The 2002 Gujarat riots politics, as the phrase is used in textbooks, examination halls and anniversary journalism, usually names one of two things: an election, or a wound. In the electoral telling, the violence of February and March 2002 is the prelude to the Gujarat assembly election of December 2002, in which the Bharatiya Janata Party won 127 of 182 seats and Narendra Modi, chief minister since October 2001, secured the mandate that began his uninterrupted rise. In the narrative telling, the same months are a sequence of atrocities: a train coach burned at Godhra station on 27 February 2002 with 59 people inside, and massacres at Gulberg Society, Naroda Patiya, Naroda Gam, Sardarpura and Ode in the days that followed. Both tellings are true as far as they go, and both stop early. Between 6 March 2002, when the Gujarat government constituted a commission of inquiry under a retired judge named K. G. Shah, and 8 January 2024, when the Supreme Court quashed the remission of the eleven men convicted in the Bilkis Bano case, the Indian state examined the events of 2002 through a commission, a court-monitored investigation team, a sequence of special courts and a criminal complaint that travelled from a police director general’s office to a three-judge bench of the Supreme Court. That twenty two year audit, and not the election or the wound alone, is the subject of this article.

The claim defended here is namable: the twenty two year institutional audit of the 2002 Gujarat violence produced a settled body of findings, on the Godhra fire, on the massacres, on the conduct of the state government and on the limits of the evidence against its chief minister, and the national political order built after 2014 operates through those findings, invoking them as institutional validation and governing inside the space they left unjudged. The audit reading does not deny the electoral reading its December 2002 verdict, nor the narrative reading its dead. It insists that a democracy’s considered answer to an atrocity is the sum of what its institutions found, failed to find, and declined to decide, and that this answer, document by document, is retrievable. Three bodies of writing contest the ground. The political-electoral reading, given scholarly form in Christophe Jaffrelot’s Modi’s India (2021), treats 2002 as the founding episode of a majoritarian order and follows the consolidation that flowed from it. The documentary reading, assembled within months in Siddharth Varadarajan’s edited volume Gujarat: The Making of a Tragedy (2002), treats the violence itself, and the administrative conduct surrounding it, as the whole story. The audit’s internal critics, most rigorously Manoj Mitta in Modi and Godhra: The Fiction of Fact-Finding (2014), hold that the fact-finding machinery was arranged so as to reach exoneration, and that its findings are therefore fictions of process. This article adjudicates among the three on the primary record: the reports of the Nanavati and Mehta Commission, the closure report of the Special Investigation Team filed on 8 February 2012, and the Supreme Court judgment in Zakia Jafri v State of Gujarat delivered on 24 June 2022. Its verdict, argued in the sections that follow, is that the audit’s findings are real as institutional facts and incomplete as moral ones, that the incompleteness is itself documented, and that the post-2014 order lives precisely in the gap between the two.

Three Readings of One Event, and Why the Audit Reading Governs

The electoral reading begins from an undeniable sequence. On 19 July 2002 the Gujarat assembly was dissolved nine months before its term expired, and the state government sought an early election on the argument that a fresh mandate would restore normalcy. The Election Commission, headed by Chief Election Commissioner J. M. Lyngdoh, refused the timetable, recorded in its order of 16 August 2002 that free and fair elections were not yet possible, and weathered a Presidential reference to the Supreme Court on the meaning of the six month rule in the Constitution before polling was held on 12 December 2002. The BJP won 127 seats on about 50 percent of the vote, and Modi, who had campaigned on the security of Gujarati Hindus after Godhra, became the longest serving chief minister in the state’s history and, in May 2014, prime minister. On this reading, 2002 is a story about what violence does to a party system: it polarised the electorate, marginalised the Congress in Gujarat for a generation, and supplied the governing template of the premiership that followed. Jaffrelot’s account adds the sociological depth the election returns alone cannot carry: his chapters on the post-2002 years describe an order in which Hindu consolidation, a weakened opposition and a disciplined party apparatus reinforced one another, and in which the events of 2002 functioned as both origin and warning.

The narrative reading begins earlier and ends sooner. Its archive is the record of February and March 2002 itself: the burning of coach S-6 of the Sabarmati Express at Godhra on 27 February, the killing of 69 people at Gulberg Society on 28 February, among them the former Congress member of Parliament Ehsan Jafri, the killing of 97 people at Naroda Patiya the same day, the attacks at Sardarpura, Ode and hundreds of other sites, and the relief camps in which more than one hundred thousand displaced persons, overwhelmingly Muslim, sheltered through the spring. Varadarajan’s volume, published in 2002, gathered journalists’ reports, official statements and survivors’ testimony while the events were months old, and it fixed the terms in which the violence would be debated: whether the massacres were spontaneous reprisals, as the state government maintained, or organised attacks conducted with the knowledge, and in places the assistance, of the administration, as the volume’s contributors argued. The National Human Rights Commission, chaired by Justice J. S. Verma, entered the same terrain on the state’s behalf and against it, reporting on 31 May 2002 that there had been a comprehensive failure of the state to protect the constitutional rights of the people of Gujarat. On this reading, everything after 2002 is epilogue: investigations, trials and commissions are judged by whether they honour or betray the record of those weeks.

The audit reading takes the epilogue as the text. It observes that the Indian republic answered 2002 not with one inquiry but with a layered architecture of them, built by different principals for different purposes across twenty two years: a state-appointed commission of inquiry that sat from 2002 to 2014 and whose final volume entered the public record only in December 2019; a set of criminal trials first stayed, then transferred, then re-investigated under the Supreme Court’s supervision from March 2008; a citizens’ tribunal that delivered its own findings within the year; and a single citizen’s complaint, filed by Zakia Jafri on 8 June 2006, which alleged a larger conspiracy at the summit of the state and which the legal system processed, stage by stage, until 24 June 2022. The audit reading’s wager is that this architecture is itself the most durable political fact of the episode. Elections turn over; anniversaries fade; but the findings of the Nanavati and Mehta Commission, the closure report of the Special Investigation Team and the judgment of 24 June 2022 remain on the record, citable and cited, and the national government formed in 2014 has treated them exactly so. Readers who know the demolition audit of 1992 will recognise the family resemblance: as the institutional-political-audit-trajectory reading of the Liberhan Commission showed for Ayodhya, a commission’s report can settle an institutional record without settling a country’s argument about the event itself.

The audit reading also has a discipline the other two lack. It can state exactly what each institution found, what each declined to find, and what each was never asked. The commission found no evidence that any minister inspired or abetted the attacks; the investigation team found no prosecutable evidence against the sixty four persons named in Zakia Jafri’s complaint; the trial courts convicted hundreds of foot soldiers and some local organisers, and acquitted, in the end, every senior political figure tried before them; the Supreme Court in 2022 upheld the closure of the conspiracy question as a matter of law. Whether one receives these findings as vindication or as failure depends on a prior judgement about the machinery, and that judgement is the disagreement this article must adjudicate. Mitta’s book is the strongest statement of the failure case, and it is not a pamphlet: it tracks the investigation team’s internal choices, the divergence of the court’s own amicus curiae, and the treatment of witnesses, and it concludes that the outcome was built into the process. The vindication case rests on the same documents read the other way: three successive forums, a magistrate in December 2013, the Gujarat High Court in October 2017 and the Supreme Court in June 2022, each examined the conspiracy allegation and each declined to disturb the closure, while the trial courts’ convictions in the massacre cases stand as proof that the machinery could and did convict where evidence survived. The sections below lay the record out stage by stage, and the adjudication comes after the evidence, where it belongs.

The Documented Record: Godhra, 27 February 2002, and the Violence That Followed

On the morning of 27 February 2002, the Sabarmati Express, running from Ayodhya toward Ahmedabad and carrying Hindu pilgrims and kar sevaks returning from the temple town, stopped at Godhra station in the Panchmahal district of eastern Gujarat. After an altercation on the platform, the train pulled away and halted again a short distance from the station, where a crowd had gathered. Coach S-6 was set on fire. Fifty nine people died in the coach, burned or asphyxiated; the dead were predominantly Hindu pilgrims, many of them women and children. What caused the fire, and whether it was planned, became the first disputed question of the entire episode, and it remained disputed through two official examinations that reached opposite conclusions. The Gujarat police charged a conspiracy: a crowd assembled with intent, petrol procured in advance, the coach doused and lit. The Nanavati Commission’s first report, submitted in September 2008, accepted the conspiracy account and named a cleric, Maulvi Hussein Haji Ibrahim Umarji, as its chief organiser. A committee appointed by the Railway Ministry under Justice U. C. Banerjee reported in January 2005 that the fire had most probably begun inside the coach by accident, a conclusion the Gujarat High Court struck down in October 2006 on the ground that the committee had been constituted without legal authority. The criminal courts, in the end, decided the matter: a special court at Ahmedabad convicted 31 accused of conspiracy and murder on 1 March 2011 and acquitted 63, and the Gujarat High Court, commuting eleven death sentences to life imprisonment on 9 October 2017, left the conspiracy finding standing. The fact that the fire was a crime, deliberately set, is therefore the settled position of the judicial record; the Banerjee committee’s accidental-fire thesis survives as a documented dissent, not as a finding any court has adopted.

Between those two institutional answers lay a forensic dispute that the audit never quite closed and that a careful reader should know exists. The Gujarat Forensic Science Laboratory’s examination of the burned coach, reported in 2002, tested the first police account, in which a large quantity of petrol had been poured into the coach from outside through its doors and windows, and concluded that the physical evidence did not support an accelerant attack of that form and scale. Each side built on the laboratory’s caution what it needed: the Banerjee committee treated the forensic findings as corroboration of an internal, accidental ignition; the prosecution, in the case it finally carried to trial, rested instead on eyewitness testimony and on accused persons’ accounts of petrol carried in containers and used at the scene, a mechanism the trial court accepted when it convicted in 2011. The Nanavati Commission, for its part, heard the forensic witnesses and maintained its conspiracy conclusion on the assembled testimony. The three treatments of one laboratory report, adopted, circumvented and absorbed, are a miniature of the whole audit: the same primary evidence, three institutional readings, and a judicial outcome that resolved the legal question while leaving the documentary tension permanently visible in the record for any later reader to weigh.

The Gujarat government of Narendra Modi, in office since 7 October 2001, responded within hours. Curfew was imposed in Godhra and then in city after city. The Vishwa Hindu Parishad called a Gujarat bandh for 28 February, which the BJP supported. The bodies of the Godhra dead were taken to Ahmedabad, a decision the government’s critics, among them the editors of the Varadarajan volume, identified as an incitement and the government defended as a requirement of identification and post-mortem procedure. On 28 February the violence became general in Ahmedabad and spread across the state. At Gulberg Society, a Muslim housing colony in the Chamanpura area, a crowd besieged the home of Ehsan Jafri through the day; Jafri, who had represented Ahmedabad in the Lok Sabha, telephoned officials and acquaintances for help that did not arrive in time, and he was killed with 68 others. At Naroda Patiya, on the city’s eastern edge, 97 people were killed, the largest single massacre of the episode; at Naroda Gam, a short distance away, 11 were killed. In Mehsana district, 33 people were killed at Sardarpura on the night of 1 March; at Ode in Anand district, mobs killed residents in two connected attacks, 23 of them burned in a single house. Attacks on Muslim homes, businesses and places of worship were reported from some twenty one districts, and there were also attacks on Hindus by Muslim crowds in localities of Ahmedabad and other towns, recorded in the government’s own reporting and in the commission’s volumes. The army was called out and staged flag marches from 1 March; Defence Minister George Fernandes toured the state; police firing killed a substantial number of rioters, a fact the state government cited in its defence and the official toll bears out in part. The killing continued at a declining rate through March and flared intermittently for weeks; relief camps held more than one hundred thousand people at their peak.

How many people died in the Gujarat violence of 2002?

The Government of India told the Rajya Sabha on 11 May 2005 that 1,044 people were killed in the post-Godhra violence: 790 Muslims and 254 Hindus, with 223 missing and about 2,500 injured. The Concerned Citizens Tribunal estimated 1,926 deaths in 2002, and other civil society counts reach about 2,000; each figure is an estimate by its named source.

The counting dispute is instructive, because it prefigures every later dispute in the audit. The official figure of 1,044 counts identified bodies registered by the state; the tribunal’s figure of 1,926 adds the testimony of survivors and camp registers; the difference between them is a difference of method, and both figures are quoted in this article only with their sources attached. The dead of Godhra, 59 in number, stand outside both totals in most presentations, a bookkeeping choice that itself carries an argument. What no source disputes is the distribution: the great majority of the dead were Muslim, and the majority of the Hindu dead after Godhra fell in the first days, many to police firing. The violence also destroyed property on a scale the state’s own reports quantify: police records list hundreds of mosques, dargahs, temples and churches damaged, thousands of homes and businesses burned, and an economy in the affected districts interrupted for weeks. A reader who wants the state’s contemporaneous answer to the question of responsibility should begin not with the parties’ statements but with the National Human Rights Commission’s report of 31 May 2002, which held, in a sentence this article quotes because every later institution had to answer it, that there was a comprehensive failure of the state to protect the constitutional rights of the people of Gujarat.

Why was Article 355 debated during the violence of 2002?

Article 355 obliges the Union to protect every state against internal disturbance and to ensure that its government is carried on under the Constitution. Opposition parties demanded in February and March 2002 that the Vajpayee government invoke it, and some demanded President’s Rule under Article 356. The Union government declined both steps.

The constitutional debate mattered because it fixed the centre’s position as a matter of record at the very opening of the audit. The government of Atal Bihari Vajpayee depended on coalition partners, several of whom, the Telugu Desam Party most audibly, expressed public disquiet; the Janata Dal (United) leader George Fernandes, as defence minister, became the government’s emissary and defender in the state. In the Lok Sabha debate of April 2002 and the Rajya Sabha debate that followed in May, where the opposition’s censure motion was defeated, Home Minister L. K. Advani defended the Gujarat government’s conduct and the centre’s restraint, and the treasury benches framed the violence as a reaction to Godhra that the state was bringing under control. The context was the post-1998 NDA centre-government context, in which a coalition government led by the BJP held office in Delhi while its party held Gujarat, and every opposition speaker drew the inference that the centre would not discipline its own. Prime Minister Vajpayee, visiting the Shah Alam relief camp in Ahmedabad on 4 April 2002, addressed the chief minister in public with a reminder that has outlived every other sentence of that spring: a ruler, he said, must follow rajdharma, the duty of governance, and rajdharma admits no discrimination between subjects. The remark was read by the government’s critics as a rebuke and by its defenders as a general exhortation; Vajpayee declined to clarify it into either. Eight days later, at the BJP national executive in Goa on 12 April 2002, Modi offered his resignation, the party declined it, and his continuation in office became the settled political fact around which the entire audit would revolve.

The Two Positions of 2002, Each at Its Strongest

Every audit begins from a dispute, and the dispute of 2002 was stated completely within the first ninety days. The state government’s position, set out by Modi in the Gujarat assembly, by Advani in Parliament, and by the state’s counsel before every later forum, had four elements. The violence was a spontaneous, unplanned reaction to the Godhra burning, of a kind no administration confronting a statewide bandh and crowds in hundreds of localities could have prevented outright. The administration responded with the instruments it had: curfews in dozens of towns, the army called out within a day, police firing that killed rioters in numbers the toll records. The chief minister appealed for peace from 27 February onward, and the charges against him rested on inference, on the testimony of interested parties, and on a delay of hours that the fog of a burning state explained. And the demand for his removal, pressed within a fortnight of the violence, was a political demand dressed as a constitutional one, aimed at a mandate the opposition had failed to win in 1998. Each element was supported by something real: the speed of the army’s deployment, the police firing figures, the absence, in the first investigations, of any document ordering inaction, and the electoral fact that the Congress had lost Gujarat three times running. The position also had a jurisprudential corollary that the government stated plainly and repeated for twenty years: in a court of law, suspicion is not evidence, and no prosecutable evidence against the chief minister had ever been produced.

The critics’ position, assembled by the National Human Rights Commission, by the Concerned Citizens Tribunal chaired by Justice V. R. Krishna Iyer and Justice P. B. Sawant, by the Editors’ Guild fact-finding mission of May 2002 and by the scholars of the Varadarajan volume, had four matching elements. The massacres bore the marks of organisation: crowds arriving with lists of Muslim properties drawn from official records, transport arranged, targets selected with a precision that spontaneity could not explain, a pattern the political scientist Paul Brass would have recognised from his work on institutionalised riot systems, where violence is produced by networks that connect politicians, criminals and police. The state’s response was uneven in a patterned way: officers who acted firmly, like Rahul Sharma in Bhavnagar, were transferred mid-crisis, and ministers were alleged by witnesses to have been present in police control rooms. The relief effort discriminated: compensation for the Godhra dead was announced at two hundred thousand rupees and for the riot dead at one hundred thousand, a differential withdrawn only after protest. And the chief minister’s own words, above all his later campaign rhetoric about relief camps, supplied the motive the documents lacked. The critics’ jurisprudential corollary matched the government’s in structure and opposed it in sign: where the state controls the police, the prosecution and the record, the absence of prosecutable evidence is a product of the system, not a finding about the facts.

Between these two positions stood the officers and witnesses whose testimony would occupy the audit for two decades. R. B. Sreekumar, the state’s additional director general of police for intelligence, filed affidavits before the Nanavati Commission alleging that officials had been instructed to let Hindus vent their anger; the state government rejected the affidavits as the fabrication of a disgruntled officer denied promotion, and both the commission and the investigation team, examining him years apart, declined to act on his account. Sanjiv Bhatt, a police officer posted in the state intelligence branch, swore in an affidavit of April 2011 that he had attended a meeting at the chief minister’s residence on the evening of 27 February 2002 at which Modi had said that Hindus should be allowed to express their anger; the state government replied that Bhatt was not present at the meeting, the investigation team tested his claim against the attendance record and found it unproved, and the commission likewise found his evidence unreliable. The two men’s claims, and the institutions’ treatment of them, became the audit’s most contested evidentiary question, and they are recorded here as the critics state them and as the record answered them, because the audit reading forbids resolving by adjective what the institutions resolved by finding. What can be said plainly is chronological: the state government’s position was stated in full by the summer of 2002, the critics’ position was stated in full by November 2002, when the Concerned Citizens Tribunal published Crime Against Humanity, and the twenty years that followed were spent by India’s institutions deciding, for specified purposes and under specified rules, which parts of each position could be converted into findings.

Parliament, the Press and the World: The Audit’s Public Stage

The first forum to adjudicate the two positions was not a commission or a court but Parliament, and its debate, held in April and May 2002 while relief camps were still full, supplied the vocabulary the later audit inherited. In the Lok Sabha, Home Minister L. K. Advani defended the Gujarat government and the centre’s decision against President’s Rule, arguing that the violence, however grave, was a state law and order failure of a kind that central intervention would not cure, and the opposition’s censure motion was defeated. The Congress, the Left parties and the allies of the ruling coalition who dissented from within it, among them the Samata Party’s George Fernandes in his capacity as defence minister, spoke to different registers of the same facts: the Congress demanded the chief minister’s removal, the Left demanded central rule, and Fernandes, returned from his tours of the state, defended the administration’s response while conceding that the first three days had outrun the police. In the Rajya Sabha the debate ran the same course to a partisan division. What the parliamentary record established, for the audit that followed, was the framing of the state government’s case in its strongest institutional form: not a denial that violence had occurred, but a claim that it was riot rather than pogrom, failure rather than design, and that the remedy lay in elections and courts rather than dismissal. The opposition’s case, in its strongest form, was the mirror: that the pattern of targets, the timing of the bandh, and the unevenness of the response described design, and that design made the chief minister’s continuance intolerable. The commission, the investigation team and the courts would be asked, in turn, to test both framings on evidence, and the parliamentary debate’s value to the audit is that it stated them before any forum had a reason to soften them.

The press formed the second public stage, and its own audit arrived within months. The Editors’ Guild of India, chaired by Aroon Purie, sent a fact-finding team to Gujarat in May 2002 and reported that parts of the Gujarati press had published inflammatory and, in some cases, fabricated material at the height of the violence, the report’s central instance being the Gujarati daily Sandesh, whose coverage in the days after Godhra carried claims about abductions and assaults at the station that the Guild found unsubstantiated and which the paper’s defenders described as reports received in good faith and later corrected. The national press, by contrast, carried the critics’ case in detail from March onward, and the Gujarat government treated much of that coverage as hostile, a complaint its supporters repeated for years in the form of the charge that English language journalism had tried the chief minister before any court. The asymmetry matters to the audit reading because the two press worlds produced two archives: the English archive, in which the Human Rights Commission report, the tribunal and the transfer petitions circulated as authority, and the Gujarati archive, in which the Godhra dead, the bandh and the threat of further attacks circulated as the governing memory. When the 2007 Tehelka recordings later entered the trial record, and when the Supreme Court’s judgments were read in 2012 and 2022, each archive heard the same document differently, and the audit’s documents were received by two publics that had long stopped sharing a description of the event.

The world formed the third stage. In March 2005 the United States Department of State denied Narendra Modi a diplomatic visa and revoked an existing tourist visa, under the provision of the Immigration and Nationality Act that bars foreign officials responsible for particularly severe violations of religious freedom, citing the findings of the United States Commission on International Religious Freedom on Gujarat. The denial, imposed on a sitting chief minister who had travelled to the United States before, was upheld through the Bush and Obama administrations, and the British government pursued a policy of non-engagement with Modi until October 2012, when the British High Commissioner James Bevan met him in Gandhinagar and the policy was formally ended. The European Union’s member states followed a similar restraint. The sanctions’ significance for the audit is not diplomatic but evidentiary: the visa denial rested on the same documentary record the audit was assembling, the Human Rights Commission report, the tribunal, the journalistic investigations, and it survived for nine years on that record alone, while the Indian courts had not yet pronounced on the conspiracy question. When Modi became prime minister in May 2014, the United States issued its invitation within weeks and received him in Washington in September 2014, the visa chapter closing on the election that the audit reading treats as the first institutional verdict. The international record therefore confirms the audit reading’s central claim in an unusual way: for nine years, the world treated the audit’s incomplete findings as sufficient for a consequential judgment, and then treated the election as sufficient to supersede them, the same sequence of institution, verdict and supersession that the domestic audit enacted at greater length.

The December 2002 Election and the Limits of the Electoral Reading

What did the Election Commission decide about the Gujarat election in August 2002?

The Election Commission under J. M. Lyngdoh, visiting Gujarat in August 2002 after the assembly’s dissolution, ordered on 16 August 2002 that elections could not be held, recording that law and order and the electoral rolls barred a free and fair poll. A Presidential reference on the Constitution’s six month rule followed; polling was held on 12 December 2002.

The election that resulted is the electoral reading’s central exhibit, and it deserves to be stated exactly. Polling took place on 12 December 2002; the results, declared on 15 December, gave the BJP 127 of the assembly’s 182 seats and the Congress 51, on a turnout of about 62 percent. Modi, sworn in again on 22 December 2002, had campaigned as the defender of Gujarati pride against what he called the defamation of five crore Gujaratis, and the campaign’s most contested episode, his September 2002 remarks at Becharaji about relief camps, was defended by his party as a statement about demographic anxiety and condemned by his opponents as a slur on the displaced; the Election Commission reprimanded the campaign’s tone, and the phrase entered the audit’s record as evidence of rhetoric, never of instruction. The verdict’s meaning was then argued in the terms that the contested-settlement framework makes familiar: for the BJP, the electorate of Gujarat had heard the charges and returned its verdict; for the Congress and the critics, a majority verdict in a polarised electorate answered a political question and left the criminal and administrative questions exactly where they stood.

The electoral reading explains the consolidation that followed better than any rival: Gujarat’s subsequent elections, the marginalisation of the state Congress, the construction of a development-centred second Modi persona after 2007, and the national campaign of 2014, in which the Gujarat record was presented as a credential. The two intervening state elections mark the audit’s electoral shadow with precision. In December 2007, Modi fought on Gujarati pride and the first version of the development record, and the Congress, unwilling to centre a campaign on 2002 five years after losing on its aftermath, raised the violence obliquely; its president Sonia Gandhi, speaking at Navsari in December 2007, described the state’s rulers as merchants of death, a phrase the BJP converted into an asset within the week, and the party won 117 seats. In December 2012 the pattern repeated in quieter form: the BJP won 115 seats, the Congress again declined to make the pending prosecutions its theme, and the chief minister’s swearing-in that month was attended by the national leadership of a party preparing his elevation. The state electorate’s three successive verdicts, 2002, 2007 and 2012, were delivered while the audit’s criminal stages were still open, the commission still sitting, the investigation team still reporting, the protest petition still unheard, and the sequence cuts in both analytical directions at once. For the government, the verdicts ratified the audit’s direction before the audit concluded; for the critics, the verdicts demonstrate how an electorate can settle the political question while the legal one remains open, leaving the institutions to audit an event whose political meaning had already been fixed by the only forum that had answered. What it cannot explain is the audit’s own persistence. A purely electoral story would predict that the inquiries died with the verdict of December 2002, their purpose exhausted. Instead the commission sat for twelve more years, the trials were stayed, transferred and re-investigated, the courts intervened at every stage, and a widow’s complaint filed in 2006 consumed the Supreme Court’s attention until 2022. The institutions behaved as though something remained undecided that elections could not decide, and in doing so they created the record this article reads. The electoral reading also cannot weigh the findings themselves: it treats the commission, the investigation team and the courts as scenery in a political story, when for the participants they were the arena. The audit reading inverts the priority. The election of December 2002 is, on the argument here, best understood as the first institutional response to the violence, a verdict by the one body that answered quickly, and the twenty years of audit that followed are the slower institutions catching up with questions the ballot could not formulate.

The Three-Stage Audit-Trajectory Map

Before the stages are examined in turn, the whole architecture can be set out in one view. The map below, called the Three-Stage Audit-Trajectory Map in this article and designed to be cited by that name, arranges the audit by stage, fixing for each stage the institution that acted, the years it occupied, the mandate it carried and the findings or positions it left on the record. A fourth row records the parallel track of criminal prosecutions, which ran beside the three stages and supplied the audit’s only convictions.

Stage	Institution and years	Mandate	Findings or positions left on the record
Stage One	Nanavati and Shah, then Nanavati and Mehta, Commission of Inquiry, from 2002 to 2014, report public in full by December 2019	To inquire into the Godhra fire and the violence that followed, including the conduct of the state administration	Godhra fire a pre-planned conspiracy; post-Godhra violence not organised by the state; no evidence that any minister inspired or abetted the attacks; police ineffective at places
Stage Two	Special Investigation Team under R. K. Raghavan, constituted by the Supreme Court on 26 March 2008	To re-investigate nine major cases under the Court’s monitoring, and to examine the Zakia Jafri complaint of 8 June 2006	Case-wise chargesheets and trials; closure report of 8 February 2012 finding no prosecutable evidence against the sixty four persons named in the complaint
Stage Three	The Zakia Jafri proceedings, from the complaint of 2006 through the Supreme Court judgment of 24 June 2022	To test the closure report through a protest petition at the magistrate, revision and appeal levels	Closure accepted by the magistrate on 26 December 2013, upheld by the Gujarat High Court on 5 October 2017 and by the Supreme Court on 24 June 2022
Parallel track	Special courts at Ahmedabad, Anand, Mehsana and Mumbai, with the Gujarat High Court and the Supreme Court on appeal	To try the individual massacre and train-burning cases on evidence	Convictions in the Godhra, Sardarpura, Ode, Naroda Patiya, Gulberg Society, Best Bakery and Bilkis Bano trials; acquittal of all accused in Naroda Gam on 20 April 2023

Stage One: The Nanavati Commission, from 2002 to the Tabling of December 2019

The commission was born modest and grew by accretion. The Gujarat government constituted it on 6 March 2002 under the Commissions of Inquiry Act 1952, with Justice K. G. Shah, a retired judge of the Gujarat High Court, as its single member and the Godhra fire as its subject. Within months the terms were widened to the violence across the state, and in September 2002 the commission was reconstituted with Justice G. T. Nanavati, a retired judge of the Supreme Court, as chairman and Justice Shah as member. In 2004 the terms were widened again, at the government’s own hand, to the conduct of the chief minister, ministers, legislators, police officers and organisations, a widening the government later cited as proof that it had hidden from nothing and its critics cited as a delayed concession extracted by pressure. The commission took evidence for years, travelled to the affected districts, and accumulated tens of thousands of affidavits; its sittings became a theatre in which the two positions of 2002 were restated under oath, with Sreekumar’s affidavits its most explosive exhibits and the government’s witnesses its most defended. Justice Shah died on 5 June 2008, and Justice Akshay H. Mehta, another retired judge of the Gujarat High Court, was appointed in his place, giving the body the name, Nanavati and Mehta Commission, by which the record knows it.

Its first fruit came on 18 September 2008, when the commission submitted the part of its report dealing with Godhra. The finding was unqualified: the burning of coach S-6 was a pre-planned conspiracy, executed by a local crowd organised in advance, with petrol procured and stored, and Maulvi Hussein Haji Ibrahim Umarji named as a principal conspirator. The report was tabled in the Gujarat assembly on 25 September 2009. Against it stood the Banerjee committee’s interim conclusion of January 2005 that the fire was accidental, a thesis the Gujarat High Court had already deprived of standing in October 2006, when Justice D. N. Patel held that the Railway Ministry’s committee had been constituted in breach of the law governing railway accident inquiries. The Godhra trial court’s judgment of 1 March 2011, convicting 31 men of the conspiracy the commission had described and acquitting 63, including Maulvi Umarji himself, aligned the criminal record with the commission on the question of design and against it on the identity of the designer, a divergence the audit reading records without embarrassment: commissions and courts answer to different standards, and on Godhra their answers converge on the fire’s deliberateness and diverge on its authorship. Manoj Mitta’s critique of the commission’s Godhra chapters, that it credited police witnesses whose testimony the trial court later discarded, is the strongest available challenge to the convergence, and it is noted here because the final report’s authority depends on how that challenge is answered; the commission’s defenders reply that the trial accepted the conspiracy the commission found, and that the acquittal of individuals is the ordinary working of the burden of proof, not a refutation of the design.

The commission’s second and final part, on the violence itself, was submitted to Chief Minister Anandiben Patel on 18 November 2014, in nine volumes, and then disappeared from public view for five years. It was tabled in the assembly only on 11 December 2019, by Minister of State for Home Pradeepsinh Jadeja, after the state government assured the Gujarat High Court, in proceedings brought by Sreekumar, that it would be. The findings, stated here by their content as the stance charter requires, were these: the post-Godhra violence was not the product of a pre-planned conspiracy and was not orchestrated by the state; there was no evidence to show that the attacks were inspired, instigated or abetted by any minister of the state; the police were ineffective at some places, through want of numbers, of equipment and in places of competence and eagerness, and action was recommended against individual officers on that account; and the evidence of the three officers whose testimony had indicted the government, Sreekumar, Rahul Sharma and Sanjiv Bhatt, was examined and found not credible. One episode of the commission’s working belongs to its audit trail in full, because both sides of the dispute cite it. Through the commission’s years, the critics demanded that it summon the chief minister himself, the one witness whose evidence could test the conspiracy allegation at its source; the commission declined, holding across repeated applications that the material before it did not justify the summons, and Modi never testified before it. The contrast with the criminal process is exact and instructive: the investigation team, a body with coercive powers the commission lacked, questioned the same man for more than nine hours in March 2010, under a Supreme Court mandate the commission never possessed. The critics read the commission’s refusal as the audit’s original sin, a state-appointed body protecting the state’s head from its questions; the commission’s defenders read it as the correct discipline of an inquiry that summons on evidence rather than on demand. The audit reading records the refusal as a fact about the first stage’s design: a commission of inquiry in India works by invitation and reputation, its powers of compulsion are thinner than its terms of reference suggest, and the Gujarat commission’s twelve years are best understood as the record of what such a body can establish, the Godhra design, the pattern of the violence, the failures of individual officers, and what it cannot, the interior of a government’s intentions on the days that mattered. That boundary, visible in the commission’s own practice, is the first stage’s most durable finding, and it explains why the republic built a second stage of a different kind.

The Congress and the critics answered that a report delivered to the government it examined, tabled five years after submission and seventeen years after the events, had audited everything except its own delay; the BJP answered that the commission had heard more evidence than any other body and cleared the government on it. Both answers belong to the record. What the audit reading adds is the structural observation the delay itself proves: a commission of inquiry reports to the government that appointed it, sits at that government’s pleasure through extension after extension, and publishes on that government’s timetable, so that even a complete exoneration arrives stamped with the architecture that produced it. The Liberhan Commission’s seventeen years and forty eight extensions had taught the same lesson in the other partisan direction, and the two reports together, on the demolition and on the riots, are the republic’s paired exhibits on what commissions can and cannot settle.

The Parallel Track: Transfers, a Tribunal and the Trial Courts

While the commission sat, the criminal process faltered, was rescued, and was rebuilt, and the story of that rebuilding is the audit’s most concrete chapter. The first trials inside Gujarat collapsed in public. In the Best Bakery case, arising from the killing of 14 people at a Vadodara bakery on 1 March 2002, a fast-track court acquitted all 21 accused on 27 June 2003 after the principal witness, Zaheera Sheikh, retracted her testimony in court. The National Human Rights Commission carried the case to the Supreme Court, which on 12 April 2004 quashed the acquittal in language that described the trial as a farce and ordered a retrial outside Gujarat, in Maharashtra. The Bilkis Bano case travelled the same road: after the attack of 3 March 2002 at Randhikpur, in which Bilkis Bano was gang-raped and fourteen members of her fleeing family group were killed, the Gujarat police closed the case for want of evidence; the Supreme Court, on the Commission’s petition, handed the investigation to the Central Bureau of Investigation in December 2003 and transferred the trial to Mumbai in August 2004. The Supreme Court stayed the trials of the nine major cases on 21 November 2003, pending its supervision of the investigation itself, and the message of those years, delivered by the court’s own orders, was that the ordinary machinery of the state could not be trusted with these files. That message is a finding too, and the audit reading counts it among the most consequential the episode produced, because it came not from the government’s critics but from the republic’s highest court in the exercise of its ordinary jurisdiction.

What did the Concerned Citizens Tribunal conclude in 2002?

The Concerned Citizens Tribunal, chaired by Justice V. R. Krishna Iyer and including Justice P. B. Sawant and Justice Hosbet Suresh, published Crime Against Humanity in November 2002. It concluded that the violence had been planned in advance, that the state administration was complicit in it, and that the death toll reached as high as 1,926.

The tribunal had no powers of summons and no standing in law, and the state government dismissed it as a partisan exercise; its convenor, the activist Teesta Setalvad of Citizens for Justice and Peace, would remain the audit’s most persistent litigant for twenty years. Yet the tribunal’s record, gathered within months while testimony was fresh, fed the later formal proceedings at every stage, and its casualty estimate framed the counting dispute the official figures never fully closed. The episode also produced an evidentiary archive of a second kind: undercover journalism. In 2007 the magazine Tehelka published recordings, made by its reporter Ashish Khetan posing as a sympathiser, in which accused men and a government counsel described the violence and its protection in their own words; the recordings were later exhibited in the Naroda Gam trial. Investigative material of that kind survives only when reporters can hold it against pressure, in the disciplined custody of protected archives of the sort that services such as the VaultBook record on how journalists protect their sources describe, and its admission into the trial record made the sting part of the audit itself.

The re-investigation the Supreme Court ordered in March 2008 set the trials running again, in special courts, with witness protection directed by the investigation team, and between 2011 and 2016 the principal verdicts arrived. The Godhra trial closed first, on 1 March 2011. At Sardarpura, where 33 people were killed, a special court convicted 31 accused on 9 November 2011 and sentenced them to life imprisonment. The Ode trials of 2012 produced 23 convictions in the principal case. On 29 August 2012, Judge Jyotsna Yagnik delivered the Naroda Patiya judgment, the audit’s heaviest single document: 97 killings proved, 32 accused convicted, among them Maya Kodnani, the sitting legislator for Naroda and a former minister, whom the judgment named a principal conspirator and sentenced to 28 years, and Babu Bajrangi of the Bajrang Dal, sentenced to life imprisonment for the remainder of his life. The Gulberg Society trial, held back for years by the pendency of the conspiracy complaint, concluded on 2 June 2016, when Judge P. B. Desai convicted 24 accused and acquitted 36. The appellate courts then did their slower work. The Gujarat High Court commuted the Godhra death sentences in October 2017, and on 20 April 2018 it acquitted Kodnani in Naroda Patiya, holding the testimony placing her at the scene too contradictory to sustain a conviction, while upholding Bajrangi’s conviction with his sentence reduced to 21 years. In the Bilkis Bano case, the Mumbai special court had convicted 11 men to life imprisonment on 21 January 2008; the Bombay High Court upheld the convictions in May 2017; and the Supreme Court on 23 April 2019 ordered the Gujarat government to pay Bilkis Bano compensation of fifty lakh rupees, with a government job and a house, the largest such award in the audit’s record. On 20 April 2023, the parallel track reached its last major verdict: Judge Shubhada Baxi, trying the Naroda Gam case in which 11 people were killed, acquitted all 67 accused who remained on trial, twenty one years after the event, in a case where the Union home minister Amit Shah had appeared in September 2017 as a defence witness for Kodnani’s alibi. Taken together, the trial record convicts the violence’s executors in their hundreds and no one above the rank of a local organiser and one legislator, whose conviction did not survive appeal. That distribution, mass guilt at the base and acquittal at every altitude, is the parallel track’s aggregate finding, and both sides of the original dispute claim it: the government as proof that the law took its course against the guilty, the critics as proof that it stopped where organisation would have had to be named. Citizens for Justice and Peace, the organisation that supported many of the prosecutions with legal aid and witness protection work, counted 157 convictions at the trial stage across the riot cases it tracked, 142 of them to life imprisonment; the accounting is the organisation’s own, and it is cited here as such, as the fullest single tally of the parallel track’s yield that any participant has published.

Stage Two: The Supreme Court-Monitored Special Investigation Team, 2008 to 2012

The second stage began with a constitutional novelty. On 26 March 2008, in the proceedings brought by the National Human Rights Commission, the Supreme Court directed the constitution of a Special Investigation Team to re-investigate nine of the gravest cases of 2002, Godhra, Gulberg Society, Naroda Patiya, Naroda Gam, Sardarpura, the Ode cases and the Mehsana cases among them, and placed R. K. Raghavan, a former director of the Central Bureau of Investigation, at its head. The state government issued its constituting notification on 1 April 2008. A court of the republic had, in effect, reached into a state’s police power and rebuilt the investigating agency for a closed set of files, because the original investigations had been discredited by the acquittals, the transfers and its own earlier orders. The team re-recorded statements, re-examined witnesses the Gujarat police had dropped, and reported to the Court in sealed covers; on 1 May 2009 the Court vacated the stay on the trials and directed that they proceed in special courts with prosecutors it approved and witnesses it protected. The investigation records the team assembled, preserved under the Court’s registry in the monitored files, are the audit’s least cited and most consequential primary source: case by case, incident by incident, they converted the allegations of 2002 into charge sheets that the trial courts then sustained in large part.

The team’s second mandate was narrower and politically larger. On 27 April 2009, the Supreme Court directed it to examine the complaint that Zakia Jafri had filed on 8 June 2006 with the state’s director general of police, naming sixty four persons, the chief minister first among them, and alleging that the violence of 2002 had proceeded from a conspiracy at the summit of the state, executed through bureaucratic inaction, police complicity and orchestrated hatred. The complaint had already travelled: the Gujarat police had declined to register it as a first information report, the Gujarat High Court had declined in November 2007 to order registration, and it reached the investigation team only because the Supreme Court routed it there. The team questioned the chief minister himself: on 27 March 2010, Narendra Modi appeared at the team’s Gandhinagar office and answered its questions for more than nine hours, the only sitting chief minister in the republic’s history to submit to interrogation in a criminal investigation of this kind. It examined the call records, the meeting records and the officers, and it tested the two testimonies on which the conspiracy allegation chiefly rested. Sanjiv Bhatt’s account of the meeting of 27 February 2002 it found unsupported by the other officers it could place in the room. Sreekumar’s affidavits it found contradicted by the record and by his own service file disputes.

The audit’s most instructive internal document belongs to this stage. The Supreme Court had appointed the advocate Raju Ramachandran as amicus curiae, its own adviser, to assess the team’s work independently. Ramachandran’s reports to the Court, later placed in the public domain during the 2022 proceedings, diverged from the team’s conclusions at the decisive point: in his assessment, the material collected could support putting the chief minister and others on trial for specified offences, and the question of their guilt or innocence was for a court to decide on evidence tested in trial, not for the investigators to foreclose. The team, in its final report, maintained its position. On 12 September 2011 the Supreme Court declined to choose between its two instruments in its own courtroom; it directed the team to place its final report, with the amicus’s comments, before the magistrate’s court at Ahmedabad, where Zakia Jafri would have the right to be heard against closure. On 8 February 2012 the team filed that report before the metropolitan magistrate. It was a closure report. On the complaint’s sixty four names, and on the chief minister at their head, the Special Investigation Team reported that it had found no prosecutable evidence.

Mitta’s book subjects this stage to its most damaging scrutiny, and the audit reading is obliged to state his case fairly. He argues that the team was built to fail the complaint: that it treated the chief minister’s interrogation as a formality, credited the denials of officials over the testimony of survivors, declined lines of inquiry the amicus pressed, and applied to the conspiracy charge a standard of proof appropriate to a trial verdict rather than to an investigation, whose office is only to ask whether a triable case exists. The team’s defenders, and the courts that later examined its work, answer on the same record: the team convicted where conviction was possible, re-investigating the massacre cases to verdicts no one had obtained before it; its interrogation of a sitting chief minister was unprecedented; and the amicus’s disagreement concerned the threshold for prosecution, a question on which investigators and advisers may honestly differ. The magistrate, the High Court and the Supreme Court each had the amicus’s view before them when they accepted the closure, so the divergence was not buried but adjudicated. What stage two settled, then, is precise and limited: under the Supreme Court’s own supervision, with an independent adviser recording his dissent, the investigative machinery of the republic reported that the evidence it could gather would not sustain a prosecution of the state’s leadership for the violence of 2002. Whether that report measures the evidence or the gatherers is the question stage three inherited.

Stage Three: The Zakia Jafri Proceedings, 2006 to the Judgment of 24 June 2022

Zakia Jafri’s complaint began, as the audit’s stage three, in an office that would not register it. The widow of Ehsan Jafri, who had been killed with 68 others at Gulberg Society on 28 February 2002, filed her complaint on 8 June 2006, naming the chief minister, ministers, legislators, police officers and officials, sixty four persons in all, and alleging a conspiracy that ran from the Godhra reaction through the massacres to the subversion of the prosecutions that followed. The legal route she then travelled defines the stage: refusal at the police station, a direction declined by the Gujarat High Court in November 2007, examination by the investigation team on the Supreme Court’s order of April 2009, the closure report of February 2012, and then the protest petition, the instrument by which a complainant asks a magistrate to reject an investigator’s closure and take cognizance herself. The metropolitan magistrate B. J. Ganatra heard the petition through 2013 and, on 26 December 2013, accepted the closure report and rejected the protest, holding that the investigation team’s conclusion was open to it on the material and that the complainant had not made out a case for cognizance against the sixty four. The Gujarat High Court, in the judgment of Justice Sonia Gokani on 5 October 2017, upheld the magistrate, recording that the investigation had been extensive and conducted under the Supreme Court’s monitoring, and declining to order the further investigation the petition sought, while granting liberty to pursue certain lesser allegations through other remedies.

The appeal to the Supreme Court, filed in 2018 and argued through the autumn and winter of 2021 with Kapil Sibal for the petitioner and Mukul Rohatgi for the investigation team, was the audit’s last full hearing of the events of 2002. The petitioner’s case was the critics’ position of 2002 stated as law: a larger conspiracy, bureaucratic inaction as its instrument, hate speech as its signal, and a closure report that had refused to look beyond the Gulberg file to the pattern of some three hundred incidents across nineteen districts that the complaint alleged. The respondents’ case was the government’s position stated as law: an investigation supervised by the Court itself, an amicus’s disagreement considered and answered, two courts below satisfied, and a protest petition that asked the judiciary to order a prosecution on material that three forums had found insufficient. On 24 June 2022 a three judge bench of Justices A. M. Khanwilkar, Dinesh Maheshwari and C. T. Ravikumar delivered the audit’s terminal judgment. Stated by its holdings, as this article states every judgment, the Court dismissed the appeal and upheld the closure; held that the magistrate’s acceptance of the closure report and the High Court’s affirmation of it disclosed no error that warranted interference; held that the investigation team’s assessment of the evidence, including its treatment of Sanjiv Bhatt’s testimony, was an assessment open to a reasonable investigator; and recorded, in observations that travelled beyond the holdings, that the proceedings had been pursued to keep the pot boiling for an ulterior design, naming the role of the co-petitioner organisation Citizens for Justice and Peace and its secretary Teesta Setalvad in terms that the Gujarat police acted upon within a day, arresting Setalvad on 25 June 2022 in a case alleging fabrication of evidence. She was released on interim bail by the Supreme Court in September 2022 and on regular bail in July 2023, and those proceedings belong to the audit’s margin rather than its text; what belongs to the text is the judgment’s institutional meaning. With it, the conspiracy question of 2002, the allegation at the summit of the critics’ case, had been carried through every forum the legal system provides and closed in each. The internal discipline of the long contest can be compared with the religious-personal-law-state-intervention renegotiation of the previous decade, where a judgment, an agitation and a statute closed a question politically while leaving it open morally; the Jafri litigation closed its question legally, and the contest over what the closure means migrated, permanently, to politics and scholarship.

What the Audit Left Outside Its Frame

An audit is defined by its exclusions as much as its findings, and the Gujarat audit’s frame, drawn by the Supreme Court’s orders and the commission’s terms, enclosed nine grave cases, one complaint, one train fire and the conduct of one administration. Outside the frame lay the mass of the violence’s legal residue: the more than four thousand criminal cases the Gujarat police registered across the state in 2002, the great majority of them investigated by the same district machinery whose conduct was in question, and closed in large numbers within the first two years as undetected or unfit for prosecution. It was the National Human Rights Commission’s litigation that forced even this residue into the light: in the course of the proceedings, the state government informed the Supreme Court that the closed cases had been reviewed and that more than two thousand of them had been reopened, a figure that stands in the record as the audit’s quietest and largest finding, the state’s own admission that the first investigation of its worst violence had failed at scale. No commission examined the pattern case by case, no investigation team re-investigated beyond its nine files, and the reopened cases worked their way through local courts largely unobserved, ending in the main in acquittals and compromises that no archive gathers. The audit reading, to be honest about its object, must register that its twenty two years of institutional attention embraced perhaps a tenth of the episode’s criminal files, and that the remainder were closed by the ordinary system acting, after one review, in its ordinary way.

Relief and repair lay outside the frame as well. The camps of 2002 emptied without a statutory rehabilitation programme; compensation moved by executive announcement, from the state government’s first differential scales, revised under protest, to the Union government’s supplementary ex gratia package of 2005, to the Supreme Court’s individualised award to Bilkis Bano in April 2019, fifty lakh rupees with employment and housing, which reads in the record less like a policy than like a court’s rebuke to the absence of one. The displaced who did not return, and the colonies on Ahmedabad’s margins where many of them settled, entered the audit only as background. Nor did the frame include the stewardship of the audit itself by the Union governments that inherited it. From May 2004 to May 2014 the Union was governed by the Congress-led United Progressive Alliance, the political formation most identified with the critics’ position, and its conduct of the audit was one of studied non-interference: it neither dissolved the state commission nor replaced the investigation’s architecture, its one intervention in fact-finding, the Railway Ministry’s Banerjee committee of 2004, being the episode’s single partisan audit instrument and the only one a court struck down. The finding the audit reading draws from that decade is bipartisan and uncomfortable for both formations: the audit’s trajectory was set by courts and commissions, and successive governments of opposed parties alike found it convenient to let the institutions carry a weight that politics preferred not to lift.

The frame also excluded, by design, the question the narrative reading considers the only one: the experience of the violence itself, in its organised cruelty and its neighbourhood intimacies, which no commission’s affidavit schedule and no charge sheet could hold. Ward Berenschot’s ethnography of Ahmedabad, Riot Politics (2011), working from years inside the city’s neighbourhoods, describes a machinery of mediation, politicians, brokers, police and strongmen linked in durable networks, through which both violence and its restraint are delivered, and his account explains a datum the formal audit repeatedly encounters and cannot use: that in 2002 the violence stopped in some places where the networks chose, and ran where they chose otherwise. The audit reading does not compete with that knowledge. It notes, instead, that the republic’s institutions audited the episode in the only vocabulary they possess, conspiracy, evidence, culpability, closure, and that the gap between that vocabulary and the event is the space in which the afterlife of 2002 has been lived. The findings inside the frame are the subject of the adjudication that follows; the exclusions are recorded here so that the adjudication is not mistaken for a verdict on the whole.

The Officers and the Witnesses: How the Audit Tested Its Evidence

The audit’s outcome turned on a small number of testimonies, and the treatment of those testimonies is where the adjudication must finally be done, because it is here that the critics’ case is at its most concrete and the institutions’ reasoning at its most exposed. The prosecution of the conspiracy theory rested on three pillars of testimony. The first was documentary: the telephone records of officials and politicians on 27 and 28 February 2002, collected by the police officer Rahul Sharma, who had served in Bhavnagar during the violence and who preserved call data records onto a compact disc that later entered the proceedings. The call records showed contacts among officials, ministers and accused persons on the critical days; the state government’s answer, accepted by the investigation team, was that the records showed a government in communication during a crisis and proved contact, not content. The second pillar was confessional in form: the Tehelka recordings of 2007, in which Babu Bajrangi described the Naroda Patiya killing to an undercover reporter, and Arvind Pandya, a counsel for the state government before the commission of inquiry, described the management of the riot cases in terms the government disowned; the recordings were exhibited in the Naroda Gam trial, their authenticity contested by the defence, and the trial court that acquitted every accused in that case in April 2023 had them before it. The third pillar was the testimony of insiders, and it is the pillar on which the institutions spent their greatest care.

R. B. Sreekumar’s affidavits, filed before the Nanavati Commission in 2002 and 2004 and expanded thereafter, alleged a meeting of senior officials on 27 February 2002 at which the state’s posture of restraint was set, and a subsequent instruction environment in which Hindu anger was to be allowed expression. The commission heard him at length, and the state government cross-examined with his service record: his supersession for promotion, the departmental proceedings against him, and the diaries he produced, whose entries the government characterised as reconstructions. The commission’s final report found his evidence unreliable, and the investigation team, examining the same claims five years later, reported that the officers named as present at the meeting denied the instruction attributed to it. Sanjiv Bhatt’s affidavit of April 2011 placed him in the same meeting in person and attributed to the chief minister the sentence that Hindus should be allowed to vent their anger; the investigation team tested the attendance itself, examining the drivers, the security logs and the officers listed, and reported that Bhatt’s presence could not be established, while the magistrate and the courts that followed treated the team’s assessment as one open to it. Bhatt’s supporters, among them the lawyer and activist voices who wrote the episode’s counter-record, answer that an officer’s presence at an informal residence meeting would leave exactly the thin trace the team demanded, and that the demand was designed to fail; the team’s defenders answer that a capital charge against a chief minister cannot rest on a presence the record cannot show. Both answers are recorded in the audit’s documents, and the reader can weigh them there, which is what distinguishes a contested record from a settled fiction.

The treatment of witnesses below the rank of officer shows the same double character at greater human cost. The Supreme Court’s orders of 2008 and 2009, creating protected prosecution conditions in the nine trials, were themselves a finding that unprotected testimony in Gujarat had failed, a finding earned in the Best Bakery collapse of 2003 and in the mass hostility of witnesses across the first trials. Within the protected trials, witnesses held: the Naroda Patiya judgment’s 327 witnesses and the Gulberg record rest on survivor testimony that the courts credited in large part, and the convictions in those cases are the audit’s standing refutation of the claim that its machinery could not convict. Outside the protected nine, in the reopened thousands, the record is one of attrition, compromises recorded as settlements between neighbours who must continue to live as neighbours, and acquittals for want of anyone willing to speak. The audit’s evidence problem was therefore never a scarcity of testimony about the violence; it was a scarcity of testimony that could carry the conspiracy’s last step, from the organisers in the street to the men in the secretariat, and every institution that faced that step, the commission, the team, the magistrate, the High Court and the Supreme Court, reported the step unproven on the material it could test. The critics’ answer, that the step is unproven because the men who would prove it controlled the material, is a claim about the world that the record can neither confirm nor finally exclude, and the honesty of the audit reading consists in leaving that claim standing, named and unanswered, beside the findings it could not move.

Adjudicating the Three Readings

What does Christophe Jaffrelot argue about the consequences of 2002?

In Modi’s India (2021), Christophe Jaffrelot argues that the violence of 2002 and its aftermath consolidated a Hindu electoral majority in Gujarat, weakened the institutions meant to check executive power, and supplied the template for a national order he calls ethnic democracy: formal democratic procedures coexisting with the subordination of the Muslim minority.

The disagreement named at the outset can be decided, because the three readings make different predictions about the record and the record discriminates among them. The electoral reading predicts that the episode’s political meaning was exhausted by its verdicts at the polls; the record contradicts it, for the audit deepened for a decade after the last Gujarat verdict it could have influenced, and its terminal judgment of 2022 was sought and fought by actors who plainly believed it mattered beyond any election. The narrative reading predicts that the institutional sequel would be mere epilogue, a betrayal or vindication of the spring of 2002; the record contradicts that too, for the sequel produced things the spring did not contain: convictions of hundreds that the first investigations had not attempted, a doctrine of court-monitored investigation that later cases inherited, an official exoneration of the state’s leadership that the first government statements had only asserted, and a citizens’ archive that outlasted every official one. The audit reading predicts exactly this: that the institutions would generate their own consequential record, irreducible to the event and to the elections, and that later politics would have to operate through it. On the evidence assembled in the preceding sections, the audit reading is sustained.

Sustaining it requires answering its strongest internal critic, and Mitta’s challenge deserves a verdict of its own. His demonstration that the investigative process was bent, in its personnel choices, its handling of witnesses and its thresholds of proof, toward the outcome it reached is, in its particulars, largely unrebutted on the record he assembles, and this article does not pretend otherwise. But his conclusion, that the findings are therefore fictions, overreaches his evidence in a way the audit reading can locate precisely. A fiction is a finding nothing supports; the closure of the Jafri complaint was supported, at each of four successive levels, by reasons given in writing, tested against an amicus’s contrary reasons, and open to the appellate correction that the petitioner in fact exercised for a decade. To call the outcome authored is accurate; every institutional outcome is authored, by rules, appointments and standards of proof. To call it fictitious is to deny the distinction between a record and its criticism, and the distinction matters practically: it is because the findings are real institutional facts, and not fictions, that they bind, validate and constrain, and that overruling them would require the very institutional labour their critics have not persuaded any forum to undertake. The complication the brief for this article posed, that the audit’s findings are themselves contested at the level of substance, is therefore answered by relocation rather than denial. The contestation is real, and it operates where such contestation belongs in a constitutional order that has closed a question in its courts: in scholarship, in journalism and in politics, outside the audit’s frame. Inside the frame, the findings stand, all of them: the Godhra conspiracy and the acquittal of its alleged author, the massacres and the hundreds convicted of them, the comprehensive failure the Human Rights Commission recorded in May 2002 and the non-culpability the investigative machinery recorded a decade later, the failure and the exoneration being compatible because they answer different questions under different rules, the first about a state’s duty in a season, the second about named men’s provable guilt in a court.

The scholarly record divides along the same seam, and its strongest voices on each side should be heard before the verdict. Martha Nussbaum, in The Clash Within (2007), argued from the documentary record that the violence bore the marks of planning and state complicity and that the episode revealed the vulnerability of India’s constitutional commitments to majoritarian capture; her account represents the critics’ position at its most philosophically ambitious, and it relies, necessarily, on the audit’s early documents, the Human Rights Commission report and the tribunal, since it predates the investigation team and the judgment of 2022. Paul Brass, whose work on the production of Hindu and Muslim violence had earlier given the study of Indian riots its concept of the institutionalised riot system, treated episodes like 2002 as products of durable networks of specialists rather than as spontaneous combustions, and the critics cite his framework as the theory their evidence fits; the state government’s defenders answer that the framework, applied to Gujarat, predicts the very organisers the trial courts did convict, at Naroda and Sardarpura, and does not by itself carry the charge the last step upward, to the secretariat. The audit reading accepts the framework as an account of how the violence was possible and notes that the institutions, applying their own evidentiary rules, convicted the network’s executors and acquitted its alleged directors, so that Brass’s theory and the closure report describe the same events at different altitudes, one explaining the machine, the other declining to convict its alleged operators.

Between Jaffrelot and the audit reading there is less distance than the framing of rivals suggests, and the adjudication should say so. Jaffrelot’s own method in the post-2002 chapters is substantially an audit method: he reads the commission, the investigation and the courts as institutions behaving in character, and his charge that the checks failed is a claim about the same documents. Where the audit reading parts from him is on the inference from failure. His account treats the audit’s exonerations as evidence for the ethnic democracy thesis, one more institution captured; the reading defended here treats them as the thesis’s boundary. The post-2014 order has indeed invoked the findings as validation, as the next section shows, but it has also been obliged to live inside them: to surrender the Bilkis convicts back to prison in January 2024 when a court held their release unlawful, to accept an acquittal map drawn by trial judges it did not appoint for that purpose, and to rest its leader’s vindication on documents whose authority depends on the independence of the very institutions its critics accuse it of bending. An order that must cite its audits has not abolished them. That, in the end, is the audit reading’s claim about the whole twenty two years: the findings of 2002’s long audit became part of the constitution of the politics that followed, not as decoration and not as destiny, but as the fixed points around which argument, including the argument of this article, is obliged to move.

The Post-2014 Architecture Operating Through the Findings

The premiership that began on 26 May 2014 was the first in the republic’s history to be carried into office by a man whose fitness for it had been litigated, in the court of institutional audit, for a decade. The audit’s findings entered the new order’s foundation documents, not as statute but as citation. Through the 2014 campaign and after it, Modi and his party stated the audit’s conclusion as their warrant: the investigation team appointed by the Supreme Court had examined the conspiracy charge and found no prosecutable evidence; the courts had accepted that closure; the accusers had had every forum and lost in each. The statement was made in interviews, in campaign speeches and in Parliament, and its form matters more than its occasions: the leader’s vindication was not asserted as a political claim, to be believed or disbelieved with one’s party, but cited as an institutional fact, with the closure report of 8 February 2012 as its exhibit. This is the operation the thesis names. Inside the integrated three-tensions framework that reads the post-2014 order as a renegotiation of the republic’s founding settlements, the audit findings function as the renegotiation’s title deeds on the religious question: the contested settlement of secularism, whose terms the contested-settlement framework sets out, could be revised after 2014 in part because the hardest case against the reviser had been formally heard and formally closed. The disentangling attempted in the tripartite-disentanglement framework applies here as well: the audit’s subject was never an ideology but a set of acts and omissions, and its exoneration was correspondingly narrow, a finding about evidence and named men, which the political order then spent as though it were a general absolution. The gap between the finding and the spending is the audit reading’s final exhibit, and it is a gap the findings themselves police only when someone litigates, which is why the order’s architects prefer the citation to the reopening.

The architecture’s day-to-day operation through the findings is visible in the post-2014 PMO architecture context: a centralised executive, jealous of institutional form, has treated the audit’s documents as the settled perimeter of the 2002 question and has resisted every attempt to redraw it. The sequence had in fact begun before the premiership, in the campaign that created it, and the timing is part of the record. The magistrate’s order accepting the closure report was delivered on 26 December 2013, ten weeks after the BJP had named Modi its prime ministerial candidate and five months before the general election of April and May 2014. Through that campaign the closure functioned exactly as this section describes: the candidate’s speeches and his party’s documents presented the investigation team’s finding, and the magistrate’s acceptance of it, as the judiciary’s answer to the charge that had defined his national reputation since 2002, and the Congress, which had governed the Union through the decade of the audit without disturbing its architecture, found its strongest rhetorical charge answered by documents its own government’s courts had produced. The election of 2014 therefore cannot be read as the electorate ignoring the audit; the audit was on the ballot, in the form of its findings, and the electorate’s verdict of 282 seats for the BJP was delivered with the closure report five months old and unappealed at the higher levels. What the Supreme Court’s judgment of June 2022 added, eight years later, was not a new fact but a finality: the closure the campaign had cited as a magistrate’s order became, in the end, the considered position of the republic’s highest court, and the citation the premiership had spent since 2014 appreciated into settled institutional capital.

The tabling of the Nanavati and Mehta Commission’s final part on 11 December 2019 belongs to this pattern. A government in its second term, under no electoral pressure on the question and acting only on its assurance to the Gujarat High Court, placed the seventeen year old exoneration on the assembly’s table and let the opposition’s protests exhaust themselves against a document whose authority the mover did not need to argue; the report argued itself. The judgment of 24 June 2022 then closed the audit’s third stage in the same register, and its reception completed the circuit: the prime minister’s supporters cited the Supreme Court’s dismissal as the final institutional word, the petitioners’ supporters cited its observations on the litigation’s conduct as a warning against future petitions, and both citations presuppose what the audit reading asserts, that the documents are the terrain. Even the audit’s post-2022 positions, the events the brief for this article mandated be followed past the judgment, confirm the pattern from both directions. The Naroda Gam acquittals of 20 April 2023 were received by the government side as the trial record’s completion and by the victims’ counsel as its final failure, and the two receptions share the premise that the verdict settles what it settles. The Bilkis Bano remission ran the circuit in reverse and is the more instructive case: a state government released the eleven convicts on 15 August 2022 under a remission policy of 1992, treating the file as closed business; the Supreme Court on 8 January 2024 held that the release had been made without jurisdiction and by a process it described in the language of usurpation, and the convicts surrendered to prison within the fortnight the Court allowed. The architecture operated through the audit’s findings when they served it, and was operated upon by them when they did not, and a political order that can be reversed by its own cited documents is, whatever else is said of it, still living inside an audit culture. That is the strongest available evidence for the claim with which this article began, and it is evidence the post-2014 order itself supplied.

The audit is therefore closed in a sense its participants rarely grant one another. No forum stands seized of the events of 2002; no commission sits; no investigation runs. What remains in force is the paper: a commission’s volumes on a Gujarat assembly shelf, a closure report in a magistrate’s record room, a judgment of June 2022 in the law reports, and the trial files of nine cases in which hundreds of convictions and acquittals distribute guilt and innocence name by name. Politics will continue to cite that paper, in both directions, for as long as the generation of 2002 holds office and memory, and each citation will confirm the argument of this article, that the audit-trajectory, and neither the election nor the wound alone, is where the republic’s answer to Gujarat 2002 is kept.

Frequently Asked Questions

Q: What happened in Gujarat in 2002?

On 27 February 2002, coach S-6 of the Sabarmati Express was burned at Godhra station and 59 people, predominantly Hindu pilgrims returning from Ayodhya, were killed. In the violence that followed across some twenty one districts, mobs killed 69 people at Gulberg Society and 97 at Naroda Patiya in Ahmedabad on 28 February, with further massacres at Naroda Gam, Sardarpura and Ode. The Government of India later reported 1,044 deaths in the post-Godhra violence, 790 Muslims and 254 Hindus, while the Concerned Citizens Tribunal estimated 1,926. The army was deployed, curfews imposed, and more than one hundred thousand people sheltered in relief camps. The events generated a twenty two year sequence of commissions, investigations and trials that closed only with the Supreme Court’s judgment of 24 June 2022 and the Bilkis Bano remission judgment of 8 January 2024.

Q: What was the Godhra train burning?

The Godhra train burning was the attack on the Sabarmati Express at Godhra station on the morning of 27 February 2002, in which coach S-6 was set on fire and 59 passengers died. The train carried Hindu pilgrims and kar sevaks returning from Ayodhya. How the fire began became the foundational dispute of the episode: the Nanavati Commission reported in September 2008 that the burning was a pre-planned conspiracy, a committee under Justice U. C. Banerjee reported in January 2005 that the fire was probably accidental, and the Gujarat High Court struck the committee’s appointment down as unlawful in October 2006. A special court convicted 31 accused of conspiracy and murder on 1 March 2011 and acquitted 63; the High Court commuted the eleven death sentences to life imprisonment in October 2017, leaving the conspiracy finding in place.

Q: What did the Nanavati-Mehta Commission examine on Gujarat 2002?

The commission, constituted by the Gujarat government on 6 March 2002 under the Commissions of Inquiry Act 1952, began with the Godhra fire alone and was widened first to the violence across the state and in 2004 to the conduct of the chief minister, ministers, legislators, police officers and organisations. Chaired first by Justice K. G. Shah alone, then by Justice G. T. Nanavati with Justice Shah, and after Justice Shah’s death in June 2008 by Justice Nanavati with Justice Akshay H. Mehta, it took evidence for years across the affected districts. Its Godhra part, submitted in September 2008, found a pre-planned conspiracy. Its final part, submitted on 18 November 2014 and tabled in the assembly on 11 December 2019, held that the post-Godhra violence was not organised by the state, that no evidence showed any minister inspiring or abetting the attacks, and that the police had been ineffective at places.

Q: What is the SIT investigation?

The SIT, or Special Investigation Team, was the body the Supreme Court directed the Gujarat government to constitute on 26 March 2008, under the chairmanship of R. K. Raghavan, a former director of the Central Bureau of Investigation, to re-investigate nine of the gravest cases arising from the violence of 2002, among them Godhra, Gulberg Society, Naroda Patiya, Naroda Gam, Sardarpura and the Ode and Mehsana cases. Working under the Court’s monitoring, it re-recorded evidence, protected witnesses and filed the charge sheets on which the special courts later convicted in most of the nine cases. On the Court’s further direction of 27 April 2009 it also examined Zakia Jafri’s complaint of a larger conspiracy, questioned Chief Minister Narendra Modi on 27 March 2010, and filed a closure report on 8 February 2012 finding no prosecutable evidence against the sixty four persons the complaint named.

Q: What is the Zakia Jafri case?

The Zakia Jafri case is the legal proceeding founded on the complaint filed on 8 June 2006 by Zakia Jafri, widow of the former Congress member of Parliament Ehsan Jafri, who was killed at Gulberg Society on 28 February 2002. The complaint, addressed to the state’s director general of police, named sixty four persons, headed by Chief Minister Narendra Modi, and alleged a larger conspiracy behind the violence, executed through bureaucratic inaction, police complicity and orchestrated hatred. The police declined to register it, the Gujarat High Court declined to order registration in 2007, and the Supreme Court routed it to the Special Investigation Team in 2009. When the team filed its closure report in February 2012, Zakia Jafri opposed it through a protest petition, which a magistrate rejected on 26 December 2013 and the Gujarat High Court declined to disturb on 5 October 2017, before the Supreme Court’s final hearing.

Q: What did the Supreme Court rule in 2022?

On 24 June 2022, a three judge bench of Justices A. M. Khanwilkar, Dinesh Maheshwari and C. T. Ravikumar dismissed Zakia Jafri’s appeal in Zakia Jafri v State of Gujarat and upheld the closure of her complaint. Stated by its holdings, the Court found no error in the magistrate’s acceptance of the Special Investigation Team’s closure report of 8 February 2012, affirmed the Gujarat High Court’s judgment of 5 October 2017, and held that the team’s assessment of the evidence, including the testimony of Sanjiv Bhatt, was open to a reasonable investigator on the material gathered. The Court also recorded observations on the conduct of the long proceedings, describing them as pursued to keep the pot boiling for an ulterior design, and criticised the role of the co-petitioner organisation Citizens for Justice and Peace.

Q: How does Gujarat 2002 affect the Modi premiership?

The premiership that began on 26 May 2014 was shaped by the audit of 2002 in a direct institutional sense. Narendra Modi entered national office as the only Indian leader whose conduct in a communal catastrophe had been examined by a commission of inquiry, re-investigated by a Supreme Court-monitored team that questioned him in person in March 2010, and closed by a closure report that a magistrate, a High Court and finally the Supreme Court in June 2022 successively upheld. His party has cited those findings as institutional validation whenever the events are raised, and the citation has structured the opposition’s choices as well, pushing criticism toward the audit’s conduct rather than toward a reopening no forum will order. The events also formed the premiership’s governing memory: the centralised, documentation-conscious administrative style associated with the post-2014 Prime Minister’s Office was built by a leader whose survival once turned on what the record could be made to show.

Q: What did the U.C. Banerjee Committee find about the Godhra fire?

The committee chaired by Justice U. C. Banerjee, a retired judge of the Supreme Court, was appointed by the Railway Ministry in September 2004 to inquire into the circumstances of the Godhra fire. Its interim report, submitted on 17 January 2005, concluded that the fire in coach S-6 had in all probability begun inside the coach and was accidental, rejecting the conspiracy account on which the Gujarat police case rested. The conclusion collided directly with the state investigation and with the Nanavati Commission’s later finding of a pre-planned conspiracy. The Gujarat High Court then removed the committee from the legal field altogether: in October 2006, Justice D. N. Patel held that the committee had been constituted in breach of the statutory scheme for railway accident inquiries and that its report carried no legal authority. The criminal courts, which convicted 31 accused of conspiracy in 2011, proceeded on the opposite conclusion from the committee’s.

Q: What did the National Human Rights Commission conclude in its report of 31 May 2002?

The National Human Rights Commission, chaired by Justice J. S. Verma, sent a team to Gujarat in March 2002 and issued preliminary observations on 1 April 2002 before its report of 31 May 2002. The report’s central sentence held that there was a comprehensive failure of the state to protect the constitutional rights of the people of Gujarat. It recorded failures of intelligence, of response and of relief, criticised the handling of the first investigations, and recommended that the gravest cases, including Gulberg Society, Naroda Patiya and the Best Bakery case, be entrusted to the Central Bureau of Investigation. The Gujarat government contested the report’s premises and resented its procedure, but the document set the terms the later audit had to answer, and the Commission’s subsequent petition to the Supreme Court in 2003, seeking transfer and supervision of the major trials, produced the stays, transfers and eventually the Special Investigation Team itself.

Q: What happened in the Best Bakery case?

The Best Bakery case arose from the attack of 1 March 2002 on a bakery in Vadodara owned by the Sheikh family, in which 14 people were killed. A fast-track court acquitted all 21 accused on 27 June 2003 after the principal witness, Zaheera Sheikh, retracted her identification of the attackers in the witness box. The National Human Rights Commission petitioned the Supreme Court, which on 12 April 2004 quashed the acquittal as a farce and ordered a retrial in Maharashtra, outside the state whose machinery had failed. Zaheera Sheikh was later sentenced to one year’s imprisonment for perjury by the Supreme Court in 2006. At the retrial, a Mumbai court convicted nine accused on 24 February 2006 and sentenced them to life imprisonment; the Bombay High Court in 2012 upheld the life sentences of four and acquitted five. The case became the template for the transfers that rebuilt the prosecution of 2002.

Q: What was the outcome of the Bilkis Bano case?

Bilkis Bano, attacked with her family at Randhikpur on 3 March 2002 and gang-raped while fourteen of her fleeing family group were killed, obtained the audit’s most travelled verdict. After the Gujarat police closed her case, the Supreme Court ordered a Central Bureau of Investigation inquiry in December 2003 and transferred the trial to Mumbai, where a special court convicted 11 men and sentenced them to life imprisonment on 21 January 2008; the Bombay High Court upheld the convictions in May 2017, and the Supreme Court ordered the Gujarat government in April 2019 to pay her fifty lakh rupees in compensation with a job and a house. The case reopened in remission: released by the Gujarat government on 15 August 2022 under its 1992 policy, the eleven convicts were ordered back to prison by the Supreme Court on 8 January 2024, which held the remission unlawful and surrendered themselves that month.

Q: What did the trial court decide in the Godhra train burning case?

A special court at Ahmedabad, presided over by Judge P. R. Patel, delivered the Godhra trial judgment on 1 March 2011, nine years after the fire. Of 94 accused tried, the court convicted 31 of criminal conspiracy, murder and connected offences and acquitted 63, including Maulvi Hussein Haji Ibrahim Umarji, whom the Nanavati Commission had named a principal conspirator. Eleven of the convicted were sentenced to death and twenty to life imprisonment. The court accepted the prosecution’s central account: that the attack on coach S-6 was planned in advance, that petrol was procured for the purpose, and that a crowd was assembled to execute it. The Gujarat High Court, hearing the appeals, commuted all eleven death sentences to life imprisonment on 9 October 2017 and broadly sustained the convictions, so the judicial record closes on a conspiracy finding for the fire that began the violence of 2002.

Q: What was the verdict in the Naroda Patiya case?

The Naroda Patiya case, arising from the killing of 97 people in Ahmedabad on 28 February 2002, produced the heaviest verdict of the trials. On 29 August 2012, Judge Jyotsna Yagnik of the special court convicted 32 accused and acquitted 29, finding a planned conspiracy behind the massacre. Maya Kodnani, the Bharatiya Janata Party legislator for Naroda and a former minister in the Gujarat government, was convicted as a principal conspirator and sentenced to 28 years’ imprisonment; Babu Bajrangi of the Bajrang Dal was sentenced to imprisonment for the remainder of his natural life. Sentences were pronounced on 31 August 2012. The convictions did not all survive appeal. The Gujarat High Court acquitted Kodnani on 20 April 2018, holding the witness testimony placing her at the scene too contradictory to sustain the conviction, while upholding Bajrangi’s conviction with his sentence reduced to 21 years.

Q: What was the verdict in the Gulberg Society trial?

The Gulberg Society trial concerned the killing of 69 people, among them the former member of Parliament Ehsan Jafri, at a Muslim housing colony in Ahmedabad on 28 February 2002. The trial was among the nine re-investigated by the Special Investigation Team, but its conclusion was delayed for years, in part because the related conspiracy complaint of Zakia Jafri was pending and the Supreme Court directed that no final judgment be pronounced while that examination ran. Judge P. B. Desai delivered the verdict on 2 June 2016, convicting 24 of the accused and acquitting 36, and on 17 June 2016 sentenced eleven of the convicted to life imprisonment for murder, with lesser terms for the others. The judgment recorded an unlawful assembly’s planned attack on the society and the failure of the police response on the day, findings that stand in the trial record beside the closure of the conspiracy complaint by the same process.

Q: What happened in the Naroda Gam case in April 2023?

The Naroda Gam case, the last of the nine major cases re-investigated by the Special Investigation Team to reach a verdict, ended on 20 April 2023 in the special court of Judge Shubhada Baxi at Ahmedabad. The case concerned the killing of 11 people in the Naroda Gam area of Ahmedabad on 28 February 2002. Eighty six persons had been charged; eighteen died during the trial’s thirteen years, one was discharged, and the 67 who stood trial, among them the former minister Maya Kodnani, the Bajrang Dal leader Babu Bajrangi and the Vishwa Hindu Parishad leader Jaydeep Patel, were all acquitted, the court finding the prosecution’s evidence insufficient to sustain the charges. Union Home Minister Amit Shah had appeared in September 2017 as a defence witness supporting Kodnani’s account of her movements that day. Counsel for the victims’ families announced that the acquittal would be challenged in the Gujarat High Court.

Q: What role did the amicus curiae play in the SIT proceedings on the Zakia Jafri complaint?

An amicus curiae, a friend of the court, advises the court independently of the parties. In the monitoring of the Special Investigation Team, the Supreme Court first worked with Harish Salve as amicus in the National Human Rights Commission proceedings, and for the examination of Zakia Jafri’s complaint it appointed the advocate Raju Ramachandran. Ramachandran reviewed the team’s material and reported to the Court an assessment that diverged from the team’s at the decisive point: in his view, the material could support placing the chief minister and others on trial for specified offences, leaving guilt to be determined at trial. The team maintained its closure position. The Supreme Court’s order of 12 September 2011 sent both documents to the magistrate’s court, where the protest petition was heard with the amicus’s comments on the record, and the magistrate, the High Court and ultimately the Supreme Court in 2022 each accepted closure with the divergence before them.

Q: Why did the Gujarat High Court acquit Maya Kodnani in 2018?

Maya Kodnani, convicted by the special court in August 2012 as a principal conspirator in the Naroda Patiya massacre and sentenced to 28 years, appealed to the Gujarat High Court. In its judgment of 20 April 2018, the High Court acquitted her, holding that the prosecution’s case on her presence and role at Naroda Patiya rested on the testimony of eleven witnesses whose statements the court found contradictory and mutually inconsistent, none of whom had named her in the earliest statements recorded after the event. On that evidentiary assessment, the court held the conspiracy charge against her not proved and extended her the benefit of the doubt. The same judgment upheld the conviction of Babu Bajrangi, reducing his sentence to 21 years, and sustained most of the other convictions in part. Kodnani, who had been on bail on medical grounds since 2014, was later acquitted in the separate Naroda Gam case in April 2023 as well.

Q: What did Atal Bihari Vajpayee tell Narendra Modi about rajdharma in April 2002?

On 4 April 2002, Prime Minister Atal Bihari Vajpayee visited Ahmedabad, toured the relief camps including the Shah Alam camp, and addressed the press in the chief minister’s presence. He said that a ruler must follow rajdharma, the dharma of governance, and that rajdharma knows no distinction between subjects by birth, caste or creed; those who govern, he said, must rise above such distinctions. The sentence was universally reported and immediately contested in meaning. The Gujarat government’s critics read it as a public rebuke of Modi’s handling of the violence, delivered in the mildest available register by a prime minister whose coalition would not dismiss him. The Bharatiya Janata Party’s defenders read it as a general exhortation addressed to all governments. Vajpayee never resolved the ambiguity. Eight days later, at the party’s national executive in Goa on 12 April 2002, Modi offered his resignation and the party declined it.

Q: What was the result of the December 2002 Gujarat assembly election?

The Gujarat assembly, dissolved early on 19 July 2002, went to the polls on 12 December 2002 after the Election Commission’s refusal of an earlier timetable and a Presidential reference to the Supreme Court on the Constitution’s six month rule. The results, declared on 15 December 2002, gave the Bharatiya Janata Party 127 of the assembly’s 182 seats on about half the popular vote, and the Congress 51 seats, on a turnout of about 62 percent. Narendra Modi was sworn in as chief minister again on 22 December 2002. The verdict was read by the Bharatiya Janata Party as the electorate’s answer to the charges of that year, and by its opponents as a majority returned by a polarised electorate on a security campaign. Modi remained chief minister until May 2014, winning the Gujarat elections of 2007 and 2012 as well, before becoming prime minister.

Q: What did Manoj Mitta argue in Modi and Godhra: The Fiction of Fact-Finding?

Manoj Mitta’s Modi and Godhra: The Fiction of Fact-Finding, published in 2014, is the most sustained critique of the audit from within its own documents. Mitta, a legal journalist, examined the Special Investigation Team’s handling of the Zakia Jafri complaint and argued that the fact-finding was arranged so as to exonerate: that the team applied a trial court’s standard of proof at the investigative stage, interrogated the chief minister without pressing the contradictions in the record, credited official denials over survivor testimony, and declined lines of inquiry that its own amicus curiae, Raju Ramachandran, had pressed. His title states the thesis: the closure report was a fiction of fact-finding, a document in the form of an investigation whose conclusion preceded its work. The book’s critics answer that three courts tested the closure with Mitta’s materials available and upheld it, and that a disagreement about prosecutorial thresholds is not proof of bad faith.