commission’s advice on the inclusion and exclusion of communities in the central list was the channel through which list disputes reached the government, and its annual reports are placed before Parliament with a memorandum of the action taken on them.

Q: How did the Seventy-seventh Amendment bring reservation into promotions?

The Seventy-seventh Amendment answered a specific judicial holding. In Indra Sawhney, decided on 16 November 1992, the Supreme Court held that Article 16(4) speaks to initial appointments and does not authorise reservation in matters of promotion. Parliament responded with the Constitution (Seventy-seventh Amendment) Act of 1995, in force from 17 June 1995, which inserted clause (4A) into Article 16. The clause empowers the State to provide for reservation in matters of promotion, with consequential seniority, to any class of posts in the services under the State in favour of the Scheduled Castes and the Scheduled Tribes which, in the State’s opinion, are not adequately represented. The amendment was deliberately narrow: it covered promotions only, named only the Scheduled Castes and Scheduled Tribes, and left the judgment’s other holdings, including the 50 percent ceiling and the creamy-layer exclusion, in force. The Eighty-fifth Amendment of 2001 later made consequential seniority retrospective to 17 June 1995.

Q: What is the carry-forward rule for unfilled reserved vacancies?

The carry-forward rule governs what happens when a reserved vacancy cannot be filled in the year it arises for want of a suitable candidate from the reserved category. In its early form, unfilled reserved vacancies were carried forward to succeeding years, so that the reserved share of appointments in a later year could rise far above the annual percentage. The Supreme Court struck that working down in T. Devadasan v. Union of India in 1964, after carry-forward produced a year in which 64.4 percent of the posts went to reserved candidates. The Eighty-first Amendment of 2000, in force from 9 June 2000, inserted clause (4B) into Article 16, restoring carry-forward on a redesigned basis: unfilled reserved vacancies of a year are treated as a separate class of vacancies, to be filled in succeeding years, and are not counted together with the vacancies of the year in which they are filled when the 50 percent ceiling is applied.

Q: What did the Supreme Court decide in M. Nagaraj v Union of India?

In M. Nagaraj v. Union of India, decided on 19 October 2006, a five-judge bench upheld the constitutional amendments that had built reservation in promotions: the Seventy-seventh Amendment’s clause (4A), the Eighty-first Amendment’s carry-forward clause (4B), the Eighty-second Amendment’s proviso to Article 335 on relaxation of qualifying standards, and the Eighty-fifth Amendment’s grant of consequential seniority. The upholding was conditional. The Court held that a State exercising the promotion power must support it with quantifiable data showing the backwardness of the beneficiary class, its inadequate representation in the services, and the maintenance of administrative efficiency under Article 335, and that the creamy-layer exclusion applies to Scheduled Castes and Scheduled Tribes in promotions. A later bench in Jarnail Singh, on 26 September 2018, removed the requirement to prove the backwardness of the Scheduled Castes and Scheduled Tribes with fresh data, holding their constitutional listing sufficient on that element, while retaining the remaining conditions.

Q: What does Article 339 require the Union to do?

Article 339 places the Union in a supervisory position over the administration of the Scheduled Areas and the welfare of the Scheduled Tribes, and it equips the Union with two instruments. Under clause (1), the President may at any time appoint a commission to report on the administration of the Scheduled Areas and the welfare of the Scheduled Tribes in the states, a power exercised in the commission chaired by U.N. Dhebar, appointed on 28 April 1960, and again in the commission chaired by Dilip Singh Bhuria, appointed in 2002. Under clause (2), the executive power of the Union extends to giving directions to a state as to the drawing up and execution of schemes specified to be essential for the welfare of the Scheduled Tribes in that state. The article thus belongs to the representational layer of the reservation framework as its oversight machinery, complementing the commissions of Articles 338, 338A and 338B.

Q: What is horizontal reservation, and how does it differ from vertical reservation?

The distinction was systematised by the Supreme Court in Indra Sawhney in 1992. Vertical reservation is the set of quotas allotted between the large constitutionally recognised categories, Scheduled Castes, Scheduled Tribes and Other Backward Classes, each forming a separate compartment of seats or posts, with the unreserved category standing as the fourth compartment. Horizontal reservation cuts across those compartments: it reserves shares within every vertical compartment for groups defined by a characteristic other than caste category, such as women, persons with disabilities and ex-servicemen. A candidate selected under a horizontal quota counts within her own vertical compartment, so that the horizontal reservation adjusts the distribution inside each compartment without adding to the total reserved share. The distinction governs the daily operation of the services and admissions layers: vertical percentages answer the claims of the listed communities, while horizontal percentages answer claims that run across all communities at once.

Q: How did the Ninety-third Amendment extend special provisions to private educational institutions?

The Ninety-third Amendment closed a gap that litigation had opened in the education flank of the framework. In P.A. Inamdar v. State of Maharashtra in 2005, a seven-judge bench held that the State could not impose its reservation policy on unaided private professional institutions. Parliament replied with the Constitution (Ninety-third Amendment) Act of 2005, in force from 20 January 2006, which inserted clause (5) into Article 15. The clause empowers the State to make special provision by law for the advancement of any socially and educationally backward classes of citizens, or of the Scheduled Castes and Scheduled Tribes, in admission to educational institutions, including private institutions whether aided or unaided by the State, with one exception: minority educational institutions protected by Article 30(1) are excluded. The Supreme Court upheld the amendment and the implementing Central Educational Institutions Act in Ashoka Kumar Thakur v. Union of India on 10 April 2008, applying the creamy-layer exclusion to admissions.