The Poona Pact of 24 September 1932 is usually remembered as the settlement that ended a fast. On 16 August 1932 Ramsay MacDonald, the British Prime Minister, announced the Communal Award and gave the Depressed Classes separate electorates. On 20 September 1932 Mohandas Gandhi, a prisoner in Yerwada Central Jail, began a fast unto death against that single provision. Four days later B. R. Ambedkar and Madan Mohan Malaviya signed an agreement at Poona that withdrew the separate electorates and replaced them with joint electorates and reserved seats. The claim defended here is architectural: the Poona Pact is best read not as a compromise between two men but as a piece of constitutional engineering whose four design choices governed Indian representation for the rest of the century.

The standard accounts arrange September 1932 as a duel. In the personal-confrontation reading, Gandhi’s fast is the event and Ambedkar’s signature its reluctant outcome; the settlement matters because of what it reveals about the two men. In the simple-compromise reading, the Pact is a midpoint: Ambedkar surrendered separate electorates, the Hindu leadership surrendered a larger block of seats, and both sides walked away with half a loaf. Neither reading can explain the document’s afterlife. A compromise explains a week; it does not explain why the Constituent Assembly of 1947 to 1949, drafting a constitution for a free republic, reached for the Poona design rather than inventing another, or why the reservation clauses of the Constitution still carry the Pact’s fingerprints in their structure, their arithmetic and their ten-year clock. The architectural reading can explain those things, because it treats the Pact as what its text shows it to be: a set of four deliberate choices about how a stigmatised minority would be represented inside a majoritarian legislature.

The four choices are these. The Pact chose joint electorates over separate electorates, so that a Depressed Classes candidate would be elected by all the voters of a constituency rather than by his own community alone. It chose a primary election, in which Depressed Classes voters first selected a panel of four candidates for each reserved seat, as the filter through which joint electorates would operate. It chose a quantum of 148 reserved seats in the provincial legislatures, more than double the 71 seats the Communal Award had allotted. And it chose a time limit: the primary-election device would lapse after ten years, while reserved seats themselves would continue until the communities concerned agreed to end them. Each choice had a price, each price was paid by somebody identifiable, and each choice survived into the Constitution of 1950 in a traceable form. Reading the Pact through its four choices also disciplines the argument about its meaning. Ambedkar’s own verdict on the settlement, delivered in his 1945 book What Congress and Gandhi Have Done to the Untouchables, was severe, and any reading of 1932 must answer it. The answer offered here is that a founder’s verdict on his own bargain and the bargain’s structural consequences are two different objects: the consequences ran on, through elections Ambedkar fought and lost and won inside the design, whether or not he forgave the week that produced it.

The Long Road to September 1932

The electorate question was thirteen years old when MacDonald settled it by decree. In January 1919, giving evidence before the Southborough Franchise Committee in Bombay, the twenty-seven-year-old B. R. Ambedkar demanded separate electorates for the Untouchables, arguing that a community shut out of temples, wells and schools could not trust its representation to the votes of the very caste Hindus who enforced the exclusion. The Government of India Act 1919 gave the Depressed Classes nominated members instead, a concession Ambedkar treated as proof that nomination delivered spokesmen chosen by officials rather than by the represented. Through the 1920s he built the organisational base that turned a demand into a constituency: the Bahishkrit Hitakarini Sabha of 1924, the Mahad satyagraha of March 1927 for the right to draw water from the Chavadar tank, and the burning of the Manusmriti at Mahad on 25 December 1927, which announced that the claim was not for a kinder place inside the Hindu order but for rights against it.

What did Ambedkar demand from the Simon Commission in 1928?

In the memorandum the Bahishkrit Hitakarini Sabha submitted to the Simon Commission in October 1928, Ambedkar held that the Depressed Classes formed a distinct and independent minority, separate from the Hindu community, and he demanded representation for them on the same basis as the Muslim minority.

The demand was conditional in form: reserved seats inside joint electorates if universal adult franchise were granted, and a campaign for separate electorates if it were not.

The conditional form matters for everything that follows. Ambedkar’s first preference in 1928 was not separation for its own sake; it was a franchise wide enough to make joint electorates safe, because a Depressed Classes voter registered without property or education tests could hold his own in a common roll. When the franchise stayed narrow, as the Lothian Committee’s recommendations of 1932 kept it, the safeguard had to move from the breadth of the roll to the structure of the constituency itself. The Simon Commission’s report of 1930 rejected separate electorates for the Depressed Classes while retaining reserved seats, a halfway house that satisfied nobody and left the question to the Round Table Conferences in London. Ambedkar attended the first conference of November 1930 to January 1931 as a Depressed Classes delegate and pressed the claim in the Minorities Committee; the Congress, boycotting the conference, was absent, which allowed him to present the demand without a rival claimant to the Depressed Classes’ voice. Gandhi attended the second conference, from September to December 1931, as the Congress’s sole representative, and the collision the Pact would later have to settle took place there in the open.

The two men had tested each other once before London. On 14 August 1931, at Mani Bhavan in Bombay, Ambedkar called on Gandhi on the eve of the Mahatma’s departure for the second Round Table Conference, in a meeting arranged so that the Congress’s representative would hear the Depressed Classes’ case from its author before sailing. Dhananjay Keer’s biography of Ambedkar preserves the exchange that both camps afterward quoted against each other: Gandhi protested that the Congress had spent its own funds and its political capital on the Depressed Classes’ uplift, and Ambedkar answered that the Congress had done nothing substantial, that its charity was a substitute for the community’s rights, and, when Gandhi spoke of their common homeland, that he had no homeland, since a country whose people treated him as less than human could not claim him. The courtesies of the meeting concealed its substance, which was a negotiation that failed before it began: Gandhi invited Ambedkar to trust the Congress’s conscience, and Ambedkar declined the currency. The London deadlock and the Poona settlement both descend from that afternoon, because the question left open at Mani Bhavan, whether the Depressed Classes would be represented by their own chosen men or by the conscience of others, was the question the Communal Award answered one way and the Poona Pact was drafted to answer another.

At the Minorities Committee meeting of 13 November 1931 at St James’s Palace, Gandhi declared that he would resist with his life the grant of a separate electorate to the Depressed Classes, and he claimed that the Congress, and in effect he himself, represented the Untouchables of India. Ambedkar’s reply, recorded in the conference proceedings, rejected the claim of representation by a body in which the Depressed Classes had no chosen voice. Eight days of deadlock ended with the minority delegates, Ambedkar among them, signing a joint memorandum on 13 November 1931 that demanded separate electorates for Muslims, Depressed Classes, Indian Christians, Anglo-Indians and Europeans as a package. MacDonald closed the conference on 1 December 1931 with the warning that, failing Indian agreement, the British government would impose its own scheme. Gandhi returned to India on 28 December 1931, was arrested on 4 January 1932 as the renewed civil disobedience campaign was suppressed, and was lodged in Yerwada Central Jail outside Poona. From that cell, on 11 March 1932, he wrote to Sir Samuel Hoare, the Secretary of State for India, to repeat the London warning in writing: the statement at the conference, he insisted, was not said in the heat of the moment nor by way of rhetoric, and if separate electorates were created for the Depressed Classes he would fast unto death. Hoare’s reply promised a decision on the merits. The merits, as London weighed them, produced the Communal Award.

Two features of the pre-1932 landscape fixed the positions from which the September negotiation would start. The first was the Congress’s own record on untouchability, which Gandhi carried into the argument as his title to speak. The Nagpur Congress of December 1920 had placed the removal of untouchability in the Congress programme, Gandhi had intervened in the Vaikom satyagraha of 1924 to 1925 in Travancore, where Depressed Classes campaigners demanded the right to use the roads around the temple, and he had insisted for a decade that the Congress’s claim to represent India stood or fell with its willingness to carry the Depressed Classes inside the national movement rather than license a separate political existence for them. The second feature was the hardening of the Depressed Classes’ own politics into an all-India form. At Nagpur in August 1930 Ambedkar presided over the All India Depressed Classes Congress and restated the demand in its conditional form, reserved seats under a universal franchise or separate electorates without one, and the conference’s existence announced that the claim no longer rested on one man’s evidence before committees: an organised constituency stood behind it, with provincial associations, newspapers and a programme. When the fast came, therefore, neither principal was improvising. Gandhi was defending a twelve-year-old Congress commitment with the only weapon his politics allowed him, and Ambedkar was defending a demand his movement had ratified in conference, which is one reason the negotiation at Poona was conducted through clauses and mechanisms rather than through appeals to sentiment, and why its product reads like a statute instead of a handshake.

The Precedent: Separate Electorates from 1909 to Lucknow

No device in Indian constitutional history arrived in 1932 without a pedigree, and the pedigree of the separate electorate shaped what both sides believed was at stake. The Indian Councils Act of 1909, the Morley-Minto reforms, had created separate electorates for Muslims, under which Muslim seats in the councils were filled by Muslim voters alone, following the Simla deputation of 1906 in which a Muslim delegation had asked the Viceroy, Lord Minto, for exactly that protection. The Congress condemned the device and then, at Lucknow in December 1916, accepted it: the Lucknow Pact between the Congress and the Muslim League traded Congress recognition of separate electorates for Muslims, with weightage in the provinces where Muslims were a minority, for League support of a common scheme of self-government. The Nehru Report of 1928 attempted to retire the device, proposing joint electorates with reserved seats for Muslims where they were a minority, and the attempt failed against the Muslim League’s insistence, restated in Jinnah’s Fourteen Points of 1929, on separate electorates as an irreducible safeguard. By 1932, therefore, the separate electorate was the established Indian technology for protecting a minority that distrusted the majority’s good faith, and its track record cut both ways in the argument. For Ambedkar, the Muslim precedent proved the device worked as designed: a community with its own electorate returned members answerable to itself, and no Muslim member owed his seat to Hindu votes. For Gandhi and the Congress, the same precedent proved the device’s danger: a machinery built to register a community as a separate political nation had, in the Muslim case, accompanied and accelerated the claim to separate nationhood itself, and to extend it to the Depressed Classes was to write the partition of Hindu society into the electoral law a generation before any partition of territory.

The Depressed Classes’ case differed from the Muslim case in the one respect Gandhi’s argument kept returning to, and the difference deserves statement in its strongest form on both sides because the adjudication turns on it. Muslims possessed, in Gandhi’s analysis, a religion, a history and a territorial concentration that made them a nation in embryo; the Depressed Classes possessed none of these, being dispersed through every village of Hindu India, bound to caste Hindu society by labour, credit and custom, and incapable of separation without ruin. A separate electorate for a dispersed and dependent population, he argued, would not create a protected minority but a licensed hostage: its members would face the majority’s resentment in the village while its representatives performed independence in a distant legislature. Ambedkar’s answer inverted the same facts. Precisely because the Depressed Classes were dispersed and dependent, joint electorates would deliver their representation to the men who controlled their livelihoods; a labourer who owed his work, his hut site and his credit to caste Hindu employers would vote as directed, and the representative so returned would be the employers’ candidate in Depressed Classes dress. Dispersion, on this reading, was not the argument against the separate electorate but the argument for it. Both readings of the same sociology were live in 1932, both were held by men who knew the villages at first hand, and the Pact can be read as the institutional compromise between two sociologies: it kept the dispersed community inside the general constituency, as Gandhi’s reading required, and gave it the primary election, as Ambedkar’s reading required, so that the direction of dependence at the decisive moment, the making of the shortlist, ran from the community outward rather than from the employers inward.

The Communal Award and the Arithmetic of Separation

The Communal Award of 16 August 1932 was MacDonald’s imposed settlement of the minorities question, and its treatment of the Depressed Classes was the clause that detonated it. The Award kept separate electorates for Muslims, Sikhs, Indian Christians, Anglo-Indians and Europeans, extended the device across the provincial legislatures, and for the first time placed the Depressed Classes inside the same machinery: 71 seats in the provincial legislatures were set apart for Depressed Classes constituencies, in which only Depressed Classes voters would choose Depressed Classes members. The Award added a refinement that its defenders quoted often in the weeks that followed. The Depressed Classes voter received a double vote, one in his special constituency and one in the general constituency where he lived, and the separate-electorate arrangement for the Depressed Classes was to run for twenty years before lapsing. Gandhi never disputed the Award’s treatment of Muslims, Sikhs or Christians; his objection, stated for a decade, was that a separate electorate for the Depressed Classes would sever them from Hindu society in law, converting a religious disability he called a blot on Hinduism into a permanent political partition of the Hindu body.

The numbers deserve the attention the standard accounts rarely give them, because September 1932 was in part an argument about arithmetic conducted by other means. Seventy-one seats spread across eight provincial legislatures, elected by a restricted franchise that registered only a fraction of Depressed Classes adults, would have produced a small bloc whose members owed their seats to nobody outside their own community. That independence was precisely what Ambedkar had demanded since 1919 and precisely what Gandhi feared: a representative who needed no caste Hindu vote could speak for the Depressed Classes without translation, and a Hindu majority that never had to solicit a Depressed Classes vote could ignore the community between elections. The Award also confirmed the category decision that underlay the arithmetic. By scheduling the Depressed Classes alongside religious minorities, the British government classified untouchability as a political separation rather than a social subordination inside Hinduism, and that classification, more than the seat count, is what Gandhi’s fast was staged to reverse. Ambedkar accepted the Award as the first recognition by the state that the Depressed Classes were, in the phrase of his 1928 memorandum, a distinct and independent minority; he had told the Round Table Conference that the community’s safeguards had to be constitutional, not charitable, and the Award wrote the safeguard into the electoral system itself.

The other Indian reactions to the Award fixed the negotiating field of September. The Congress Working Committee, meeting with its president in prison and its organisation battered by the ordinances of 1932, adopted the posture it had held toward communal settlements since the late 1920s and resolved neither to accept nor to reject the Award, a studied neutrality that left Gandhi’s personal vow as the only operative Congress position on the Depressed Classes clauses and left the Hindu leadership outside the Congress free to negotiate. The Hindu Mahasabha, whose president B. S. Moonje had signed the February pact with Rajah, opposed the Award’s separation of the Depressed Classes as a British design against Hindu consolidation and stood ready to pay seats for unity, which is why Malaviya, a Congress veteran presiding over the Mahasabha’s politics in those years, could convene the Bombay conference with both constituencies behind him. The Muslim League’s leadership, absorbed in the Award’s confirmation of Muslim separate electorates and weightage, treated the Depressed Classes clauses as a Hindu domestic quarrel in which its support was available to whichever side courted it, a reserve of indifference that Ambedkar noted when he weighed, and rejected, the option of letting the Award stand and building a minorities’ bloc upon it. Every party to the September negotiation thus entered it with its preference already priced: the British would accept an agreed alternative, the Mahasabha would pay in seats, the Congress would not obstruct its own Mahatma, and Ambedkar held the single asset the others needed, a signature that could retire the Award.

Gandhi’s response moved on the timetable he had fixed in March. On 18 August 1932, two days after the Award, he wrote to MacDonald from Yerwada that he had read the government’s decision and slept over it, and that in pursuance of his letter to Hoare and his declaration at St James’s Palace he had to resist the decision with his life, by a perpetual fast unto death from food of any kind save water with or without salt and soda, beginning at noon on 20 September 1932 unless the scheme of communal electorates for the Depressed Classes was withdrawn. The letter was cabled to London. The Award’s remaining provisions, little noticed in the storm over the Depressed Classes clauses, show how completely the communal principle organised MacDonald’s scheme: separate electorates were confirmed or created for a schedule of interests that ran from Muslims, Sikhs, Indian Christians, Anglo-Indians and Europeans to special constituencies for commerce, industry, mining, planting and labour, with a small reservation of seats for women distributed among the communal rolls, and with the Marathas of Bombay awarded a bloc of seats inside the general constituencies as a caste interest with a corporate claim. The scheme’s philosophy was a census philosophy, India as a federation of enumerated communities whose shares a British arbiter could fix. It is worth pausing on the fast’s target, because the personal-confrontation reading habitually blurs it: the fast was addressed to the British government, which alone could revise the Award, and its pressure fell on the Hindu leadership and on Ambedkar only because MacDonald had already announced that an alternative agreed between the communities would be accepted in place of the Award’s scheme. The architecture of the following week was set by that sentence. If Hindus and Depressed Classes could sign a common scheme before Gandhi died, London would substitute it; the negotiation therefore had to happen in Poona, at speed, with the fasting man’s life as its clock.

Six Days at Yerwada

The negotiation that produced the Pact was the second attempt in 1932 to settle the Depressed Classes question by agreement, and its predecessor explains its shape. In February 1932 M. C. Rajah, a Depressed Classes leader from Madras, and B. S. Moonje of the Hindu Mahasabha had signed an agreement for joint electorates with reserved seats, the Rajah-Moonje Pact, which the Hindu Mahasabha then advertised as proof that the community accepted joint electorates. Ambedkar repudiated it, on the ground that Rajah commanded no following comparable to his own and that a settlement signed without the Depressed Classes’ acknowledged leadership settled nothing. When Gandhi’s fast was announced, the Hindu leadership therefore knew both the formula that could replace the Award and the man whose signature alone could make the replacement stick. Madan Mohan Malaviya convened a conference of Hindu leaders in Bombay on 19 September 1932, the day before the fast began, and the conference’s work moved to Poona as Gandhi’s condition became the country’s front-page measure of time.

The mediators who carried proposals between the Yerwada cell and the negotiators outside were Tej Bahadur Sapru, M. R. Jayakar and Ghanshyam Das Birla, with C. Rajagopalachari working at Gandhi’s side; the detail matters because the Pact’s clauses bear the marks of lawyers negotiating a mechanism, not of two principals exchanging concessions. Ambedkar came to Poona and met Gandhi at Yerwada, and the conversation between them, retold in every biography, turned on the panel: Ambedkar’s price for surrendering separate electorates was a device that would let Depressed Classes voters control the shortlist from which the joint electorate chose. He is reported, in accounts of the meeting preserved by his biographer Dhananjay Keer, to have told Gandhi that he wanted the Depressed Classes to hold the power of selecting their representatives in the first instance, and Gandhi, who had said he was born an Untouchable by choice, pressed for a settlement he could accept without abandoning the principle of a common electorate. The panel of four was the mechanism that squared the two demands. By 24 September 1932 the terms were complete, and the agreement was signed by Ambedkar for the Depressed Classes and by Malaviya for the caste Hindu leadership, with Palwankar Baloo, the Depressed Classes cricketer whose standing crossed both worlds, among those who had organised the meetings at Yerwada.

The pressure around the negotiators was organised as deliberately as the drafting. Prayer meetings ran in Bombay and Poona through the week, telegrams reached the jail and the conference from Hindu organisations across the country, and temples in a number of towns were opened to Untouchables during the fast days in gestures meant to demonstrate the reforming will on which Gandhi’s side of the bargain depended. Depressed Classes opinion was not unitary, and the week’s record preserves its divisions: M. C. Rajah defended the joint-electorate formula he had signed in February and pressed Ambedkar to settle, provincial Depressed Classes associations telegraphed support for Ambedkar’s stand and instructed him not to yield the separate electorate at any price, and the knowledge that Gandhi’s death would be answered on Depressed Classes bodies in the villages hung over the conference as its unspoken agenda item, named in Ambedkar’s own later account as the consideration no negotiator in his position could wave aside. The settlement that emerged was thus negotiated by three parties, only two of whom were in the room: the Hindu leadership with its seat quantum, Ambedkar with his panel, and the crowd outside, whose anticipated violence fixed the deadline more absolutely than Gandhi’s weakening pulse.

The document’s route out of the jail was as carefully staged as its drafting. The British Cabinet considered the agreed scheme in London and announced its acceptance on 26 September 1932, substituting the Poona terms for the Award’s Depressed Classes clauses while leaving the Award’s other separate electorates untouched, and Gandhi broke his fast that evening, taking orange juice from Kasturba Gandhi in the presence of Rabindranath Tagore. The six days had killed nothing and settled the form of Indian minority representation for a generation, yet the reactions recorded at the time cut against any simple story of relief. Jawaharlal Nehru, in prison himself, wrote in the account later published in his autobiography that he had felt annoyed with Gandhi for choosing a side issue for his final sacrifice, a reaction that registered how completely the fast had redirected the national movement’s attention from the confrontation with the Raj to the internal constitution of Hindu society. In Bengal, Hindu leaders protested that the Pact’s Bengal allocation of 30 reserved seats surrendered too much of the province’s general representation, a grievance the province’s politics carried into the elections of 1937 and beyond. The settlement’s costs, in other words, were visible to its contemporaries; what was not visible, because it had not happened, was the settlement’s durability.

The Settlement Beyond the Jail: The Resolution of 25 September 1932

The Pact was an electoral instrument, and its signatories on the Hindu side understood that an electoral instrument alone would not discharge the obligation Gandhi’s fast had laid on them. On 25 September 1932, the day after the signing, a public meeting in Bombay adopted the resolution the Hindu leadership had promised: henceforth, amongst Hindus, no one should be regarded as an untouchable by reason of birth, and those classed as Depressed Classes should have the same rights in public institutions, wells, schools, roads and temples as other Hindus. On 30 September 1932 the All India Anti-Untouchability League was founded to carry the resolution into practice, with Ghanshyam Das Birla as its president and Amritlal V. Thakkar as its secretary; renamed the Harijan Sevak Sangh, it became the organisational vehicle of Gandhi’s campaign for the rest of the decade. Gandhi founded the weekly Harijan in February 1933, was released from Yerwada on 8 May 1933 as he began a further fast of twenty-one days for the Harijan cause, toured the country for the movement through 1933 and 1934, and pressed the temple-entry question from the Guruvayur satyagraha in Malabar to the Travancore Temple Entry Proclamation of 12 November 1936, by which the Maharaja of Travancore opened the state temples to all Hindus.

The social programme matters to this article because it marks the boundary of what the Poona settlement settled. The Pact decided how the Depressed Classes would be represented; it did not decide, and its clauses did not touch, whether untouchability would be abolished, whether caste itself would survive, or whether the Depressed Classes would remain Hindus. Gandhi’s Harijan campaign was his attempt to settle those questions on the terrain of Hindu reform, by penance and persuasion addressed to caste Hindu conscience. Ambedkar declined the terrain. He had been invited to join the Sangh’s work and refused the invitation’s terms, holding that an organisation of caste Hindu reformers dispensing uplift to the Depressed Classes reproduced the hierarchy it claimed to dissolve, and that the Depressed Classes needed power, not patrons: the annihilation of caste, not its softening, was the programme he had announced in the 1920s and would restate in his undelivered address Annihilation of Caste of 1936. The divergence is usually narrated as temperament, the reformer against the revolutionary. Read architecturally, it is a divergence about where change is manufactured: Gandhi’s post-Poona programme located change in the majority’s conscience and therefore accepted the majority’s timetable, while Ambedkar’s located it in enforceable shares of power, which is why he guarded the Pact’s clauses so jealously even while condemning the circumstances of their birth. The two programmes ran in parallel for the rest of both men’s lives, and the republic inherited both: the Constitution abolished untouchability in Article 17, a Harijan-programme clause carried into fundamental law, and reserved seats and posts on the Poona design, an Ambedkar-programme clause carried into the same text.

Choice One: Joint Electorates and the Price of Integration

The first clause of the Pact is a single sentence of electoral architecture: there shall be seats reserved for the Depressed Classes out of the general electorate seats in the provincial legislatures. The word that does the work is general. A reserved seat inside a general constituency is a seat for which every voter in the constituency votes, caste Hindu and Depressed Classes together, with the restriction that only a Depressed Classes candidate may hold it. Separate electorates inverted the restriction: only Depressed Classes voters voted, and the member they returned answered to them alone. The Pact’s choice between the two devices is the choice that every later Indian argument about reservation has had to re-litigate, because it fixed the direction of accountability in reserved representation. Under joint electorates, the Depressed Classes member of a provincial legislature holds his seat on the sufferance of a majority that is not his community; under separate electorates, he would have held it on the mandate of his community alone.

How do joint electorates change a reserved-seat candidate?

A candidate for a reserved seat under joint electorates must assemble a majority from an electorate in which Depressed Classes voters are a minority, so his campaign, his promises and his legislative conduct are addressed to caste Hindu voters as much as to his own community.

The joint electorate therefore prizes the Depressed Classes candidate acceptable to the majority and penalises the candidate who campaigns on confrontation with it, whatever his standing inside the community itself.

Gandhi’s case for the device was stated in his letter to Hoare of 11 March 1932 with unusual completeness. He was not against Depressed Classes representation, he wrote; he favoured registering every Depressed Classes adult, male and female, as a voter irrespective of education or property qualification, under a franchise wider for them than for others if need be. What he opposed was the electoral separation, because the Depressed Classes were distributed among caste Hindu populations and dependent on them in daily life, and a separate roll would harden a social subordination into a constitutional boundary between two electorates that would thereafter face each other as rival political nations. The argument was continuous with the position the Congress had carried since its founding generation first debated communal representation, the inheritance of the pre-Independence INC architecture context, and with the post-1920 Gandhi-Congress organisational architecture that made Hindu social reform a stated charge on the national movement. Ambedkar’s counter-case, stated at the Round Table Conference and restated for the rest of his life, was that the integration Gandhi prized was the integration of the prisoner into the prison: a community that could not draw water from a common well would not be emancipated by being folded into a common roll, and its representatives, chosen by the majority’s votes, would be the majority’s nominees wearing the community’s name. Both arguments were arguments about the same mechanism, and the mechanism is what the Pact actually decided: the accountability of the reserved-seat member would run outward to the whole constituency, and the community’s control over its representation would have to be exercised, if at all, before the general vote, at the stage of candidate selection. That is why the second choice, the panel, is not a detail.

The device behaved after 1937 broadly as its critics and its defenders had each predicted, which is what makes the first elections under the Pact such clean evidence. Across the provinces, the Congress party, with its unmatched organisation and its claim to the national succession, took the bulk of the reserved seats in the elections of early 1937, seating Depressed Classes members who owed their tickets to the Congress organisation and their majorities to the joint electorate; Ambedkar’s charge that the joint electorate manufactured pliable representatives was, in those provinces, a description of the result. In Bombay, where the Independent Labour Party contested the primaries with a disciplined Depressed Classes organisation of its own, the same machinery returned his candidates instead, which is the counter-demonstration: the joint electorate did not determine the character of reserved representation by itself, it transmitted the balance of organisation inside the community, rewarding whichever force had built the deeper roots among Depressed Classes voters. The integration the device produced was therefore real but conditional, a standing invitation to majority parties to organise the community and to the community to organise itself faster. Gandhi had wanted joint electorates to oblige caste Hindus to seek Depressed Classes votes; the record of 1937 shows the obligation running the other way as often as not, with Depressed Classes candidates seeking caste Hindu votes on majority-party tickets, and the difference between the two directions is the difference between his sociology and Ambedkar’s, still unsettled by the design that tried to hold both.

Choice Two: The Panel of Four and the Primary Election

Clause two of the Pact built the filter. All the members of the Depressed Classes registered in the general electoral roll of a constituency would form an electoral college, and that college would elect a panel of four candidates belonging to the Depressed Classes for each reserved seat, by the method of the single vote; the four persons getting the highest number of votes in the primary election would be the candidates for election by the general electorate. The design is exact about where power sits at each stage. Depressed Classes voters, and only they, decide which four names the constituency may choose among; the general electorate, voting jointly, decides which of the four sits in the legislature. Neither side can impose a member on the other: the majority cannot seat a candidate the community’s primary has rejected, and the community cannot seat a candidate the general electorate refuses. Ambedkar’s own summary of the device, in the statements he issued in the weeks after the Pact, stressed the veto the primary conferred: the panel ensured that no Depressed Classes representative could reach the legislature over the recorded opposition of Depressed Classes voters themselves.

The panel is the clause the simple-compromise reading cannot accommodate, because a compromise splits a difference and the panel instead created a new institution. Nothing like it existed in the Communal Award, in the Government of India Act 1919, or in the Congress’s own schemes; it was drafted in Poona in four days, and it converted the question of September 1932 from who votes to who shortlists. Its consequences ran in both directions, and the article’s adjudication below depends on holding both in view. In one direction, the primary gave Depressed Classes voters a collective instrument they had never held: a candidate unacceptable to the community could be kept off the ballot entirely, which meant that the joint electorate’s winner, whoever he was, carried a certificate of minimum acceptability signed by the community’s own votes. In the other direction, the primary distributed its power unevenly across the community’s internal divisions: a primary electorate organised by sub-caste, locality and patronage could return panels dominated by the larger and better-organised Depressed Classes groups of a province, and the general electorate could then pick the most pliable of the four. Eleanor Zelliot, in From Untouchable to Dalit (1992), traced how the device worked in the Bombay elections of 1937, where the panel system obliged Ambedkar’s Independent Labour Party to win the community’s own vote before it could ask for anybody else’s, and where the double test filtered candidates in ways neither the Award nor a bare reserved seat would have done.

The post-1947 fate of the panel is the sharpest single measure of what the republic kept from the Pact and what it discarded, and it is taken up in the inheritance section below. The panel survived into the Government of India Act 1935 and operated in the provincial elections of 1937 and 1946; the Constitution of 1950 dropped it, leaving the reserved seat to be filled by the joint electorate without the community’s preliminary veto. The dropping was done without a recorded Constituent Assembly debate of any length on the point, and it reversed the balance the panel had struck: after 1950, a Depressed Classes candidate for a reserved seat needed his party’s ticket and the general electorate’s majority, and the community’s collective power over the shortlist, the one power Ambedkar had salvaged from the wreck of separate electorates, existed no longer. A reading of the Pact that treats it as a transaction between two men has nothing to say about the panel; a reading that treats it as architecture can say precisely what was lost when the panel was removed, and when.

The Narrow Roll: Franchise Arithmetic and the Panel’s Real Weight

The panel cannot be priced without the franchise it operated on, because the Pact’s clauses were drafted against a roll that excluded most of the people they concerned. The Indian Franchise Committee under the Marquess of Lothian, whose report of 1932 supplied the Award’s electoral machinery, recommended a franchise resting on property, tax and education qualifications, extended to wives of qualified voters, which enfranchised roughly a tenth of the population; the Government of India Act 1935 carried an electorate of about thirty million into the elections of 1937. Inside a reserved constituency, therefore, the voters who mattered at the primary stage were only those Depressed Classes adults who had cleared the qualification hurdles and reached the general electoral roll, and clause two of the Pact is careful to say so: the electoral college consists of the members of the Depressed Classes registered in the general electoral roll of the constituency. The primary electorate was a narrow, enumerable, organisable body, a few thousand names in most constituencies, whose addresses a disciplined party could canvass and whose votes a disciplined party could deliver.

Why did the narrow franchise make the primary election decisive?

On a roll restricted by property and education tests, the Depressed Classes voters registered in a reserved constituency were few enough for a party organisation to know them by name and street. Winning the primary turned on organisation rather than opinion at large, and the panel carried the stamp of whichever Depressed Classes organisation had done the local work first.

That narrowness explains both the panel’s power and the ferocity of the later argument about abolishing it. On a narrow roll, organisation is destiny: the party with branches in the Depressed Classes quarters of a constituency, with the caste associations and the labour unions wired into its committee rooms, would sweep the primary and write the panel, and the general electorate’s choice among the four would be a choice among that party’s nominees. Ambedkar understood the arithmetic from the beginning, and the Independent Labour Party’s Bombay organisation of 1936, built ward by ward through the mill districts, was constructed to exploit it. The Congress understood it too, and Congress ministries after 1937 invested in Depressed Classes organisation of their own, through the Harijan Sevak Sangh’s networks and through patronage, precisely to contest primaries the party had once dismissed as a colonial complication. When the Constitution’s drafters dropped the panel in 1949, they did so in the first flush of universal adult franchise, under which the Depressed Classes share of a reserved constituency’s roll rose from a registered remnant to roughly the community’s share of the population, on the argument that numbers had replaced the veto. The substitution is the cleanest experiment the design affords: the panel was a device for making a small electorate count, and it was retired at the moment the electorate stopped being small. Whether the retirement was premature, in constituencies where the community’s voters remain a dispersed minority of the roll and party tickets remain in other hands, is the question the sub-categorisation and representation debates keep re-asking in new vocabulary.

Choice Three: A Quantum of 148 Seats

Clause one of the Pact distributes its 148 seats by province: Madras 30, Bengal 30, the United Provinces 20, the Central Provinces 20, Bihar and Orissa 18, Bombay with Sind 15, Punjab 8 and Assam 7. Clause four carried the device to the centre, reserving 18 per cent of the seats allotted to the general electorate for British India in the Central Legislature for the Depressed Classes, with the same panel procedure applied through clause three. Set against the Communal Award’s 71 provincial seats, the Pact’s quantum was not a marginal improvement but a doubling, and the doubling was the Hindu negotiators’ substantive contribution to the bargain: Malaviya’s conference could not give Ambedkar the separate electorate he had come with, so it paid in seats. The payment changed the political meaning of the reserved bloc. Seventy-one members scattered through eight legislatures, returned by special constituencies, would have formed a pressure group at the margin of each assembly; 148 members returned by general constituencies formed, in several provinces, a bloc large enough to matter to ministry-making, and large enough for the Congress ministries of 1937 to 1939 to discover that reserved-seat members could not be taken for granted in the division lobbies.

The quantum also fixed a ratio that outlived the instrument that created it. The Pact’s provincial distribution tracked, roughly, the Depressed Classes share of each province’s population as the Lothian Committee had estimated it, and the habit of tying the reserved quantum to population proportion passed from the Pact into the Government of India Act 1935 and from there into the Constitution, whose Articles 330 and 332 reserve seats for the Scheduled Castes and the Scheduled Tribes in the Lok Sabha and the state assemblies in proportion, as nearly as may be, to their share of the population. What the Pact did not do is as instructive as what it did. It reserved seats in legislatures and said nothing of a population-proportionate share of the executive, of the services beyond the fair-claims language of its seventh clause, or of the private economy in which most Depressed Classes labour was sold. The eighth clause directed that a sum be earmarked from provincial educational grants for Depressed Classes education, a recognition that representation without schooling would reproduce itself as tokenism; yet the earmark depended on provincial budgets voted by the very legislatures whose composition the Pact had just redesigned. The architecture, in short, was a legislature architecture. Ambedkar’s later insistence, through the 1940s, on statutory shares in the services and in education grants was an attempt to extend the Pact’s principle into the two domains its text had touched only with exhortation, and the Constitution’s Article 335, which carries the claims of the Scheduled Castes and Scheduled Tribes to services and posts into the constitutional text itself, is the trace of that attempt.

One part of the quantum never operated as drafted, and its failure is instructive. Clauses three and four assumed a Central Legislature elected under the new constitutional scheme, with 18 per cent of the general-electorate seats for British India reserved and filled through the panel procedure. The federation projected by the Government of India Act 1935 never came into force, foundering on the princes’ refusal to accede on the terms offered, so the central provisions of the Pact remained an allotment without an election, while the provincial provisions ran in 1937 and 1946 and built the practice the Constituent Assembly later codified. The asymmetry left a mark on the inheritance: Articles 330 and 332 reserve for the Lok Sabha and the assemblies by a population formula generalised from the provincial arithmetic of 1932, not from the percentage device of clause four, which died with the federation that was to have housed it. The design that governed the republic was thus the design that had been tested in the provinces, and the centre’s untested percentage survives only as evidence of how completely the Pact’s authors expected their architecture to scale from a Bombay constituency to the legislature of a federated India.

Choice Four: The Ten-Year Clock and the Mutual-Agreement Lock

Clauses five and six of the Pact set two different clocks running, and conflating them has confused commentary on the settlement ever since. Clause five provided that the system of primary election to a panel of candidates should come to an end after the first ten years, unless terminated sooner by mutual agreement. Clause six provided that the system of representation by reserved seats should continue until determined by mutual agreement between the communities concerned in the settlement. The distinction is deliberate and it is the fourth architectural choice: the safeguard of the shortlist was experimental and time-limited, while the safeguard of the seat was open-ended and locked against unilateral repeal. A caste Hindu majority in a future legislature could not abolish reserved seats by its own vote, because clause six made their continuance the default and their termination a matter for agreement between the communities; but the panel, the device that gave the community its grip on candidate selection, was built to expire in 1942 unless the parties renewed it by the same mutual agreement.

Why did the Poona Pact give the panel a ten-year life?

The panel was the clause the Hindu negotiators accepted most reluctantly, since it preserved a separate Depressed Classes vote inside the joint electorate they had fought to create, so the ten-year term was the price of their signature: the primary would prove itself in two provincial elections and then lapse unless both communities agreed to continue it.

The reserved seat carried no such term because Ambedkar would not sign a settlement whose central guarantee a future majority could switch off alone.

The clock’s actual behaviour is a study in how constitutional defaults govern outcomes that nobody votes on. The panel ran in the elections of 1937, was overtaken by the war and the suspension of normal provincial politics after 1939, and was still in the text of the Government of India Act 1935 when the elections of 1946 were held under it; no mutual agreement renewed it into the republic, and the Constitution’s drafters let it die with the Act that carried it. The reserved seat, protected by the clause-six default, crossed into the Constitution with its continuance presumed, and when the Constituent Assembly wrote Article 334 it replaced the mutual-agreement lock with a parliamentary clock of its own: reservation of seats would cease ten years from the commencement of the Constitution unless the period was extended. The substitution deserves emphasis, because it changed who held the key. Under the Pact, termination required the Depressed Classes’ consent; under Article 334, extension required only Parliament’s vote, and the history of the seven extensions that followed is the history of a safeguard that survives by the majority’s periodic permission rather than by the protected community’s veto. Ambedkar, who sat in the Constituent Assembly as chairman of the Drafting Committee and piloted the reservation clauses, accepted the parliamentary clock as the price of a constitution that had to be amendable by the republic’s own institutions; the acceptance was consistent with everything he wrote about the difference between safeguards granted by an imperial arbiter and safeguards carried by a democratic legislature’s repeated decision.

The Four-Choice Map of the Poona Pact

Table 1 sets out the four architectural choices of the Poona Pact, the immediate consequence each choice carried in the 1930s, and the mechanism by which each passed into the post-1947 framework. The table is the article’s findable artifact: a reader who needs the Pact’s design in one view, for citation in the reservation debates that run through this pillar, can take it from here.

Table 1. The Four-Choice Map of the Poona Pact (24 September 1932)

Architectural choice	Clause and mechanism	Immediate consequence, 1932 to 1947	Post-1947 inheritance
Joint electorates over separate electorates	Clause 1: reserved seats carved out of general electorate seats	Depressed Classes members answerable to whole constituencies; the Award’s separate Depressed Classes roll withdrawn	Articles 330 and 332: reserved seats inside general constituencies in the Lok Sabha and state assemblies
Primary election to a panel of four	Clause 2: Depressed Classes electoral college shortlists four candidates by single vote	Community veto over the shortlist; no member seated over the community’s recorded rejection	Dropped by the Constitution of 1950; reserved seats filled by joint electorate and party ticket alone
Quantum of 148 provincial seats and 18 per cent at the centre	Clauses 1 and 4: provincial distribution from Madras 30 to Assam 7	Reserved bloc doubled against the Award’s 71; bloc weight in ministry-making after 1937	Population-proportionate reservation under Articles 330 and 332, extended to the Scheduled Tribes
Ten-year panel sunset; reserved seats locked to mutual agreement	Clauses 5 and 6: panel lapses at ten years; seats continue until the communities agree otherwise	Panel ran in 1937 and 1946 and lapsed unrenewed; seats crossed 1947 with continuance presumed	Article 334: ten-year parliamentary clock, extended seven times, the seventh renewal running to 2030

Two cells of the map carry the argument of this article in miniature. The panel row records an inheritance that failed: the one device in the Pact that belonged to the Depressed Classes alone was the one device the republic discarded. The sunset row records an inheritance that mutated: the mutual-agreement lock of 1932, which made the community a party to any termination, became the parliamentary clock of Article 334, which made the community a petitioner for each extension. Between those two cells lies the whole subsequent history of Indian reservation politics, from the first extension debate of 1959 to the extension carried by the 104th Amendment in January 2020. The Mandal-era conflicts over the backward classes, examined in the Mandal moment of 1990, were fought inside this architecture rather than against it: the question in 1990 was never whether the state would reserve, but for whom, in what quantum, and under whose control of the list of beneficiaries.

The map also records what the Pact was not. It contains no row for the abolition of untouchability, because the Pact’s text does not mention the practice it existed to answer; no row for shares in the executive or the services, beyond the fair-claims clause and the education earmark; and no row for the social programme of temple entry and interdining that the Hindu signatories pledged in the resolution of 25 September, which lived outside the instrument and died, as a living campaign, with the decade. The absences are the boundary between the settlement as architecture and the settlement as redemption. Gandhi’s public language in September 1932 ran to redemption, the fast as an appeal to the Hindu heart; the document his negotiators signed contains none of it, only electorates, panels, seats and clocks. Ambedkar’s public language ran to power; the document he signed contains the power clauses and nothing else. That both men accepted a text confined to machinery, while each preached a larger gospel around it, is the surest internal evidence that the machinery was the settlement, and that the gospels were bids for its interpretation addressed to audiences the text itself did not need to persuade.

Ambedkar Inside the Architecture, 1932 to 1947

The Pact’s first consequence fell on its own signatory. Ambedkar had returned from London with the Communal Award in his pocket, the state conceding that the Depressed Classes were a political element distinct from the Hindu mass; he left Poona having signed the concession away, and the correspondence of 1932 and 1933 between Ambedkar and Depressed Classes leaders across the provinces, preserved in the Babasaheb Ambedkar: Writings and Speeches volumes, shows him working through the wreckage in real time: defending the panel as a real acquisition, conceding the loss of the separate electorate as a fact accomplished under the pressure of the fast, and redirecting his followers toward the conquest of the new machinery. The letters matter because they date his disillusion precisely. The strategic adjustment came first; the indictment came later, when the machinery had been tested. In the interim he performed the most consequential act of the Pact’s first decade, which was to win under its rules.

What did the Independent Labour Party’s 1937 result prove?

Ambedkar founded the Independent Labour Party in August 1936 to contest the first provincial elections under the Government of India Act 1935, and in the Bombay elections of early 1937 the party won 15 of the 17 seats it contested, eleven of them reserved seats filled through the Pact’s panel procedure.

The result proved the panel could deliver a community-chosen leadership to the legislature when a disciplined Depressed Classes organisation contested the primary, and it made Ambedkar the leader of the opposition benches in Bombay.

The 1937 victory is the strongest evidence for the panel as Ambedkar had bargained for it, and it is evidence the architectural reading can weigh exactly: in reserved constituencies where his organisation fought the primary, the community’s shortlist carried his candidates, and the joint electorate, Congress-dominated as Bombay’s was, seated them. The sequel is equally exact. When Ambedkar converted the Independent Labour Party into the All India Scheduled Castes Federation in 1942, broadening the claim from labour to caste and taking office the same year as Labour Member of the Viceroy’s Executive Council, where he served from 1942 to 1946, he was building the all-India machine the panel system rewarded: primaries are won by organisations, and the Federation was an organisation for winning them. In the provincial elections of 1946 the machine failed almost everywhere outside Bengal, undone by the Congress’s capture of the joint electorate and by the Federation’s thin provincial roots, and Ambedkar reached the Constituent Assembly in July 1946 through a Bengal seat arranged with Jogendra Nath Mandal’s support rather than through a Federation majority anywhere. The architecture had done what architecture does: it had set the terms on which his politics succeeded in 1937 and the terms on which it was checked in 1946, and it had done so impartially, through the primary’s demand for organisation and the joint electorate’s demand for majority acceptability.

It was between those two elections that Ambedkar wrote the verdict the architectural reading must answer. What Congress and Gandhi Have Done to the Untouchables, published in Bombay in 1945, worked through the fast and the Pact in prosecutorial detail: the fast, he argued, had been aimed not at the British but at the Depressed Classes’ one moment of leverage, and the joint electorate had delivered exactly what he had predicted in 1931, representatives who were the nominees of caste Hindu parties rather than the choice of the community, men who would sit silent in legislatures while the community’s questions went unasked. The book’s evidence was the conduct of Congress-elected reserved-seat members through the 1937 to 1939 ministries, and much of it was true of those members in those years. What the book could not yet weigh, because it had not happened, was the framework’s behaviour across a change of regime: the same joint electorates, the same reserved seats, carried into a republic with adult franchise, where the Depressed Classes vote in a reserved constituency was no longer a small registered minority of the roll but, under universal suffrage, a bloc every candidate for the seat had to court in the open. Ambedkar’s own trajectory ran through that change: re-elected to the Constituent Assembly from Bombay in 1947 after his Bengal seat lapsed at Partition, appointed Law Minister, and placed in the founding triangulation context of Patel, Nehru and Ambedkar as chairman of the Drafting Committee, he became the custodian of the clauses the Pact had drafted in embryo.

One entry in the trajectory sits outside the electoral story and defines its limit. At Yeola in the Nashik district, on 13 October 1935, addressing a conference of Depressed Classes youth, Ambedkar declared that he had been born a Hindu, which was not in his power to prevent, but that he would not die a Hindu, and the conference resolved to leave the Hindu fold. The declaration is the obverse of the Pact. Joint electorates presuppose a shared political community: the reserved seat works because Depressed Classes voters and candidates remain inside the general constituency, and the whole design of 1932 was Gandhi’s wager that the community could be held inside Hindu society by a redesign of its representation. Yeola announced that the wager’s premise was itself in question, that a leader who had signed the architecture was preparing his people’s exit from the religious body the architecture assumed. The exit took twenty-one years to complete, at Nagpur on 14 October 1956, when Ambedkar embraced Buddhism with several hundred thousand followers, weeks before his death. The republic had by then constitutionalised the Pact’s design for a population a large part of whose leadership had renounced the identity the design presumed, and the design survived the renunciation without amendment, which is perhaps the strongest evidence that what was built in 1932 was a piece of state machinery rather than a Hindu family settlement: it went on operating, indifferently and effectively, for citizens who had ceased to describe themselves in its terms.

The Post-1947 Inheritance

The Constituent Assembly’s handling of reservation was the Pact’s architecture passing through a second founding. The Advisory Committee on Fundamental Rights and Minorities, chaired by Vallabhbhai Patel, reported in August 1947 in favour of reserved seats for the Scheduled Castes and the Scheduled Tribes in the legislatures, inside joint electorates, for an initial period of ten years; the Assembly adopted the scheme inside the founding-period long-arc context: the three tensions between unity, democracy and social transformation that structured its whole design. When the Assembly returned to the minorities question in May 1949 and abolished reserved seats for religious minorities, it expressly retained the Scheduled Castes and Scheduled Tribes provisions, on the reasoning Sardar Patel stated to the House: the Depressed Classes’ reservation rested not on religion but on a backwardness the republic had pledged itself to remove, and it would stand until removal made it needless. The distinction between a communal safeguard, struck down, and a compensatory safeguard, retained, is the Constitution’s own commentary on 1932: what Gandhi had resisted as the communal electorates’ logic was refused, and what Ambedkar had extracted at Poona as the community’s floor was kept.

The drafting translated the Pact clause by clause, and the translations are traceable. Articles 330 and 332 enact choice one and choice three together: seats reserved for the Scheduled Castes and the Scheduled Tribes in the Lok Sabha and the state legislative assemblies, filled from general constituencies by joint electorates, in proportion as nearly as may be to population, the Pact’s 148-seat arithmetic generalised into a formula and extended to the Scheduled Tribes, whom the Pact had not covered. Article 334 enacts choice four in mutated form: the reservation of seats ceases after ten years unless Parliament extends the period, as Parliament has done by the 8th Amendment of 1959, the 23rd of 1969, the 45th of 1980, the 62nd of 1989, the 79th of 1999, the 95th of 2009 and the 104th of 2019, carrying the terminus from 1960 to 2030 across seven renewals. Two administrative devices completed the translation, and both descend from the Pact’s arithmetic. The first is the list: a population-proportionate quantum is meaningless until the state names the communities to which it applies, and Articles 341 and 342 empowered the President to specify the Scheduled Castes and Scheduled Tribes by public notification, done in the Constitution (Scheduled Castes) Order and the Constitution (Scheduled Tribes) Order of 1950, with Parliament alone competent to vary the lists thereafter. The list is the architecture’s gatekeeper, the point at which a caste’s inclusion or exclusion decides its share of every reserved seat and post, and the politics of list-making, from the state-level additions of the 1950s to the sub-categorisation disputes of later decades, is conducted at that gate. The second device is rotation: a reserved constituency is not a permanent communal pocket, because delimitation commissions redrew the legislative map in the 1950s, the 1960s and the 1970s, and again on the census of 2001, reassigning the reserved designation among constituencies so that the burden and benefit of reservation circulated across the map of each state. Rotation has no ancestor in the Pact’s text, where the 148 seats sat in named provinces without a circulation rule, and it is the republic’s own amendment to the design, softening the joint electorate’s integration logic one step further: not only must the reserved-seat member answer to the whole constituency, the constituency itself holds the designation in trust for a delimitation cycle rather than in perpetuity. Article 335 carries the Pact’s seventh clause, the fair-claims language on the services, into constitutional text, and Article 16(4) empowers the state to reserve appointments in the services for backward classes inadequately represented, the provision whose administrative career runs from the state lists of the 1950s to the Mandal Commission of 1980. The First Amendment of 1951, which added Article 15(4) after the Supreme Court’s judgment in State of Madras v. Champakam Dorairajan of 9 April 1951 struck down the Madras communal order, completed the framework’s extension from legislatures to education and the services, inside the compensatory logic the Assembly had articulated in May 1949.

Choice two is the inheritance’s negative space. The panel did not survive the translation, and the reasons the record gives are thin: the Constituent Assembly’s advisers regarded the primary as a colonial complication, an election before the election, and adult franchise was expected to do the panel’s work by numbers, since a fifth or more of a reserved constituency’s voters under universal suffrage would be Scheduled Castes electors whose support no candidate could ignore. The expectation was not absurd, and in constituencies with large Scheduled Castes populations it has often been fulfilled; yet the substitution changed the community’s power from a veto over the shortlist, exercisable collectively before a single general vote was cast, to a bargaining weight inside other parties’ ticket distribution, exercisable only through the parties themselves. Christophe Jaffrelot, in Dr Ambedkar and Untouchability (2005), drew the balance that the architectural reading adopts: the post-1947 system guaranteed the Scheduled Castes a presence in every legislature and simultaneously routed that presence through party organisations the community did not control, so that the question Ambedkar had posed in 1931, whether a reserved-seat member is the community’s representative or its host party’s, remained open inside the republic that had inherited his committee’s draft. His chairmanship of the Drafting Committee placed the man who had signed away separate electorates in charge of the text that constitutionalised the surrender’s terms, which is why the inheritance reads less like a defeat administered to him than like a design he spent his last decade trying to operate, extend and, in the Hindu Code Bill battles, push past its own limits.

The design’s first test under universal franchise produced the result its author had predicted for it in 1931. In the general election of 1951 to 1952, Ambedkar contested the Bombay North parliamentary constituency, a reserved seat, as a Scheduled Castes Federation candidate, and lost to Narayan Sadoba Kajrolkar, a Depressed Classes candidate of the Congress, in a constituency where the joint electorate’s majority and the Congress organisation together outweighed the community’s most famous name. The defeat is the inheritance’s founding parable, quoted by both traditions ever since: for the Ambedkarite reading, proof that a reserved seat under joint electorates returns the majority party’s nominee rather than the community’s tribune, even when the tribune is Ambedkar; for the design’s defenders, proof that the seat performed its stated function, sending a Depressed Classes member to Parliament on a general mandate in a constituency no separate electorate could have made more representative of the whole. Ambedkar entered the Rajya Sabha in April 1952 and served there until his death, legislating inside the architecture from its upper house. The decennial rhythm of Article 334 then began its repetitions: the extension of 1959 passed the Lok Sabha with the support of every major party, as did those of 1969, 1980, 1989, 1999, 2009 and 2019, and the debates’ recurring feature is the absence of the argument of 1932, no bench in any extension debate proposing separate electorates or abolition, the contest confined to the clock’s length and to the company reservation should keep. A safeguard renewed seven times by the institution it constrains has become part of the institution’s own furniture, which is either the architecture’s triumph or its capture, and the pillar’s remaining studies inherit that question rather than settle it.

Three Readings of the Pact, and a Verdict

The scholarship on September 1932 sorts into the three readings the brief for this pillar names, and the architectural reading wins the adjudication on the evidence of the text and the record rather than on sympathy for either signatory. The personal-confrontation reading, carried by the great biographies from Dhananjay Keer’s life of Ambedkar (1954) to Ramachandra Guha’s Gandhi II: The Years That Changed the World (2018), renders the week as a contest of wills in which Gandhi’s moral weapon met Ambedkar’s constitutional claim and the claim yielded. Its strength is fidelity to the actors’ experience: both men understood the week as a confrontation, and Ambedkar’s correspondence shows the wound it left. Its weakness is that it cannot price the document. On the confrontation reading, the difference between 71 seats and 148, between a panel and no panel, between a ten-year clock on the device and a mutual-agreement lock on the seats, is stage decoration around the duel, yet those differences are the settlement, and they are what the elections of 1937 and the Constitution of 1950 actually inherited.

The simple-compromise reading, the staple of textbook summaries, prices the document but prices it as a split difference: separate electorates surrendered here, seats conceded there, honours even. Its weakness is the panel. A compromise does not invent institutions; it trades existing goods. The primary election to a panel of four existed in nobody’s programme before 20 September 1932, and it structured Depressed Classes politics for fifteen years; a reading that files it as a detail of the trade has mistaken the settlement’s most original clause for small print. The compromise reading also fails the quantum test in the other direction: the doubling from 71 to 148 was not a midpoint between the parties’ positions, since the Hindu side’s opening position was the Rajah-Moonje formula at far fewer seats, and the Award’s figure had been set by London, not by Ambedkar. What happened at Poona was not the averaging of two demands but the design of a third thing under deadline.

The architectural reading, whose scholarly line runs from Eleanor Zelliot’s From Untouchable to Dalit (1992) through Christophe Jaffrelot’s Dr Ambedkar and Untouchability (2005) to Aishwary Kumar’s Radical Equality (2015), treats the four choices as the event and the fast as the deadline that forced them. Kumar’s treatment is the most demanding version: he reads the 1932 confrontation as a contest between two theories of equality, Gandhi’s equality as the dissolution of untouchability inside a reformed Hindu moral order and Ambedkar’s equality as a share of sovereign power guaranteed in law, and he shows the Pact recording the collision of the two theories rather than the victory of either. That is the register in which the Pact’s clauses become legible as philosophy with seat numbers attached. The architectural reading also absorbs the strongest counter to itself, which is Ambedkar’s 1945 verdict in What Congress and Gandhi Have Done to the Untouchables that the Pact was an inadequate compromise imposed by coercion. The distinction the reading insists on is between the bargain’s justice, which Ambedkar was entitled to judge and did, and the bargain’s consequences, which ran through institutions indifferent to his judgment: the panel filtered candidates in 1937 whether or not he forgave its origin, and Articles 330 and 332 reproduced the joint-electorate design in 1950 whether or not the design deserved reproduction. A founder’s indictment of his own settlement is evidence about the settlement’s origins; it is not evidence about the settlement’s effects, and the effects are the architectural reading’s subject.

The coercion charge deserves its own paragraph in the adjudication, because it is the strongest form of the counter-argument and the one Ambedkar pressed hardest. In his 1945 account the fast figures as a weapon aimed at a helpless people: the Depressed Classes were told that the price of the separate electorate was Gandhi’s life, and that the violence which would follow his death would fall on Untouchables in ten thousand villages, so that consent extracted under that shadow was no consent. Gandhi’s stated defence, in his letters and in his statements during the fast, was that the fast was addressed to caste Hindus and to his own conscience, a penance to sting the Hindu conscience awake, and that he had announced it six months before the Award, when no negotiation existed for it to coerce. The historical record supports parts of both accounts. The fast was pre-announced in March 1932, as Gandhi said; it also functioned in September 1932 exactly as Ambedkar described, compressing a constitutional negotiation into the timetable of one man’s pulse, and telegrams poured into Poona holding Ambedkar answerable for the Mahatma’s life. The architectural reading does not need to acquit the fast in order to sustain itself, because coercion explains the negotiation’s haste and asymmetry without explaining its product. Pressure can make a man sign; it cannot draft clause five’s distinction between a ten-year device and a locked guarantee, or the panel’s division of power between a community primary and a general vote. Those bear the fingerprints of the lawyers and of Ambedkar’s own bargaining over mechanism, Sapru and Jayakar shuttling drafts, Ambedkar holding out for the primary as the irreducible core. A settlement dictated by one side reads like a dictate; the Poona text reads like a negotiation in which the weaker party, bargaining under duress he never forgave, nonetheless extracted the two clauses, the panel and the quantum, that his stronger opponents spent the following decade trying to live with and the following generation quietly discarded or kept.

The readings are not antiquarian, because the week is still commemorated from opposed directions. In Maharashtra and wherever Ambedkarite organisations keep the movement’s calendar, the anniversary of the Pact in late September is marked by protest meetings and pamphlet literature that renew the 1945 indictment, treating the Pact as the founding dispossession of Dalit politics, the moment an independent electorate was traded away under threat. In the Gandhian and Congress memory, the same week is kept as a founding act of national unity and social conscience, the fast that held Hindu society together and set the Harijan programme in motion. The two commemorations cannot both be descriptions of a compromise, since a compromise does not generate protest on one side and thanksgiving on the other across the decades since. They are better evidence for the architectural reading’s premise: the week distributed real power, in clauses that still operate, and commemorations persist precisely where institutions persist, because each generation of Dalit politics must decide what to do with a design it did not choose and cannot ignore, and each generation of the majority’s politics must decide whether the design’s obligations have been discharged. Historiography follows the institution, not the other way round.

Two further exhibits discipline the verdict. The first is the sequel for the other minorities of 1932: the separate electorates the Award confirmed for Muslims, Sikhs, Christians, Anglo-Indians and Europeans were swept away by the Constituent Assembly, and the post-1947 plural-civilisational settlement refused communal electorates altogether while keeping the Pact’s compensatory design for the Scheduled Castes, a sorting that only makes sense if the republic’s founders read 1932 as the Pact’s defenders describe it, as a social-compensation architecture rather than a communal partition. The second exhibit is comparative silence: no Indian debate since 1932, however bitter, has proposed restoring separate electorates for the Scheduled Castes, while every decade has re-argued the Pact’s own variables, the quantum, the creamy layer, the sub-categorisation of the lists, the extension clock. Politics argues about the dials a design leaves exposed; it does not argue about designs it has discarded. The design of September 1932 is the one still on the table. Readers who work through the Pact’s eight clauses in the printed text, as the negotiators’ generation read it in the newspapers of that September, will find the four choices easier to track with the document open beside the commentary; scanned prints of the 1932 agreement circulate as office files, and an office file viewer for working with Excel, Word and PowerPoint documents opens them in a browser without a document suite installed.

A final test audits the settlement as its inheritors audited it, against the purposes a founding text declares for itself. The Poona Pact declared one purpose in its structure: to keep the Depressed Classes inside the common political life of the country while guaranteeing their presence in its legislatures on terms their own voters had first approved. Measured against that purpose across the first decade of the republic, the design scores unevenly in ways the map predicts. Presence it delivered absolutely: no legislature of the Indian Union has sat without Scheduled Castes members since 1952, a continuity no other safeguard of the colonial period can match. Approval it delivered only until 1946, because the panel that embodied it was allowed to lapse, and the post-1950 member’s mandate runs from his party and his general electorate, with the community’s prior consent nowhere recorded. Presence without the approval mechanism is exactly the condition Ambedkar’s 1945 book forecast and exactly the condition the joint electorate’s defenders reply to by pointing at the member’s dependence on a constituency in which his community’s votes, under adult franchise, are too numerous to insult with impunity. The audit, fairly conducted, therefore ends in the tension the design was built to hold: the architecture guarantees that the community is represented and declines to guarantee by whom, and Indian politics has been arguing the value of that bargain, in those terms, without interruption since the week it was struck.

Inside the Architecture: Mandal, the EWS Amendment and Sub-Categorisation

The proof of an architecture is the load it carries without redesign, and the reservation conflicts of the decades after 1990 were all carried by the frame of 1932 and 1950. The three-reading adjudication framework applied to the Mandal moment shows the same sorting at work: the Mandal Commission’s report of 1980 and its implementation in 1990 extended reservation from legislatures to central services for the Other Backward Classes, a new beneficiary class, through the service-side door that Article 16(4) had opened, while the legislative side of the house, Articles 330 and 332, continued to reserve for the Scheduled Castes and Scheduled Tribes alone on the Poona pattern. The Supreme Court’s judgment in Indra Sawhney v. Union of India of 16 November 1992 upheld the services extension, capped total reservation at 50 per cent and required the exclusion of the creamy layer, reasoning throughout inside the compensatory logic of May 1949 rather than against it. The question the Mandal decade argued was never the joint electorate’s legitimacy; it was the list of beneficiaries and the size of the quantum, which are precisely the variables the 1932 design leaves open to politics.

The later amendments repeat the pattern at the frame’s edges. The Constitution (103rd Amendment) Act of January 2019 created a 10 per cent reservation for the economically weaker sections in education and public employment, the first beneficiary class in the framework’s history defined by income rather than by caste, and it was carried as an addition to the services-and-education code of Articles 15 and 16, leaving the legislative architecture of 330 and 332 untouched; the sub-categorisation question, whether the Scheduled Castes’ reserved share may be divided among the castes on the list so that its benefits reach the most disadvantaged among them, argues about the distribution of the quantum inside the beneficiary class, again a variable the design exposes rather than a challenge to the design. The courts then ratified both edge-extensions inside the same logic. In Janhit Abhiyan v. Union of India, decided on 7 November 2022, a five-judge bench of the Supreme Court upheld the 103rd Amendment by a majority of three to two, holding that reservation on economic criteria alone breached neither the 50 per cent ceiling treated as inflexible nor the basic structure, and that the exclusion of the Scheduled Castes, Scheduled Tribes and backward classes from the new quota was a permissible classification within the compensatory code. In State of Punjab v. Davinder Singh, decided on 1 August 2024, a seven-judge bench held by six to one that states may sub-classify the Scheduled Castes within the reserved quota to direct benefits toward the most disadvantaged castes on the list, overruling the contrary bar laid down in E. V. Chinnaiah in 2004. Both judgments reason from the framework’s premises outward: the EWS majority treated the 1932 to 1950 architecture as a floor of compensatory justice to which new floors could be added, and the sub-categorisation majority treated the list, the architecture’s gatekeeping device, as an instrument to be refined until the quantum reached its intended beneficiaries. Neither bench entertained the proposition that reservation itself stood in need of a new foundation. The extension debates supply the pattern’s most pointed exhibit. The 104th Amendment of January 2020 did two things in one Act: it extended the reservation of seats for the Scheduled Castes and the Scheduled Tribes by a further ten years, to 2030, and it discontinued the nominated representation of the Anglo-Indian community in the Lok Sabha and the state assemblies, the last institutional survival of the Communal Award’s minorities scheme of 1932. In a single parliamentary decision, the republic renewed the Poona architecture for its ninth decade and retired the Award’s machinery for good, sorting the inheritance of 1932 into what was compensatory, and kept, and what was communal, and closed. No member in the debates of December 2019 proposed the reverse operation. Each of these contests belongs to its own study in this pillar, and each will be read there on its own evidence; what belongs here is the observation that all of them presuppose the settlement of September 1932. The joint electorate with reserved seats is the water in which Indian reservation politics swims: so completely did the Poona architecture become the republic’s default that its alternatives, separate electorates on one side and unreserved majoritarianism on the other, survive only as positions in the historiography of a single week. The Pact established joint electorates with reserved seats as the architectural framework of Indian compensatory representation; the post-1947 reservation system operates within that architecture, extends it, argues inside it, and has never replaced it.

Frequently Asked Questions

Q: What was the Poona Pact?

The Poona Pact was the agreement signed at Poona on 24 September 1932 between B. R. Ambedkar, for the Depressed Classes, and Madan Mohan Malaviya, for the caste Hindu leadership, during Gandhi’s fast unto death at Yerwada Central Jail. It replaced the Communal Award’s separate electorates for the Depressed Classes with joint electorates and reserved seats: 148 seats in the provincial legislatures, filled by all voters of each constituency from candidates of the Depressed Classes, with a primary election in which Depressed Classes voters first chose a panel of four candidates for each seat. The British government accepted the Pact on 26 September 1932 and amended the Award accordingly, and Gandhi ended his fast the same day. Its four design choices, joint electorates, the panel, the doubled quantum and a divided time limit, passed into the reservation framework of the Indian Constitution.

Q: What was the Communal Award?

The Communal Award was the scheme of minority representation announced by the British Prime Minister, Ramsay MacDonald, on 16 August 1932, after the Round Table Conferences failed to produce an agreed Indian settlement. It confirmed separate electorates for Muslims, Sikhs, Indian Christians, Anglo-Indians and Europeans, created special constituencies for commercial and labour interests, and extended separate electorates to the Depressed Classes for the first time, allotting them 71 seats in the provincial legislatures to be filled by Depressed Classes voters alone. Depressed Classes voters were also to receive a second vote in the general constituencies, and the separate-electorate arrangement for them was to lapse after twenty years. MacDonald had announced in advance that any alternative scheme agreed between the communities would be accepted in its place, and that undertaking opened the legal space in which the Poona Pact was negotiated six weeks later.

Q: Why did Gandhi go on fast in 1932?

Gandhi fasted against the Communal Award’s grant of separate electorates to the Depressed Classes, and against that provision alone. He had declared at the Round Table Conference on 13 November 1931 that he would resist the grant with his life, and he repeated the warning in a letter from Yerwada Central Jail to Sir Samuel Hoare, the Secretary of State for India, on 11 March 1932, insisting the statement was a religious one and not rhetoric. His argument was that a separate electorate would sever the Depressed Classes from Hindu society in law, hardening a social disability into a permanent political partition. He was not opposed to their representation, and he favoured registering every Depressed Classes adult as a voter regardless of property or education. When the Award was announced on 16 August 1932, he wrote to MacDonald fixing the fast from noon on 20 September 1932.

Q: What did Ambedkar agree to in the Poona Pact?

Ambedkar agreed to surrender the separate electorates the Communal Award had granted the Depressed Classes, accepting joint electorates in their place. In exchange he secured three things. The reserved quantum in the provincial legislatures rose from the Award’s 71 seats to 148, with 18 per cent of the general-electorate seats in the Central Legislature also reserved. A primary election was built into every reserved seat, under which Depressed Classes voters alone would choose a panel of four candidates, so that no Depressed Classes member could reach a legislature without first passing his own community’s vote. And the reserved seats were protected against unilateral repeal, continuing until the communities concerned agreed to end them, even though the primary-election device itself was limited to ten years. He signed the agreement on 24 September 1932, while maintaining, in statements of the following years and in his book of 1945, that the surrender had been extracted under the pressure of Gandhi’s fast.

Q: What is joint electorates with reserved seats?

Joint electorates with reserved seats is an electoral design in which a constituency’s voters of every community vote together on a common roll, while the seat itself may be held only by a candidate from a specified community. It differs from a separate electorate, in which only the specified community’s voters, on a separate roll, choose the member, and from an unreserved seat, in which any candidate may stand. The Poona Pact of 1932 adopted the design for the Depressed Classes in place of the Communal Award’s separate electorates, adding a primary election in which Depressed Classes voters shortlisted a panel of four candidates for the joint electorate to choose among. The Indian Constitution carried the design into Articles 330 and 332, reserving seats for the Scheduled Castes and the Scheduled Tribes in the Lok Sabha and the state assemblies inside general constituencies, though without the Pact’s primary-election stage.

Q: How did the Poona Pact shape Indian Constitution?

The Constitution of 1950 took its reservation architecture from the Pact clause by clause. Articles 330 and 332 reproduce the Pact’s joint electorates with reserved seats, extending them to the Scheduled Tribes and fixing the quantum by population proportion rather than by a negotiated seat list. Article 334 reproduces the Pact’s time-limit logic in altered form, setting reservation to lapse after ten years unless Parliament extends it, as Parliament has done seven times, carrying the provision to 2030. Article 335 carries the Pact’s clause on the claims of the Depressed Classes in the public services into the constitutional text, and Article 17, abolishing untouchability, enacts the pledge of the Hindu leadership’s resolution of 25 September 1932. The one Pact device the Constitution dropped was the primary election to a panel of four, leaving reserved seats to be filled by the joint electorate and the party ticket alone.

Q: Was the Poona Pact a defeat for Ambedkar?

Ambedkar’s own verdict, in What Congress and Gandhi Have Done to the Untouchables (1945), was that the Pact was a defeat imposed by coercion: the fast, he argued, confronted the Depressed Classes with Gandhi’s life as the price of their separate electorate, and the joint electorates then delivered representatives who were nominees of caste Hindu parties. On the measure of his London demand, the verdict holds, since the separate electorate he had won in the Communal Award was signed away. On the measure of the settlement’s text and afterlife, the account is mixed: he doubled the reserved quantum to 148 seats, extracted the primary election that gave Depressed Classes voters a veto over the shortlist, and locked the seats against unilateral repeal, and the Constitution of 1950 was built inside the design he had negotiated. Historians from Eleanor Zelliot to Christophe Jaffrelot have therefore treated the Pact as a defeat in the negotiation and a foundation in the sequel.

Q: Who signed the Poona Pact?

The Pact was signed on 24 September 1932 by B. R. Ambedkar on behalf of the Depressed Classes and by Madan Mohan Malaviya on behalf of the caste Hindu leadership. Gandhi did not sign: he was a prisoner in Yerwada Central Jail, fasting, and the agreement was negotiated to a text he could accept and communicated to him at the jail, where Ambedkar met him during the negotiations. The mediation that produced the text was conducted by Tej Bahadur Sapru, M. R. Jayakar and Ghanshyam Das Birla, with C. Rajagopalachari working at Gandhi’s side, and the Depressed Classes cricketer Palwankar Baloo helped organise the meetings at Yerwada between Ambedkar’s camp and the Hindu leadership. The agreement took legal effect when the British government announced its acceptance on 26 September 1932, substituting the Poona terms for the Communal Award’s Depressed Classes clauses, whereupon Gandhi broke his fast.

Q: How many seats did the Poona Pact reserve for the Depressed Classes?

The Pact reserved 148 seats for the Depressed Classes in the provincial legislatures, distributed as Madras 30, Bengal 30, the United Provinces 20, the Central Provinces 20, Bihar and Orissa 18, Bombay with Sind 15, Punjab 8 and Assam 7. The figure more than doubled the 71 seats the Communal Award had allotted a month earlier, and the increase was the caste Hindu negotiators’ substantive payment for Ambedkar’s surrender of separate electorates. Clause four of the Pact added a central provision, reserving 18 per cent of the seats allotted to the general electorate for British India in the Central Legislature, to be filled by the same joint-electorate and panel procedure. The provincial figures were based on the strengths of the councils announced in the Award and tracked, roughly, the Depressed Classes share of each province’s population, a proportionality the Constitution later generalised into the formula of Articles 330 and 332.

Q: What was the panel of four in the Poona Pact?

The panel of four was the Pact’s primary-election device for reserved seats. Under clause two, all Depressed Classes voters registered on the general electoral roll of a constituency formed an electoral college, which voted by single vote to select four Depressed Classes candidates for each reserved seat; the four candidates with the highest votes in that primary then stood before the joint electorate, which made the final choice. The device divided the power of selection between the two electorates: the community controlled the shortlist, so no candidate its voters had rejected could be imposed on the seat, and the general electorate controlled the outcome among the four. The panel operated in the provincial elections of 1937 and 1946 under the Government of India Act 1935. The Constitution of 1950 dropped it, so reserved-seat candidates have since been chosen by party ticket and the joint electorate without the community’s preliminary vote.

Q: What was the Rajah-Moonje Pact of 1932?

The Rajah-Moonje Pact was an agreement of February 1932 between M. C. Rajah, a Depressed Classes leader from Madras, and B. S. Moonje of the Hindu Mahasabha, under which Rajah accepted joint electorates with reserved seats for the Depressed Classes in place of separate electorates. The Hindu Mahasabha publicised the agreement as evidence that the Depressed Classes themselves preferred joint electorates, and it supplied the formula the Hindu leadership carried into the Poona negotiations seven months later. Ambedkar repudiated it at the time and afterward, arguing that Rajah did not carry the Depressed Classes’ mandate and that a settlement signed by a leader without a following settled nothing. The episode mattered in September 1932 because it proved a text existed that the Hindu side could sign, while fixing the question the fast week had to answer: whether Ambedkar, whose following Rajah lacked, could be brought to sign a version of it with the panel and the doubled quantum added.

Q: What happened to the primary election system after 1947?

The primary election to a panel of four lapsed without renewal and was not carried into the Constitution. Clause five of the Poona Pact had limited the device to ten years unless the communities agreed to continue it, and no such agreement was made; the device operated in the provincial elections of 1937 and 1946 under the Government of India Act 1935 and then expired as a living institution. When the Constituent Assembly framed Articles 330 and 332, it retained reserved seats inside joint electorates but filled them by the ordinary party ticket and general vote, reasoning that universal adult franchise had changed the arithmetic that made a preliminary community veto necessary: under a universal roll, Scheduled Castes voters formed a substantial bloc in every reserved constituency. The change shifted the community’s power from a formal veto over the shortlist to bargaining weight inside party organisations, a shift critics in the Ambedkarite tradition have treated as the quiet reversal of 1932’s one unambiguous Depressed Classes safeguard.

Q: How long was legislative reservation originally meant to last?

The Poona Pact set two different periods for its two devices. The primary election to a panel of candidates was to end after the first ten years unless terminated sooner or continued by mutual agreement. Reserved seats themselves carried no fixed term: clause six provided that representation by reserved seats should continue until determined by mutual agreement between the communities concerned, which protected the seats against repeal by a future majority acting alone. The Constitution replaced the mutual-agreement lock with a parliamentary clock. Article 334 provided that reservation of seats for the Scheduled Castes and the Scheduled Tribes would cease ten years from the commencement of the Constitution, in 1960, unless Parliament extended the period. Parliament has extended it seven times, by the 8th, 23rd, 45th, 62nd, 79th, 95th and 104th Amendments, each time for a further ten years, carrying the current terminus to 2030.

Q: What did Ambedkar argue in What Congress and Gandhi Have Done to the Untouchables?

Ambedkar’s book of 1945, published in Bombay, is the fullest contemporary indictment of the Poona settlement by its own signatory. Its chapters on the fast and the Pact argue that Gandhi’s fast was directed not at the British government but at the Depressed Classes, whose one moment of constitutional leverage it destroyed, and that the joint electorate then produced the result Ambedkar had predicted in 1931: reserved-seat members chosen by caste Hindu party organisations and answerable to them, present in the legislatures in the community’s name and silent on the community’s questions. The book marshals the record of the Congress ministries of 1937 to 1939 as its evidence. Read beside the Pact’s text, the book is best treated as two documents at once: a participant’s testimony about the coercion of September 1932, which historians continue to weigh, and a polemic whose forecast about reserved representation the subsequent record has partly confirmed and partly overtaken.

Q: Why did Ambedkar found the Independent Labour Party?

Ambedkar founded the Independent Labour Party in August 1936 to fight the first provincial elections held under the Government of India Act 1935, due in early 1937. The party’s programme joined the Depressed Classes’ cause to a labour platform, opposing landlord power in the countryside, including the khoti system of the Konkan, and demanding state action on workers’ conditions, which allowed Ambedkar to contest general seats alongside reserved ones and to build the ward-level organisation the Pact’s primary elections rewarded. In the Bombay elections of 1937 the party won 15 of the 17 seats it contested, eleven of them reserved seats won through the panel procedure, and Ambedkar entered the Bombay Legislative Assembly as the leader of a disciplined bloc. The result demonstrated the strategy behind the founding: inside the Poona architecture, representation belonged to whichever Depressed Classes organisation contested the primaries with the deepest local roots.

Q: What is the difference between the Communal Award and the Poona Pact?

The two documents of August and September 1932 allocated the same subject, Depressed Classes representation, by opposite machinery. The Communal Award of 16 August 1932 gave the Depressed Classes separate electorates: 71 provincial seats filled by Depressed Classes voters alone, on a separate roll, with a second vote in the general constituencies, the arrangement to run for twenty years. The Poona Pact of 24 September 1932, agreed between the communities and substituted by the British government for the Award’s Depressed Classes clauses, gave joint electorates with reserved seats: 148 provincial seats inside general constituencies, filled by all voters jointly from Depressed Classes candidates shortlisted through a primary election of Depressed Classes voters. The Award treated the Depressed Classes as a separate political element akin to the religious minorities; the Pact treated them as a disadvantaged section of a single electorate whose presence the constitution would guarantee. Every later Indian reservation provision descends from the Pact’s model, not the Award’s.

Q: Did the Poona Pact abolish untouchability?

The Pact itself did not address untouchability: its clauses concern electorates, seats, a primary election, service claims and an education grant, and the practice of untouchability appears nowhere in its text. The abolition pledge of September 1932 stands in a companion document, the resolution adopted at a public meeting in Bombay on 25 September 1932, in which the Hindu leadership declared that no one should thereafter be regarded as an untouchable by reason of birth and that the Depressed Classes should have equal rights in public wells, schools, roads and temples. The All India Anti-Untouchability League, founded on 30 September 1932 and renamed the Harijan Sevak Sangh, carried the pledge into campaign work. Legal abolition arrived with the Constitution: Article 17 abolished untouchability and forbade its practice in any form, a provision the Constituent Assembly adopted from the reform programme the fast had set in motion rather than from the electoral text of the Pact.

Q: What was Gandhi’s letter to Samuel Hoare of March 1932?

Writing from Yerwada Central Jail on 11 March 1932 to Sir Samuel Hoare, the Secretary of State for India, Gandhi put the British government on notice that his declaration at the Round Table Conference, that he would resist with his life the grant of a separate electorate to the Depressed Classes, had not been said in the heat of the moment or by way of rhetoric and was meant as a religious statement. The letter set out his position in full: he favoured registering every Depressed Classes adult, male and female, as a voter irrespective of education or property qualification, but held a separate electorate harmful to the Depressed Classes and to Hinduism, since it would fix their separation in law. He concluded that if separate electorates were created he would fast unto death. The letter matters because it dates the fast six months before the Communal Award, which is the foundation of the defence that the fast was a pre-announced vow rather than an improvisation to coerce the September negotiations.

Q: How did the Poona Pact treat education and the public services?

Two clauses of the Pact reach beyond the legislatures. Clause seven provided that there should be no disabilities attaching to membership of the Depressed Classes in regard to elections to local bodies or appointment to the public services, and that every endeavour should be made to secure fair representation of the Depressed Classes in those respects, subject to the qualifications the services required. Clause eight directed that in every province a sum be set apart from the educational grant for the educational facilities of the Depressed Classes. Neither clause carried a percentage or an enforcement machinery, and both depended on the provincial governments the Pact’s electoral clauses were busy composing. Their constitutional descendants are stronger than the originals: Article 335 of the Constitution carries the claims of the Scheduled Castes and the Scheduled Tribes to services and posts into the text itself, and Article 16(4) empowers reservation in public appointments for backward classes inadequately represented in the services.

Q: Why is the Poona Pact still debated?

The Pact is debated because its machinery still operates and its bargain still distributes power. The joint electorates with reserved seats that it designed in September 1932 stand in Articles 330 and 332 of the Constitution, renewed under Article 334 into the present century, and the arguments of 1932 recur inside its clauses: whether a reserved-seat member answers to his community or to his party, whether the quantum should grow or be sub-divided among the castes on the list, and whether the safeguard should continue at all. The two commemorations of the week state the unresolved verdicts: Ambedkarite organisations mark the anniversary as a dispossession, the surrender of an independent electorate under the threat of the fast, while the Gandhian memory keeps it as the settlement that held Hindu society together and launched the campaign against untouchability. A document whose clauses decide elections and whose anniversary draws protest has not passed into history in the ordinary sense.