In the night of 2 and 3 December 1984, methyl isocyanate gas escaped from a pesticide plant operated in Bhopal by Union Carbide India Limited, the Indian subsidiary of the Union Carbide Corporation of the United States, which held 50.9 per cent of the subsidiary’s shares. The Bhopal disaster killed thousands of people in the surrounding settlements within days. The Government of Madhya Pradesh confirmed 3,787 deaths as related to the gas release, and every other count in circulation, from the figures in the Supreme Court’s settlement reasoning to the estimates of victims’ organisations, is a rival number that this article keeps separate, because the quarrel over the count became part of the legal history itself. What followed the leak, over nearly four decades, was not one legal response but a sequence of them, and the sequence is the subject here: a statute that took every victim’s claim away from the victim and gave it to the state, an umbrella environmental statute enacted within eighteen months, a new chapter on hazardous processes in factory law, a doctrine of absolute liability announced in a different city’s case, a no-fault insurance statute, a compensation tribunal that Parliament created on paper and the executive never brought into existence, and a compensation question that the Supreme Court finally closed in March 2023.

The test this history has to pass is a tracing test. A reader who finishes it should be able to say which instrument answered which failure, in which order, and which of the answers never operated at all. Four statutes and one doctrine came out of Bhopal. One of the statutes was repealed fifteen years after its enactment without having been in force for a single day. The criminal case ended, twenty-six years after the leak, in a two-year sentence under a negligence provision. The compensation case ended, thirty-nine years after the leak, with a Constitution Bench declining to reopen a settlement that had been struck in 1989 on assumptions about the dead and the injured that the claims process itself later disproved. Each element of that sequence has a statutory or judicial anchor, and each anchor is set out in turn.

The legal landscape the Bhopal disaster exposed

The first thing the disaster exposed was how little law there was to expose. India in December 1984 had no general environmental statute. The Water (Prevention and Control of Pollution) Act, 1974 and the Air (Prevention and Control of Pollution) Act, 1981 regulated discharges and emissions through boards and consent orders, but neither created a compensation remedy for the victims of an industrial accident, and neither applied a general duty of care to the handling of dangerous substances. The Factories Act, 1948 contained general duties about fencing machinery, safety officers in larger factories and the health of workers inside the gate; it said nothing systematic about hazardous processes as a class, nothing about the population outside the gate, and nothing about disclosing to either workforce or neighbourhood what substances a plant held and what they could do. The law of tort, which in England and the United States supplied the machinery for mass injury claims, was in India undeveloped to a degree that the Union of India itself later pleaded in the American proceedings, arguing that its legal system lacked the experience and the procedural tools for litigation of this scale. There was no no-fault relief scheme, no environmental fund, no specialist forum, and no settled Indian doctrine on the liability of an enterprise whose lawful activity escaped and killed. The doctrine that did exist in the common law inheritance was the English rule in Rylands v Fletcher, strict liability for the escape of dangerous things brought onto land, hedged with exceptions for acts of God, acts of third parties and statutory authority, and never tested in India against a multinational defendant.

What regulation existed was sectoral and permissive in form. The Insecticides Act, 1968 licensed the manufacture of pesticides, the activity the Bhopal plant existed for, through registration and licensing machinery directed at product quality and trade, not at the catastrophic potential of the process. The Factories Act’s own hazardous-operations provision, Section 87, empowered state governments to make rules for operations declared dangerous, a power exercised unevenly and framed around the safety of the persons employed. The Explosives Act and the petroleum and inflammable substances legislation governed their own narrow subjects. None of these instruments asked the questions a disaster asks: who bears the loss when the permitted activity escapes, on what standard, in which forum, and with what money available in the interval before liability is decided. The patchwork regulated the plant’s inputs and its paperwork. The risk it stored was, in legal terms, unpriced and unallocated, and it stayed that way until the risk materialised.

The plant itself illustrated the gap between the permissions regime and the risks inside it. Methyl isocyanate was an intermediate in the manufacture of the pesticide carbaryl, stored in bulk in tanks at the Bhopal plant. A legal regime built on licences and inspections had permitted the storage; no legal regime had priced the consequence of its escape, assigned in advance who would pay for mass injury, or said which court, applying what standard, would decide the claims of a city’s worth of victims. When the escape came, every one of those questions had to be improvised under emergency conditions, and the improvisations are the legislation. The order in which they came matters. Compensation machinery came first, because the victims existed before any principle did. Preventive and regulatory law came next. The liability doctrine came from a court, in a different dispute. The insurance statute came six years after the disaster. The specialist tribunal came last and never arrived at all, and the criminal law, which ran underneath the whole sequence, was shaped at every stage by decisions taken in the compensation case.

The emergency weeks

The state’s first responses were administrative, and they set the pattern of improvisation the legislation then codified. The surviving stock of methyl isocyanate in the plant’s other tanks remained a live danger to a traumatised city, and between 16 and 22 December 1984 the remaining gas was converted into pesticide in a controlled operation, codenamed Operation Faith, conducted under scientific direction with the population evacuated from the surrounding area. Warren Anderson, the chairman of the parent corporation, arrived in Bhopal on 7 December 1984, was arrested, was released on bail the same day, and left India, a sequence that fixed the criminal case’s central absence at its very beginning. The state government paid out ex gratia relief to the bereaved and the hospitalised from its own funds, in amounts that were emergency assistance rather than compensation. American plaintiffs’ lawyers arrived in the city within days and signed retainers with thousands of victims, a spectacle that offended Indian official opinion and supplied one of the practical arguments for consolidating the claims: if the victims’ rights were not gathered into a single hand, they would be gathered, piecemeal and on contingency terms, in foreign courtrooms. Proceedings were in fact begun in the United States almost at once. The Union government, confronted with a liability contest already internationalised, claims already being marketed, and a domestic legal system with no mass tort machinery, chose consolidation by statute, and chose it at ordinance speed. The ordinance of 20 February 1985 preceded the Act, and the Act followed within weeks, a passage timetable that tells its own story about the balance of the argument in Parliament: the displacement of the individual claimant, radical as it was, was debated against a visible alternative of chaos, and chaos lost.

The medical aftermath ran on a parallel track of uncertainty that later entered the legal record. The treatment of mass methyl isocyanate exposure had no established protocol, and a scientific controversy developed over the composition of the gas cloud and the appropriateness of sodium thiosulfate as an antidote, a controversy conducted between the Indian Council of Medical Research’s scientists, treating physicians in Bhopal and the company’s medical advisers. The dispute was not academic. What the gas had done to lungs, eyes and immune systems determined what the claims categories had to recognise, and the long argument over whether the exposure produced continuing systemic illness, or a bounded injury that time would close, is one root of the irreconcilable casualty and morbidity figures examined later in this article. The Council of Medical Research mounted a long-term epidemiological programme following the exposed population through the following decade, and the discontinuation of that follow-up in the mid-1990s was itself criticised by the victims’ organisations as the abandonment of the only systematic evidence base; a dedicated National Institute for Research in Environmental Health was established at Bhopal under the Council in 2010, a quarter-century after the exposure whose consequences it exists to study.

The first statute: the Bhopal Gas Leak Disaster (Processing of Claims) Act, 1985

Parliament’s first answer was procedural, and it was drastic. The Bhopal Gas Leak Disaster (Processing of Claims) Act, 1985 was promulgated first as an ordinance and enacted as Act 21 of 1985, receiving the assent of the President on 29 March 1985 and taking effect from 20 February 1985. Its preamble purpose was to confer powers on the Central Government so that claims arising out of the disaster would be dealt with speedily, effectively, equitably and to the best advantage of the claimants. The operative provision, Section 3, did something no Indian statute had done on this scale. It gave the Central Government the exclusive right to represent, and to act in place of, every person who had made or was entitled to make a claim, whether within India or outside it, for all purposes connected with the claim and with the same effect as the claimant personally. The purposes were spelled out to include instituting or withdrawing any suit or proceeding anywhere in the world, and entering into a compromise. A claim was defined widely, covering loss of life, personal injury, damage to property, the expenses of containing the disaster, and loss of business or employment. The individual victim’s right to sue Union Carbide, in India or in America, in person, was extinguished by the statute and replaced by the state.

Section 4 was the concession built into the displacement. In representing a claimant, the Central Government was required to have due regard to any matter the claimant wished to have urged, and a claimant who desired it could have a legal practitioner of his own choice associated with the conduct of the proceedings, at the claimant’s own expense. Section 6 provided for a Commissioner and officers to administer the claims machinery. Section 9 empowered the Central Government to frame a Scheme, and under it the Bhopal Gas Leak Disaster (Registration and Processing of Claims) Scheme, 1985 created the working apparatus: registration of claims, their categorisation by deputy commissioners into classes running from death and permanent disability through temporary disability and minor injury to loss of property and livestock, determination of the compensation payable in each category, appeals within the machinery, and disbursement through dedicated claims courts. The scale of the paperwork alone was a legislative fact. The claims filed eventually numbered 10,29,517 on the official count given in government replies, and the categorisation exercise, conducted across forty claims courts, ran for the better part of two decades. Anyone who works through the digitised residue of that record, claim registers and categorisation data held in spreadsheet and document form, handles it through ordinary office software, and a browser-based office file viewer for Excel, Word and PowerPoint documents is the practical way to open such files without the full suite installed. The account that follows describes a historical statutory scheme of general application; it is an account of what the law did, not advice on any individual claim.

Why did Parliament put the state in charge of every Bhopal claim?

Because the alternative was six hundred thousand individual lawsuits against a foreign corporation, brought by poor, ill and bereaved claimants facing a defendant able to litigate causation victim by victim for decades. One plaintiff speaking for all could negotiate, settle and distribute at a scale no aggregation of individual suits could match, abroad as well as at home.

The justification had a doctrinal name, parens patriae, the state as parent of the nation, guardian of those unable to protect themselves. The government’s case was that mass disaster victims are not ordinary litigants: they could not fund American litigation, could not assemble medical and engineering evidence against a multinational’s experts, and could not wait out a contested trial on liability, causation and quantum that would run for a generation. The critics’ case, put most forcefully by the jurist Upendra Baxi, who wrote on the litigation from its first phase, was that the statute converted victims into beneficiaries of state discretion, stripped them of any voice in the decisions that disposed of their rights, and made the Indian state simultaneously the claimants’ representative, a regulator whose own failures were part of the background, and a government courting foreign investment with reasons of its own to close the matter. Both cases were honestly available on the facts. The statute’s defenders could point to the outcome that no purely private litigation in India had ever produced anything comparable for mass tort victims, because there was no class action culture, no contingency fee profession of the American kind, and no developed tort jurisprudence to run. The critics could point to what the exclusive mandate made possible four years later, when the state, acting alone, settled every claim and every criminal proceeding in a bargain the victims had no part in making. The displacement question and the settlement question are the same question at two removes, and the constitutional challenge to the Act forced the Supreme Court to answer the first of them before the settlement tested the second.

The machinery the Scheme built deserves a closer account, because it became the largest claims adjudication exercise in Indian legal history and its design choices decided what the settlement was worth to each family. At its head stood the Welfare Commissioner for Bhopal gas victims, the office charged with supervising registration, categorisation and disbursement. A claimant filed a claim, supported by whatever evidence of death, injury or loss existed, hospital records, death certificates, ration cards, or, in their frequent absence, the testimony of neighbours and the records of the relief operations. Deputy commissioners categorised each claim after inquiry, assigning it to a class that carried its own compensation range, and the claimant dissatisfied with the categorisation or the amount could appeal within the machinery to the Additional Commissioner, with further recourse structured through the claims courts established for the purpose. The design was administrative rather than adversarial: no claimant had to prove negligence, identify a tortfeasor or survive a contest on causation in the individual case, because liability had been globalised in the settlement and the only questions were identity, exposure and category. That was the scheme’s great mercy and its great defect in one. It paid people whom no tort system would have reached, the undocumented, the migrant, the family whose breadwinner’s name appeared in no register, on evidence a court would have rejected. It also compressed radically different injuries into a handful of categories, so that the categorisation decision, taken by an official on a file, mattered more to a family’s recovery than any legal argument, and the disputes that tort litigation conducts about liability reappeared inside the machinery as disputes about category boundaries. The exclusions cut hardest: claims for deaths that occurred years later from gas-related illness had to fight the category definitions, and the adjudication returns record that a substantial number of death claims were awarded at injury rates on evidential grounds, a grievance the victims’ organisations carried into every later proceeding. Section 11 gave the Act overriding effect over other law, foreclosing parallel routes, and Section 3(3) reached backwards to claims already filed abroad, subject to the foreign court’s permission, so that the consolidation was as complete as Indian legislation could make it. When the Act is called a monopoly, the description is statutory fact; when it is called a necessity, the description is the judgment the next section examines.

The passage of the Claims Act

The Act’s passage history explains its shape. It began as an ordinance because the emergency would not wait for a session: promulgated on 20 February 1985, eleven weeks after the disaster, while American proceedings were multiplying and the first interim relief payments were being made from state funds. When Parliament took up the bill in March 1985, the argument concentrated, as it had to, on Section 3. The government’s case was administrative and diplomatic together: a single claimant could negotiate with a multinational from sovereign equality, could appear in foreign courts without the spectacle of competing Indian plaintiffs, could bind every victim to one bargain and so make a bargain possible at all, and could distribute a fund through machinery designed for the purpose. The objections, voiced in the debates and elaborated in the constitutional challenge that followed, were that the bill expropriated a vested right, the citizen’s own cause of action, without consent; that the power to compromise in Section 3(2) was a power to surrender; and that Section 4’s safeguards, due regard to the claimant’s wishes and the right to associate one’s own lawyer at one’s own expense, were courtesies rather than controls, since the expense condition put them beyond the reach of the destitute claimants who were the Act’s justification. The gap between the introduced design and any alternative Parliament might have chosen, a statutory fund with individual opt-outs, a court-supervised class mechanism on the American model, a claims commission without the settlement power, was never seriously explored, because no such machinery existed in Indian law to be adapted and the government argued, with force, that invention under fire would produce delay without protection. The bill passed substantially in the ordinance’s form, received assent on 29 March 1985, and was deemed to operate from the ordinance’s date. A legislative history has to record what the timetable did to the text: a measure extinguishing the litigation rights of several lakh people was law within four months of the disaster, drafted in weeks, and its most consequential clause, the compromise power, was the clause under which, four years later, both the claims and the criminal prosecutions were closed in a single order. Passage records predict operation; this one’s speed predicted its use.

The contrast between the Act’s speed and the machinery’s pace is the first lesson of the claims record. The statute that took weeks to enact governed an adjudication that took the better part of two decades to complete: the forty claims courts worked through more than ten lakh filed claims over some fourteen years, with the bulk of adjudication complete only by 2003, and disbursement, including the pro-rata second round of 2004, running behind adjudication. For the individual claimant, the sequence had a human arithmetic. A victim categorised in the early 1990s received the first payment years after the injury, and the pro-rata enhancement a decade after that, so that the settlement’s promise of speed over litigation was redeemed, if at all, only against the counterfactual of a contested tort suit, not against the needs the money was for. The legislative history should record this without flinching in either direction. The consolidation did what it was built to do: it produced a single fund, a single defendant-facing negotiation, and a distribution that reached more than five and a half lakh recognised claimants, a reach no Indian litigation of the era could have approached. It also normalised a timescale, disaster to completed payment inside twenty years counted as administrative success, that the instruments of 1991 and 1995 were designed to abolish and did not.

Charan Lal Sahu and the defence of parens patriae

The constitutional challenge reached a Constitution Bench in Charan Lal Sahu v Union of India, decided in December 1989 and reported at (1990) 1 SCC 613. The petitioners attacked the 1985 Act as a violation of the claimants’ rights, principally on the ground that the state could not lawfully confiscate a citizen’s cause of action and conduct it, compromise it and conclude it without the citizen’s consent or even a hearing. The Bench, speaking principally through Chief Justice Mukharji, upheld the Act. A victim of a mass disaster, the Court reasoned, is no match for a multinational corporation or for the conditions of litigation after a catastrophe, physically, mentally, financially or in access to proof, and people in that position need the protection of the state and come within its sovereign umbrella to assert and maintain their rights against the wrongdoer. The parens patriae principle, applied in that setting, was jurisprudentially permissible, and the exclusive representation provision survived.

The judgment is more interesting for what it conceded. The majority accepted that the victims had not been heard before the settlement, examined the complaint that justice had been done without appearing to be done, and held that although a settlement concluded without notice to the victims was not quite proper, a post-decisional hearing would serve no useful purpose on the material available, since the victims had been unable to show any further data that would impeach the basis on which the figure had been reached. That is a holding with a hard edge. It validated the structure of the 1985 Act, the state as sole representative, while recording, in terms, that the represented had been excluded from the act of representation at the decisive moment. Later argument about Bhopal returns to that tension repeatedly: the doctrine was defensible in the abstract, the Court in substance said, and its application produced a bargain the beneficiaries never approved. The distinction between the doctrine and its application, stated in the brief for this article as the honest framing, is the distinction the Court itself drew.

The judgment also located the Act within the Constitution’s own distribution of burdens. The Bench treated the disaster as engaging the state’s obligations toward life and livelihood under Article 21, and the parens patriae power as an incident of sovereignty that a legislature could organise by statute for a defined class of claims. Critics of the decision, Baxi among them, answered that sovereignty language concealed a conflict of interest the Court declined to examine: the state that appointed itself the victims’ guardian was also the licensor of the plant, a defendant in waiting on its own regulatory failures, and a government with macroeconomic reasons to reassure foreign capital. That critique has never been judicially answered, because Charan Lal Sahu was the last occasion on which the Act’s validity was seriously in issue, and its doctrinal legacy is therefore double. It stands as the authority that mass disaster claims in India may be collectivised by statute, the foundation on which every later group compensation scheme implicitly rests. It also stands as the authority that the collectivised claimant’s right to be heard before his claim is compromised may be satisfied, in the catastrophe setting, by the opportunity to be heard afterwards. Both propositions were available to the Constitution Bench that closed the compensation question in 2023, and the second of them is the one the victims’ organisations cite when they describe the legal history of Bhopal as a history of being represented without being consulted.

The American detour and the suit at Bhopal

The claims began where the defendant’s parent was. Suits were filed in the United States, including consolidated proceedings before the District Court for the Southern District of New York, where the Union of India, acting under the 1985 Act, pursued Union Carbide Corporation. Judge Keenan dismissed the consolidated action in 1986 on the ground of forum non conveniens, the doctrine under which a court declines a case that belongs more appropriately elsewhere. The reasoning rested on the location of the evidence, the witnesses, the plant and the victims in India, on Union Carbide’s undertaking to submit to Indian jurisdiction, and on a pointed observation that for an American court to retain the litigation would be a form of imperialism, displacing the courts of a sovereign nation judging harm done to its own people. The Indian government had argued the opposite of its later position at home: that Indian courts lacked the procedural machinery, the tort experience and the speed for litigation of this magnitude. The argument failed in New York, and the dismissal was substantially upheld on appeal in the Second Circuit in 1987, with modifications to the conditions attached to it, and the case came to Bhopal.

The conditions attached to the dismissal were the bridge on which the case crossed to India, and they deserve record because later Indian proceedings rested on them. Judge Keenan dismissed on terms that Union Carbide consent to the jurisdiction of the Indian courts, waive any defence based on limitation, and accept a measure of exposure to the enforcement of an Indian judgment, conditions the federal appellate court adjusted in detail when it upheld the dismissal. The architecture was deliberate: the American court would surrender the case only against the defendant’s surrender of the procedural advantages of the move, so that India would receive not merely the litigation but a defendant already bound to appear, already stripped of the time-bar, and already committed on enforcement. The irony that followed is part of the record. Within two years the defendant, present in India on those undertakings, was contesting the power of the Bhopal court to order interim payment at all, and within three the entire structure of undertakings and counter-undertakings had been absorbed into a settlement that the Supreme Court brokered in the appeals from that very order. Forum non conveniens decided where Bhopal would be litigated; it could not decide how, and the how consumed the following thirty-seven years.

Why was the Bhopal compensation case first fought in the United States?

Because the parent company, its assets, documents and decision-makers were there, because American tort practice offered contingency fees, juries and discovery on a scale Indian procedure did not, and because the victims’ representatives judged that a New York courtroom was where Union Carbide could least afford to lose. The forum was decided before liability was ever reached.

In the District Court at Bhopal, the Union of India filed Suit No. 1113 of 1986, claiming damages of approximately three billion US dollars against Union Carbide Corporation and Union Carbide India Limited. The suit’s first great battle was not over liability but over interim compensation, money for the victims before any trial. In December 1987, by an order dated 17 December 1987, the District Judge directed Union Carbide to deposit Rs 350 crore as interim compensation, invoking the court’s inherent powers. Union Carbide challenged both the amount and the very idea of interim compensation in a damages suit where liability was denied. The Madhya Pradesh High Court, by its order of 4 April 1988, reduced the sum to Rs 250 crore. Both sides appealed to the Supreme Court, the Union against the reduction and the company against the permissibility of the award itself. It was those appeals, with the interim compensation question open and the liability question untouched, that the Supreme Court had before it when it chose to end the entire litigation instead.

The interim compensation episode deserves its place in the doctrinal record, because it was an innovation the settlement prevented from maturing. In the common law tradition a plaintiff in a damages action waits for judgment; interim payment before liability is established was exceptional, and Union Carbide’s appeal raised the objection in its pure form, that no court could order payment on account of damages whose very foundation, liability itself, the defendant denied and no court had tried. The District Judge had rested the order on the court’s inherent powers, and the High Court, while reducing the amount, had sustained the power, in substance reclassifying the payment as interim damages against the enormity of uncompensated suffering. Had the Supreme Court decided the appeals, Indian tort law would have received, in 1989, an authoritative answer on interim compensation in mass disaster litigation, an answer of obvious value for every later catastrophe. The settlement withdrew the question. The point is not antiquarian. The instruments Parliament built afterwards can be read as statutory substitutes for the missing doctrine: the insurance Act of 1991 is interim compensation generalised and depersonalised, payable without any court order at all, and the tribunal of 1995 was to have had the power to move money before final adjudication as a matter of design. The judiciary’s improvised interim remedy of 1987 and 1988, contested and unfinished, is the bridge between the Bhopal suit and the statutory scheme, and the settlement, which answered the money question for Bhopal’s victims in one global payment, is the reason the bridge was never completed as doctrine.

The settlement of February 1989

On 14 February 1989, a five-judge bench of the Supreme Court, headed by Chief Justice Pathak, recorded an overall settlement. Union Carbide Corporation was to pay US$470 million to the Union of India in full settlement of all claims, rights and liabilities related to and arising out of the disaster, payable by 31 March 1989. All civil proceedings stood transferred to the Supreme Court and concluded in terms of the settlement. And, in the clause that defined the controversy ever after, all criminal proceedings related to and arising out of the disaster stood quashed wherever they were pending. The detailed reasons followed on 4 May 1989, reported at (1989) 3 SCC 38. The sum, the Court explained, lay between the offers the parties had exchanged, which had run from US$426 million to US$500 million, and it had been computed on assumptions the Court stated openly: that the death cases were about 3,000 and the injured about 105,000, with a higher standard of compensation applied than ordinary tort principles would have produced. The Court also recorded the immediate toll figure of 2,660 deaths that stood in the material before it, a number worth pausing on, because it shows the informational world inside which the largest industrial compensation bargain in Indian history was struck: the Court was pricing a disaster whose dimensions were still estimates, and estimates supplied largely by the parties.

Why did the Supreme Court accept a settlement instead of deciding the case?

Because the claimants were destitute, and the alternative was a contested liability trial against a foreign defendant whose assets within Indian reach were limited, followed by appeals measured in decades. A certain fund, paid within weeks, was worth more to people needing money for treatment and survival than a larger judgment that might never be executed.

The Court’s reasoning deserves its full statement, because later criticism often answers a weaker version of it. The litigation, the Court said in substance, was unusually complex in law and fact, involving novel questions of multinational liability, causation across a mass exposure, and the quantification of loss for a population with almost no documentary record of income or health before the disaster. Delay was not an abstraction for the claimants; people were dying and being treated, or not treated, while the interim compensation appeals alone had consumed more than a year. Union Carbide’s undertaking to submit to Indian jurisdiction had not brought with it assets in India sufficient to satisfy a decree of the size the Union claimed, and execution against the American parent would have opened a second generation of litigation in the United States. A settlement converted all of that risk into a fund. Judges approving compromises apply that calculus routinely, and the five judges applied it to a case in which every element of the calculus, the desperation of the claimants, the evidential difficulty, the enforcement risk, cut the same way. The payment was made; with the exchange rate of the day the fund was worth a little over Rs 700 crore, and as interest accrued and the rupee moved, the amount ultimately disbursed to claimants grew to roughly Rs 3,000 crore on the Union’s own later account, a fact the Union itself placed before the Court in 2023 and to which this history returns.

The mechanics of the fund explain the growth. The US$470 million was deposited with the Reserve Bank of India in dedicated accounts, converted into rupees as disbursement required, under an arrangement by which the Union bore the exchange risk, so that the depreciation of the rupee and the accumulation of interest enlarged the rupee fund well beyond the Rs 700 crore it represented in 1989. That enlargement is the reason the final disbursement figure and the settlement figure are both true and entirely different numbers, and the reason arguments that compare them directly, in either direction, mislead. The company’s side can point to a fund that ultimately paid out roughly four times its original rupee value. The victims’ side can answer that the enlargement came from exchange movement and interest on money withheld from claimants for years, not from any addition by the corporation, and that a fund divided among five times the assumed number of victims left each recognised claimant with a fraction of what the settlement’s own arithmetic had contemplated per person. The arithmetic is common ground; the characterisation is the dispute.

One feature of the road to February 1989 should be fixed before the criticism is assessed, because it conditions the whole event. The settlement was not simply a bargain the parties brought to a passive court. The appeals against the High Court’s interim compensation order were the occasion, and the Supreme Court’s order of 14 February 1989 records the Court’s indebtedness to counsel for accepting the terms of settlement the Court itself had suggested. The initiative, the figure and the global scope, civil and criminal together, were products of the hearing, shaped by the Bench’s assessment that the case was pre-eminently fit for overall settlement. That origin matters to both sides of the later argument. For the settlement’s critics, it meant the bargain wore the authority of the Court from the first day, making it harder to attack than an executive deal would have been, and it entangled the judiciary’s own prestige in the figure’s defence at every later stage, in the review of 1991 and the curative proceedings of 2023 alike. For its defenders, the Court’s authorship is the answer to the charge of a sell-out by the executive: five judges, with the mass of data before them, fixed the number. The settlement was, in form and in fact, a judicial act, and the litigation that followed for thirty-four years was in substance an argument about how far a judicial act of that kind can be revisited.

The criticism of the settlement, stated at full strength

Against that reasoning stands a body of criticism that has never been answered on its own terms, and the settlement’s defenders do the record no service by understating it. The figure was built on a census of harm that proved wrong. The Court assumed about 3,000 deaths; the claims machinery, applying the statute’s own categories over the following years, awarded compensation in 5,295 death cases, and recognised injury on a scale five times the assumption, with awards in 5,74,376 cases in all out of 10,29,517 claims filed, figures given in official replies concerning the Welfare Commissioner’s adjudication. A fund sized for about 105,000 injured was distributed across more than five and a half lakh recognised claimants, so the per-person value of the settlement collapsed in the distribution, whatever its aggregate size. Amnesty International, in its account of the disaster’s aftermath, put the deaths in the immediate aftermath at more than 7,000 and the deaths in the following years at a further 15,000, figures irreconcilable with the settlement’s arithmetic and never reconciled with it. Victims’ organisations, among them the Bhopal Gas Peedith Mahila Udyog Sangathan and the Bhopal Group for Information and Action, made the second objection, of process: the claimants had no seat at the table, because the 1985 Act had taken their claims, and the settlement of their claims without their consent or a prior hearing was, in the phrase the Charan Lal Sahu majority itself could not avoid, justice done without appearing to be done. Upendra Baxi’s criticism ran deeper still, to the proposition that a state which had licensed and inadequately regulated the plant was not a disinterested parent, and that its eagerness for a global settlement, closing criminal liability along with civil claims, served interests beyond the victims’.

The third objection was constitutional in character and proved to be the one the legal system itself accepted. The quashing of the criminal proceedings was not incidental to the bargain; it was a term of it, and it traded the prosecution of individuals for a corporation’s payment in a way that sat uneasily with the principle that criminal liability is owed to society and is not the private property of litigants to sell. Review petitions attacked the settlement orders, and on 3 October 1991 the Supreme Court delivered its judgment in review, reported at (1991) 4 SCC 584. The review dismissed the challenges to the civil settlement and left the US$470 million bargain intact. But it held the quashing of the criminal proceedings to be unsustainable and restored the prosecutions. The partial correction matters to the narrative: the system conceded that one element of February 1989 had exceeded what a settlement may lawfully do, while holding the money bargain closed. The same review judgment directed the Union to arrange insurance cover for the victims against a shortfall in the settlement, a direction whose non-implementation the Court would still be recording, in severe language, thirty-two years later. With the civil question closed and the criminal question reopened, the legislative sequence resumed, and it is to the statutes, rather than the litigation, that the disaster’s permanent legal legacy belongs.

The review judgment’s division of the settlement into a sustainable civil part and an unsustainable criminal part is worth one further paragraph, because it fixed the boundary of what a court-brokered compromise may do. The petitioners had attacked the settlement orders as an exercise of power beyond the Court’s authority, including under Article 142, the power to pass orders necessary for doing complete justice. The Bench upheld the civil settlement as a lawful compromise of the claims the Union was statutorily empowered to compromise, reasoning that the power to settle carried the power to conclude the civil proceedings. The quashing of criminal proceedings stood on different ground. Criminal liability is not a private asset; a prosecution for offences arising out of thousands of deaths is not the Union’s to surrender as consideration in a civil bargain, and the term that extinguished the prosecutions could not be sustained as part of a settlement of claims. The distinction the review drew, between disposing of claims, which the 1985 Act authorised, and disposing of prosecutions, which nothing authorised, is the reason the criminal history in this article exists at all after 1991, and the reason the settlement’s most criticised feature is also the one the legal system itself disowned. The money stayed; the immunity went. Whatever else is said about February 1989, that severance is the system’s own verdict on its most extraordinary term.

The umbrella statute: the Environment (Protection) Act, 1986

The second legislative response was general, and it came fast. The Environment (Protection) Act, 1986 received the assent of the President on 23 May 1986 as Act 29 of 1986, less than eighteen months after the leak, and was brought into force on 19 November 1986. It was enacted under Article 253 of the Constitution, the power to legislate for implementing international agreements, and its preamble recites the decisions taken at the Stockholm Conference of 1972; Article 114 of this series owns the Act’s architecture in depth, and the full account of its powers, its rules and its enforcement record is in the Environment (Protection) Act, 1986 explained. What belongs to this history is its place in the sequence. Where the Water Act and the Air Act had regulated single media through boards, the 1986 Act gave the Central Government omnibus power over the environment as a whole: to set standards for emissions and discharges, to restrict the location of industries, to prescribe procedures and safeguards for handling hazardous substances, to enter, inspect, sample and analyse, and, under Section 5, to direct the closure of an industry or the stoppage of its electricity and water. It created the framework inside which hazardous substances rules, chemical accident rules and environmental impact assessment would later be made.

Two features of the Act are legible only against Bhopal. The first is its speed and its form. A legislature reacting to catastrophe does not draft a code of standards, because a code takes years and the next accident resembles the last one in no particular; it drafts a power, and reserves almost nothing to itself. The 1986 Act is twenty-six sections of delegated authority, and Bhopal is the reason it exists in that shape. The second feature is what it did not contain. The Act created no compensation remedy, no victim fund and no liability rule. The Central Government could close a plant; a widow could not use the Act to recover a rupee. That omission was not an oversight. Compensation for industrial accidents was being handled, in the legislature’s mind, by the claims machinery for Bhopal itself and by two instruments still to come, the insurance statute of 1991 and the tribunal statute of 1995, and the umbrella Act’s silence on compensation is the space those two statutes were meant to fill.

The Act’s delegated space filled quickly with the disaster’s specific lessons. The Manufacture, Storage and Import of Hazardous Chemicals Rules, 1989, made under the Act, imposed on the occupier of an industrial activity involving hazardous chemicals a regime of notification, safety reports and safety audits, on-site emergency plans approved by the designated authority, and the communication of information to persons outside the site who could be affected by a major accident, a direct statutory descendant of the Bhopal failure to warn. The Chemical Accidents (Emergency Planning, Preparedness and Response) Rules, 1996 added the crisis-management architecture, crisis groups at central, state, district and local level charged with preparing off-site emergency plans for chemical accident districts. Neither instrument is a statute, and neither belongs in the count of four Acts, but both belong in the legislative history, because they show the 1986 Act performing the function for which its shape was chosen: the framework stood, the specification followed, and the specification is where Bhopal’s operational lessons, the emergency plan, the safety audit, the duty to inform the neighbourhood, were actually written into enforceable law.

The framework form carried a cost that the legislative history should also record, because it shaped everything the Act could not do. A statute of powers depends entirely on the energy of the power-holder: rules are made when the ministry makes them, standards bite when the boards enforce them, and the citizen’s only lever is the constitutional court. In the years after 1986 the Supreme Court’s environmental benches became, in effect, the Act’s enforcement engine, converting its general powers into specific directions in case after case, which is why the jurisprudence article in this series carries so much of the 1986 Act’s real history. Bhopal’s umbrella statute is therefore best characterised as a transfer of initiative, from a Parliament that had legislated in emergency to an executive armed with every power and a judiciary willing to supervise its use. That characterisation explains the Act’s durability, twenty-six sections never substantially expanded because they never needed to be, and its recurring criticism, that a framework so dependent on delegated energy reproduces, at the level of the whole environment, the same vulnerability the tribunal statute showed at the level of one forum: what is not commenced, notified or enforced under it does not exist, whatever the statute book says.

The hazardous processes chapter in the Factories Act, 1948

The third response addressed the inside of the factory gate. The Factories (Amendment) Act, 1987, Act 20 of 1987, inserted a new Chapter IVA, Sections 41A to 41H, into the Factories Act, 1948, with effect from 1 December 1987, the permissible-limits provision following on 1 June 1988. The chapter was built around a statutory category that had not existed before, the hazardous process, defined by reference to a schedule of industries whose materials and processes could, unless special care was taken, cause material impairment to the health of persons inside the factory or in the general public, a definition that for the first time in factory law treated the surrounding population as within the statute’s concern. The detailed operation of the parent Act is treated in the Factories Act, 1948 explained; the chapter’s provisions, which are the Bhopal-specific layer, repay summary here.

Section 41A requires a Site Appraisal Committee, chaired by the Chief Inspector and including representatives of the pollution control boards and the local authority, to advise the state government on applications for the initial location of a factory involving a hazardous process or for its expansion, so that siting became a considered decision rather than a licensing formality. Section 41B imposes compulsory disclosure: the occupier must reveal all information about dangers and health hazards, and the measures to overcome them, to the workers, to the Chief Inspector, to the local authority and to the general public in the vicinity, must lay down a written health and safety policy, must disclose the quantity and characteristics of wastes and their manner of disposal, and must draw up an on-site emergency plan and make its safety measures known to workers and neighbours alike. Section 41C fixes specific responsibility on the occupier in relation to hazardous processes, including accurate and up-to-date health records for workers exposed to chemical and toxic substances, supervision of such processes by persons with qualifications and experience in handling hazardous substances, and medical examination of workers before assignment to a hazardous process and at prescribed intervals thereafter. Section 41D empowers the Central Government to appoint an inquiry committee where a hazardous factory’s standards are in question. Section 41E provides for emergency standards where none have been prescribed. Section 41F writes permissible limits of exposure to chemical and toxic substances into a schedule to the Act. Section 41G requires a safety committee in every hazardous factory with equal representation of workers and management, the first statutory foothold for workers’ participation in safety management. Section 41H gives workers the right to warn the occupier and, through the prescribed channel, the inspectorate, about imminent danger, with a duty on the occupier to act and to record the action.

Read against the disaster, the chapter is a point-by-point negation of the conditions that had made Bhopal possible in law: a plant sited without appraisal of what surrounded it, substances whose dangers were not disclosed to workforce or neighbourhood, no statutory health surveillance of the exposed, no emergency plan owed to the public, and no institutional channel through which a worker who saw the danger coming could speak with legal protection. Whether the inspectorates acquired the capacity to run the chapter is an implementation question that belongs to the Factories Act article; the legislative fact is that the duties exist, in that form, because of December 1984.

The chapter’s teeth were set by the same amending Act. A new Section 96A made contravention of Sections 41B, 41C or 41H, the disclosure duty, the occupier’s specific responsibility and the workers’ warning right, punishable with imprisonment extending to seven years and a fine extending to two lakh rupees, with a further daily fine for a continuing contravention and imprisonment extending to ten years where the contravention continued beyond a year after conviction. Measured against the general penalty provision of the parent Act, which capped most contraventions at two years, the seven-year term was Parliament’s statement that the hazardous process duties were of a different order from the ordinary run of factory regulation: the failure to disclose what a plant holds, or to maintain the health records of the exposed, was made one of the most severely punishable offences in factory law. The disclosure philosophy is the thread to notice. Before 1987, information about industrial danger moved, if it moved at all, inside the inspectorate. After Chapter IVA, the occupier owed the truth about his substances to three audiences at once, his own workers, the state, and the public in the vicinity, and the third audience is the Bhopal audience, the people outside the gate who in December 1984 had no legal right to know what was stored across the road or what it would do to them. A statute cannot retrospectively inform the dead; it can, and this one did, make ignorance of the neighbourhood unlawful for every comparable plant that followed.

Absolute liability: the doctrine Bhopal provoked and never received

The doctrinal response came from a court, and from a different leak. In the night of 4 December 1985, a year and a day after Bhopal, oleum gas escaped from the Shriram Foods and Fertilizer Industries plant in Delhi, and a further leak followed on 6 December 1985; an advocate died and many were injured. The environmental lawyer M. C. Mehta already had a petition under Article 32 before the Supreme Court seeking the closure or relocation of the plant, and the leak converted that petition into the occasion for the most consequential Indian judgment on industrial liability. On 20 December 1986, a Constitution Bench of five judges, speaking through Chief Justice Bhagwati, delivered judgment in M. C. Mehta v Union of India, reported at (1987) 1 SCC 395.

The Court refused to let Indian law be governed by the English rule in Rylands v Fletcher. That rule, strict liability for the escape of dangerous things, came encumbered with exceptions, for act of God, for the act of a stranger, for statutory authority, which had been developed for a nineteenth-century economy of mills and reservoirs. An enterprise engaged in a hazardous or inherently dangerous activity, the Court held, owes an absolute and non-delegable duty to the community to ensure that no harm results from the activity. If harm does result from an accident in the operation of that activity, the enterprise is strictly liable, without exception, to compensate everyone affected, and the liability is subject to none of the exceptions that qualify the English rule. The Court added a second proposition that was equally deliberate: the measure of compensation must be correlated to the magnitude and capacity of the enterprise, so that the damages have a deterrent effect proportioned to the enterprise’s ability to pay, a deep-pocket principle that rejected the idea of valuing mass injury by the poverty of the injured. The passage of the doctrine into the wider jurisprudence, and its place among the principles the Green Tribunal is directed to apply, is traced in the environmental jurisprudence of the Supreme Court.

Why did the oleum case, and not Bhopal, produce the absolute liability rule?

Because the Bhopal claims never reached a liability adjudication in which a rule was needed: they were displaced into a state-run suit, fought over forum and interim money, and then settled, and a settlement states no principle. The oleum petition put the adequacy of Rylands v Fletcher directly before a Constitution Bench, which had to answer the doctrinal question.

The irony is structural and complete. Absolute liability was articulated because Bhopal had happened; Bhagwati CJ’s reasoning is unintelligible without the catastrophe a year earlier and a few hundred miles away, and the Court said in terms that a rule built for Victorian England could not govern an economy of multinational hazardous industry. Yet the doctrine was never applied to Bhopal itself, because the Bhopal compensation question was closed by bargain before any court was asked to decide liability, and the settlement’s arithmetic owed nothing to the deterrent principle the oleum bench had announced. In the oleum litigation itself, the compensation ordered was modest, twenty lakh rupees together with a bank guarantee of fifteen lakh against future claims, a gap between doctrinal ambition and monetary outcome that runs through this entire field. The doctrine nonetheless became the load-bearing principle of Indian environmental tort law, absorbed into the strict liability scheme of the tribunal statute of 1995 and later into the National Green Tribunal’s governing principles, and it is the answer Indian law gives in examination halls and courtrooms to the question Bhopal posed. The disaster’s most cited legal product was decided in someone else’s case.

The content of the rejected English rule shows what was at stake in the choice. Rylands v Fletcher, decided in 1868, made the person who brings and keeps a dangerous thing on his land liable for its escape, without proof of negligence, but the exceptions that gathered around the rule were wide: an act of God, the default of the plaintiff, consent, statutory authority, and, most dangerously for a mass disaster case, the act of a stranger over whom the occupier had no control. Each exception was an invitation to a decade of litigation, and one of them mapped exactly onto the defence Union Carbide was preparing in Bhopal. The corporation’s own account of the disaster, developed through its investigations, was that the water that entered the methyl isocyanate tank had been introduced deliberately, an act of sabotage by a disgruntled employee. Had the Bhopal liability trial been fought under Rylands, the stranger’s-act exception would have stood at the centre of it, and the compensation of a city’s victims would have turned on proof about the identity and motives of a single unknown workman on a single night. Absolute liability abolished that terrain. Under the oleum rule, the enterprise’s answer to an escape is not a story about who else was at fault; there is no such answer, only the fact of the activity, the escape and the harm, and a measure of compensation scaled to the enterprise’s capacity rather than to its culpability. That is why the doctrine, and not any proposition decided in the Bhopal litigation itself, is the liability law Bhopal left behind: it is the rule a legal system writes when it has watched the exceptions swallow a disaster.

Two features of the oleum judgment complete its account. The first is jurisdictional. The petition had been brought under Article 32, and the Court held that compensation could be awarded in such a proceeding for infringement of the right to life under Article 21, treating monetary redress as an available constitutional remedy rather than remitting victims to the delays of a civil suit. For mass injury litigation, that holding mattered almost as much as the liability rule: it located a forum, the constitutional court itself, in which the victims of a hazardous enterprise could be compensated without the procedural attrition that the Bhopal claimants’ statutory displacement had been designed to avoid. The second feature is the doctrine’s onward path. The tribunal statute of 1995 took strict liability for hazardous accidents as its foundation, in substance the oleum principle codified for a specialist forum. The National Green Tribunal Act, 2010, which replaced the 1995 scheme, provides for relief and compensation in accident cases on a no-fault basis in its own liability provisions and directs the Tribunal to apply the principles of sustainable development, the precautionary principle and the polluter pays principle, the family of doctrines of which absolute liability is the Indian head. Later benches applied the oleum principle in ordering compensation and remediation in pollution cases, and no later bench has disowned it. A doctrine announced in a case about a comparatively small leak, in a proceeding in which the compensation ordered was twenty lakh rupees, became the stated liability philosophy of the Indian environmental state. That disproportion between the occasion and the consequence is the normal way of doctrinal history, and Bhopal’s role in it, provoking a principle it never received, is the sharpest illustration in this series of the difference between a disaster’s litigation and its legacy.

The capacity principle within the doctrine drew its own controversy, which a balanced history should state. To correlate damages with the magnitude and capacity of the enterprise is to make the wealthy defendant pay more for the same injury than the poor one, and critics within the legal profession objected that this converted compensation into punishment without the safeguards of a criminal trial, and would bear hardest on the public sector and the indigenous enterprise while the multinational priced the risk into its insurance. Bhagwati CJ’s reasoning answered the objection at its root: an enterprise that carries on a hazardous activity for profit internalises the profit and must internalise the loss; deterrence that does not scale with capacity does not deter; and the cost of the activity’s accidents is a cost of the activity, to be spread through its prices and its insurers across everyone who benefits from it, rather than concentrated on the neighbours who merely live beside it. That is a complete economic theory of tort in three moves, and its presence in a 1986 judgment, articulated under the pressure of Bhopal, is why the oleum case is taught wherever Indian environmental law is taught, and why the examination literature treats the disaster’s doctrinal response as a subject in itself.

No-fault relief: the Public Liability Insurance Act, 1991

The fourth response attacked the problem that the settlement had exposed from the other end: the years a victim waits while liability is argued. The Public Liability Insurance Act, 1991, Act 6 of 1991, brought into force on 1 April 1991, is a small statute with an exact idea. Where death or injury to any person other than a workman, or damage to property, results from an accident while a hazardous substance is being handled, Section 3 makes the owner liable to give the relief specified in the Schedule on the principle of no fault: the claimant is expressly not required to plead or establish that the death, injury or damage was due to any wrongful act, neglect or default of any person. Section 4 makes the promise collectible by requiring every owner, before starting to handle a hazardous substance above the quantities notified under the Act, to take out insurance against that liability. Claims are made to the Collector of the district, who inquires and makes the award under Section 7, and relief under the Act is in addition to, and does not extinguish, any other right to claim compensation under another law, so the Schedule payment is a floor of immediate money and not, in form, a ceiling on recovery.

The Schedule is where the statute’s ambition meets its arithmetic. The relief it specifies is reimbursement of medical expenses up to Rs 12,500; Rs 25,000 for a death, in addition to medical expenses; for permanent total disability Rs 25,000, and for permanent partial disability a proportion of that sum according to the percentage of disablement certified, again with medical expenses; for loss of wages through temporary partial disability a fixed monthly relief not exceeding Rs 1,000 for up to three months, where the victim was hospitalised for more than three days and is above sixteen years of age; and up to Rs 6,000 for damage to private property. The insurance rules cap the insurer’s aggregate liability at Rs 5 crore for claims arising out of one accident and Rs 15 crore in a year. These are sums designed for the ordinary industrial accident, a boiler explosion, a chlorine leak at a small works, and they have never been revised into proportion with a catastrophe; a Bhopal-scale event would exhaust the insurance cap before the first category of claimants was paid. The statute’s defenders answer that its purpose was never to price a disaster but to abolish the wait for the common run of accidents, putting some money in a victim’s hands on proof of the accident and the injury alone. That defence is worth taking seriously, because the Bhopal claims experience, years of categorisation before payment, was precisely a failure of speed rather than of aggregate funds alone.

The Act’s second institution is the Environmental Relief Fund. Section 7A provides for the establishment of the fund, financed under the rules by a contribution from every owner equal to the premium payable to the insurer, collected with the premium, and the fund stands behind the insurance: where an award exceeds the amount payable under the policy, the excess is met from the fund, and any remaining shortfall falls on the owner. The fund is therefore a deliberate piece of socialisation, a levy on the whole hazardous sector standing behind each owner’s policy. Its history, however, repeats the pattern that this article tracks. The statutory provision stood on the book from the early 1990s, but the Environment Relief Fund Scheme that made the fund operational was notified only on 4 November 2008, a gap of more than a decade and a half between the enactment of the mechanism and the notification that gave it working form, during which contributions collected rested with insurers awaiting a scheme. The fund exists; the delay in constituting it is not an allegation but a matter of gazette dates.

The working of a claim shows the statute’s design intent. The victim, or the legal representative, applies to the Collector of the district where the accident occurred; the Collector holds an inquiry, giving the owner and the insurer their say, and makes an award of the Schedule relief; the insurer pays; where the award in an accident exceeds the policy’s cover, the Environmental Relief Fund meets the excess within the rules’ limits, and the owner bears any residue, so that the claimant is not the one left chasing the shortfall through the layers. The threshold quantities that trigger the Act were specified by notification in 1992, a list of hazardous chemicals with quantities above which the insurance duty bites, tying the Act’s reach to the same chemicals regime the hazardous substances rules administer. The Act’s definitions show the Bhopal statutes interlocking into a single system. Its central term, hazardous substance, is not defined afresh: Section 2 borrows the definition from the Environment (Protection) Act, 1986 and adds the quantity threshold notified under the insurance Act itself, so that the umbrella statute supplies the concept and the relief statute supplies the trigger. Its accident is defined as a fortuitous, sudden or unintended occurrence while handling such a substance, resulting in continuous, intermittent or repeated exposure, a drafting choice that reaches beyond the single explosion to the slow leak and the repeated emission, while excluding accidents by reason only of war or radioactivity. Its owner is the person who owns or has control over the handling of the substance at the time of the accident, extended, in the case of a company, to the directors, managers and officers directly in charge of and responsible for the conduct of the business, language borrowed from the standard template of Indian regulatory offences and aimed at the Bhopal problem of the absent decision-maker: the statute identifies, in advance and by office, the human beings through whom the duty to insure and the liability to pay will run. The design is derivative in the best sense. Each Bhopal instrument assumes the others, the Factories chapter policing the process, the umbrella Act the standards, the insurance Act the first money, and the stillborn tribunal the forum, and the coherence of the whole is the strongest evidence that the sequence was, at the level of drafting, a single considered response spread across six years.

One boundary of the Act is a deliberate policy choice with a Bhopal resonance. Section 3 excludes the workman, the employee of the very undertaking whose substance escapes, because the workman has his own compensation code in the workmen’s compensation legislation. The boundary is defensible as social insurance design, each risk to its own fund, but its consequence is that the statute enacted in Bhopal’s shadow would not have covered the plant’s own dead and injured workers, who were among the disaster’s victims and whose claims under the 1985 Act had to be run, like everyone else’s, through the settlement. The Act is best understood as a statute for the public outside the gate, the bystander and the neighbour, the class Bhopal made impossible to ignore, and its exclusions mark the limits of what its drafters thought a relief statute, as distinct from a compensation code, should attempt.

The Act’s record in operation is the weakest part of its history, and it should be described in the pattern terms the evidence supports rather than in figures that are not reliably published. The reported use of the machinery has been sparse relative to the incidence of industrial accidents: the sums are small enough that many victims and their advisers have not found the Collector’s route worth its own delays, awareness of the insurance duty among smaller handlers of hazardous substances has been uneven, and enforcement of the duty to insure has depended on the same inspectorates whose thinness the Factories chapter already assumes. The design lesson is the inverse of the tribunal’s. The 1995 Act failed by never commencing; the 1991 Act commenced and was simply too small for its subject, a relief mechanism calibrated to the single accident at the single works, enacted by a Parliament whose imagination had been formed by a catastrophe five orders of magnitude larger. Between them, the two statutes mark the distance between legislating a principle and funding it, and they explain why, when the next great industrial accident claims are argued in India, they will be argued, as Bhopal’s were, in the constitutional courts and under the general law, with the specialist instruments playing supporting parts.

The National Environment Tribunal Act, 1995: the statute that never began

The fifth response was the one aimed directly at the Bhopal experience, and it is the one that never operated. The National Environment Tribunal Act, 1995, Act 27 of 1995, was enacted, in the words of its long title, to provide for strict liability for damages arising out of any accident occurring while handling any hazardous substance, and for the establishment of a National Environment Tribunal for the effective and expeditious disposal of cases arising from such accidents, with a view to giving relief and compensation for damage to persons, property and the environment. The idea had a pedigree. The Supreme Court’s environmental benches had repeatedly called for specialised environmental courts. The United Nations Conference on Environment and Development at Rio de Janeiro in 1992, in Principle 13 of its declaration, had called on states to develop national law on liability and compensation for victims of pollution and environmental damage. The 1995 Act was India’s answer: a specialist tribunal, with judicial and expert members, before which the victim of a hazardous-substance accident could claim compensation on a strict liability basis, without proving negligence, from a forum designed for expedition, with an appeal lying to the Supreme Court only on the limited grounds available under Section 100 of the Code of Civil Procedure.

The Act was never brought into force. Its commencement was left, in the standard form, to notification by the Central Government, and the notification was never issued. No tribunal was constituted, no chairperson appointed, no bench sat, no claim was ever filed under it. The Law Commission of India, in its 186th Report of September 2003 on a proposal to constitute environment courts, recorded that the 1995 Act had remained unnotified for eight years, and treated that fact as evidence that the legislative response to industrial accident compensation had stalled at the point of executive action. The Commission’s own proposal is part of the explanation for the stall: it recommended against reviving the narrow accident tribunal and in favour of environmental courts of general jurisdiction, with judicial and expert membership, a recommendation that redirected the reform project toward breadth and made the uncommenced 1995 Act obsolete in the eyes of the very ministry that held its commencement notification. An Act can thus die of its successor’s conception as surely as it can die of opposition, and the interval between the two designs, 1995 to 2010, is the period in which India had, on paper, a strict liability compensation tribunal and, in fact, none. The reasons for the stall are contested and should be stated as contested. Commentators on the Act have alleged pressure from industries handling hazardous substances, which faced a strict liability forum with environmental restoration within its remit. The official trajectory suggests a different and possibly additional explanation: policy moved on. The 1995 tribunal had a narrow brief, accidents involving hazardous substances only, and within a few years the reform conversation had shifted toward a comprehensive environmental tribunal of general jurisdiction, the project the Law Commission’s 186th Report advanced and which eventually produced the broader statute of 2010. A narrow Act awaiting notification is easy to leave awaiting when a wider replacement is being designed. Neither explanation flatters the record. A Parliament that had legislated strict liability compensation for the next Bhopal left the instrument uncommenced for fifteen years, and the victims of every hazardous accident in that interval were remitted to the ordinary courts, the insurance Schedule, and the general law.

The repeal, when it came, was almost silent. The National Green Tribunal Act, 2010, Act 19 of 2010, created the comprehensive tribunal, and its Section 38 repealed the National Environment Tribunal Act, 1995 together with the National Environment Appellate Authority Act, 1997, with a savings clause for anything done under the repealed Acts, of which, in the case of the 1995 Act, there was by definition nothing. The companion repeal tells its own contrasting story: the National Environment Appellate Authority, created in 1997 to hear appeals against environmental clearances, was at least constituted and functioned, within a remit so narrow that its existence barely registered, so that the tribunal-building programme of the 1990s produced one body that never sat and one that sat to little effect, and the comprehensive tribunal of 2010 inherited the ambitions of both. The non-commencement device deserves its place in the wider pattern of this series, because Bhopal’s tribunal is not an isolated instance of an Indian statute dying unnotified: the Whistle Blowers Protection Act, 2014 carries the same pattern into a different field, enacted amid public demand and then left uncommenced while amendments were contemplated. The mechanism is worth naming precisely. A commencement clause that leaves the appointed day to the executive converts enactment into an option, and the option, exercised silently by doing nothing, attracts none of the scrutiny that repeal in Parliament would attract. The tribunal statute is the purest case in the Bhopal sequence because the subject was compensation for the next disaster, and the cost of the option was borne, invisibly, by people who never knew the forum had been created for them.

The design that was lost can be stated concretely. The Tribunal was to be a national body sitting in benches, presided over by a chairperson drawn from the higher judiciary, with judicial members and technical members bringing the scientific competence that accident litigation demands and ordinary civil courts lack. Its jurisdiction was the mirror image of the Bhopal suit’s agonies: a claim for compensation where death, injury or damage to property or the environment resulted from an accident while handling a hazardous substance, decided on strict liability, so that the claimant’s case was the accident, the substance and the harm, and not a decade of discovery into the defendant’s state of mind and maintenance records. It could award compensation for damage to the environment itself and direct its deposit toward restoration, an element no Indian forum then possessed. Read beside the other instruments, the 1995 Act was the keystone: the umbrella Act regulated, the Factories chapter prevented, the insurance Act paid small sums fast, absolute liability supplied the principle, and the tribunal was to be the place where the principle was applied to full compensation by a specialist bench. Remove the keystone and the arch stands as a set of disconnected spans, which is a fair description of the law the victims of later industrial accidents actually found. The 2010 Act eventually supplied a tribunal of wider jurisdiction, and the compensation jurisdiction for accidents forms part of its work, but it arrived twenty-six years after the disaster and sixteen years after the Parliament that voted the 1995 Act had accepted the need. Legislative histories usually record delay in years of debate. This one’s decisive delay happened after the debate ended, in a notification that was never signed.

What the disaster produced

The sequence, compressed, is this. Each instrument answered a distinct failure, and the table records the answers together with the two facts that decide their weight: whether they were brought into force, and what they would do if the same night recurred.

Response	Year	What it required	In force?	In a comparable accident
Bhopal Gas Leak Disaster (Processing of Claims) Act	1985	Exclusive representation of all claimants by the Central Government, with a statutory scheme for registration, categorisation and payment of claims	Yes, from 20 February 1985	Nothing: the Act is specific to the Bhopal disaster and creates no general machinery
Environment (Protection) Act	1986	Omnibus central power to set standards, restrict siting, regulate hazardous substances and direct closure of offending industry	Yes, from 19 November 1986	Prevention, inspection and closure powers, but still no compensation remedy for victims
Factories (Amendment) Act, Chapter IVA	1987	Site appraisal, compulsory disclosure to workers and the public, occupier responsibility, health records, safety committees and a right to warn of imminent danger	Yes, from 1 December 1987	Duties owed before and during operation, enforceable by the inspectorate, breach punishable under factory law
Absolute liability doctrine, M. C. Mehta v Union of India	1986	Liability of a hazardous enterprise for accident harm without the exceptions of the English strict liability rule, damages correlated to the enterprise’s capacity	Yes, as judge-made law from 20 December 1986	A liability rule a court can apply to the enterprise, with no defence of act of God or third-party act
Public Liability Insurance Act	1991	Compulsory insurance and immediate relief on a no-fault basis at Schedule rates, backed by the Environmental Relief Fund	Yes, from 1 April 1991; the Relief Fund scheme followed in 2008	Immediate small payments without proof of fault, while fuller claims proceed under other law
National Environment Tribunal Act	1995	A specialist tribunal awarding compensation on strict liability for accidents involving hazardous substances	Never brought into force; repealed in 2010	Nothing: the forum was never constituted, and claims would go to the Green Tribunal or the ordinary courts under other law

The criminal case: restoration, reduction, conviction

The criminal proceedings ran on their own track, repeatedly bent by the civil case. A first information report was registered on 3 December 1984, the investigation passed to the Central Bureau of Investigation within days, and the charge sheet followed on 1 December 1987. Warren Anderson, the chairman of Union Carbide Corporation, was arrested in Bhopal shortly after the disaster, released on bail, and left India; he never returned, was declared an absconder, and the two foreign companies, Union Carbide Corporation and Union Carbide Eastern, likewise never stood trial. The February 1989 settlement quashed the criminal proceedings entirely, and their restoration by the review judgment of 3 October 1991 restarted a prosecution that had to be rebuilt around the charge. The Sessions Court framed charges including culpable homicide not amounting to murder under Section 304 Part II of the Indian Penal Code, which carries up to ten years, together with charges of voluntarily causing hurt and grievous hurt by dangerous means and mischief.

On 13 September 1996, a Division Bench of the Supreme Court, in Keshub Mahindra v State of Madhya Pradesh, reported at (1996) 6 SCC 129, intervened at the stage of charge. Examining the prosecution material, the Bench held that it did not make out a prima facie case under Section 304 Part II or under the hurt and mischief provisions, because those charges required a degree of knowledge or intention about the fatal outcome that the material, directed at failures of maintenance, staffing and safety systems, did not establish against the accused. It quashed those charges and directed that the trial proceed under Section 304A, causing death by a rash or negligent act, which carries a maximum of two years, with the allied endangerment provisions. The decision is the most resented single step in the entire sequence, and the resentment has a precise content: it converted the legal character of the disaster from homicide to negligence before a single witness had been examined, on the prosecution’s own threshold material, and it capped the punishment for the worst industrial disaster in history at the level the Penal Code sets for a fatal road accident. The Court’s answer, that charges must answer the evidence and not the enormity of the event, is orthodox criminal jurisprudence; the victims’ groups’ answer, that the material about a plant run with disabled safety systems and known hazards supported the knowledge the graver charge requires, was pressed again for fifteen years, in a review attempt dismissed in 1997, in an application to alter the charge during the trial itself, rejected in April 2010, and in a curative petition by the Central Bureau of Investigation against the 1996 judgment, which the Supreme Court dismissed on 11 May 2011, holding that the 1996 decision rested on the material then before the Court and furnished no ground for the curative remedy. At no point did the graver charge return.

The prosecution case that the reduced charge carried to trial is worth stating, because it explains both the conviction and the controversy. The Central Bureau of Investigation’s case rested on the documented condition of the plant’s safety systems on the night of the leak: the refrigeration unit for the methyl isocyanate storage had been shut down, the flare tower that should have burned off escaping gas was out of service, the vent gas scrubber was not in a condition to neutralise a release of that scale, staffing and supervision at the plant had been run down, and safety audits had recorded defects that were not remedied. On that material the prosecution said the disaster was the foreseeable product of an enterprise operated with its defences dismantled, and the magistrate’s conviction accepted the case in its negligence form. The distance between the two characterisations of the same facts, a plant stripped of its safeguards awaiting the accident its own audits predicted, and a rash or negligent act within Section 304A, is the distance the 1996 judgment created and the trial could not cross. The victims’ organisations’ critique of 1996 is at bottom a critique of that translation: that a knowledge-laden course of conduct, documented over years, was reclassified at the charge stage as inadvertence, and that the reclassification, once made by the Supreme Court on threshold material, governed everything that followed, including the two-year ceiling that made the 2010 verdict read to the public as an acquittal dressed as a conviction.

The trial, under the reduced charge, ended on 7 June 2010 in the court of the Chief Judicial Magistrate at Bhopal, twenty-six years after the leak and after 178 prosecution witnesses had been examined. Seven former officials of Union Carbide India Limited, including its chairman Keshub Mahindra, its managing director and its works and production managers, were convicted under Sections 304A, 336, 337 and 338 read with Section 35 of the Penal Code, and the company itself was convicted as an accused. An eighth official had died during the trial. Each convicted individual was sentenced to two years’ imprisonment, the maximum Section 304A allows, and fined Rs 1 lakh; the company was fined Rs 5 lakh. The sentences were immediately suspended on bail pending appeal. The honest summary of the criminal law’s answer to Bhopal is therefore tripartite: the corporation’s chairman in America was never tried; the Indian officials were convicted of negligence and received the highest sentence the negligence provision permits, which is two years; and the Supreme Court’s 1996 reading of the charge ensured that no court was ever asked to decide whether the disaster was a homicide. Each element is a matter of record; the evaluation of the whole is a judgement on which the named participants remain divided, with the victims’ organisations describing the outcome as a parody of justice and the defence maintaining throughout that a maintenance failure, however grave its consequences, is negligence in law and was prosecuted as such.

The Anderson absence deserves its own record, because it is the element of the criminal history most often misstated. Anderson was an accused, the first named in the case, and he was within Indian custody for a matter of hours in December 1984 before bail released him and he departed. He was declared a proclaimed offender when he failed to appear, the charges against him were separated, and the Indian government pursued his extradition from the United States through diplomatic and legal channels over many years; the requests did not succeed, and he died in the United States in 2014, an accused person who had never been tried. The State of Madhya Pradesh and the Central Bureau of Investigation carried appeals and revisions against the 2010 outcome to the sessions court, seeking enhancement of the sentences and revival of the graver charge; the sessions court declined to disturb the magistrate’s work, and no enhancement followed. The criminal record therefore closes in an unusual shape. The only completed trial in the disaster’s history was of the Indian subsidiary’s officers, under a negligence charge framed fourteen years after the event, producing sentences at the statutory maximum of a provision designed for individual rashness. The parent corporation, its chairman and its eastern affiliate, all named as accused, were never tried at all, protected first by a settlement term the courts later disowned and thereafter by the ordinary difficulties of extradition and corporate process. Responsibility, in the criminal law’s final accounting, stopped at the level of the plant’s Indian management, not because any court held that responsibility ended there, but because every route upward was closed by a decision, a bargain or a border.

Counting the dead: the figures and why they cannot be merged

Every number attached to Bhopal belongs to a set, and the sets answer different questions. The Government of Madhya Pradesh confirmed 3,787 deaths as related to the gas release; that is the state’s official confirmation figure, produced through its own recording processes. The Supreme Court, explaining the settlement in May 1989, worked on the assumption of about 3,000 deaths and about 105,000 injured; those are litigation assumptions, the pricing basis of the bargain, not findings of fact. The claims machinery, adjudicating over the following two decades, produced a third set: 10,29,517 claims filed, compensation awarded in 5,74,376 cases, of which 5,295 were death cases, with the balance spread across minor injury, temporary and permanent disability, loss of property and livestock, figures carried in official replies about the Welfare Commissioner’s work. Amnesty International’s account gives a fourth set, more than 7,000 deaths in the immediate aftermath and a further 15,000 in the years that followed from gas-related illness, an estimate built on the proposition that official recording missed the undocumented dead and the later mortality of the exposed. A single death toll for Bhopal cannot be honestly stated, because the four sets measure an official confirmation, a settlement assumption, an adjudication outcome and an epidemiological estimate, and they diverge because the underlying records diverge, not because one of them is a lie.

What can be stated is the consequence of the divergence for the money. The fund was sized on the second set and distributed on the third. The amount disbursed grew beyond the original fund, to approximately Rs 3,058 crore by July 2011 on the figures in official replies, because interest and exchange movements had enlarged the account, and because the Supreme Court in 2004 directed that the accumulated balance be distributed pro rata among the claimants already categorised, roughly doubling what each had first received. The Union’s curative petition of 2010 nonetheless contended that the settlement had been struck on a radically understated census of harm and that disbursement, however enlarged, divided a fixed bargain among five times the expected number of recognised victims. The victims’ organisations put the comparative point in its sharpest form: the average recognised claimant received, across both rounds, a sum in the region of a few hundred pounds sterling equivalent for injuries that in many cases were permanent. The company side answered, in the curative proceedings, that the settlement was a compromise of disputed claims at a figure between the parties’ own offers, that the Union had attested its sufficiency, and that a bargain cannot be repriced a generation later because one side’s estimates proved optimistic. Both accounts are internally consistent. They disagree about what a settlement is: the closing of a risk, or a provisional payment against a truth not yet known.

The composition of the adjudicated set, from the same official replies, is instructive about where the disaster’s harm actually fell in the machinery’s eyes. Of the 5,74,376 awards, the overwhelming majority, 5,27,894, were for minor injury; 35,455 were for temporary disability; 4,902 for permanent disability; 42 for injury of the utmost severity; 5,295 for death; and the remainder, a few hundred each, for loss of property and loss of livestock. The distribution is the quantitative form of the critics’ argument. A disaster remembered for its dead was processed, in the main, as a mass minor-injury event, because that is what the evidence available to the machinery could sustain at scale: hospital attendance and recorded exposure could be shown for lakhs, while the gravest harms, the deaths in unregistered households, the illnesses that declared themselves years later, pressed against the category boundaries and the evidential rules. The settlement’s assumption of about 105,000 injured had imagined a smaller, graver claimant population than the one that filed. What filed was a population five times larger and, in the machinery’s classification, mostly less gravely hurt, and the fund spread itself accordingly. Any single-sentence account of what the average Bhopal victim received conceals that distribution, and this article accordingly states the sets and their sources rather than an average.

The successor question and the abandoned site

Two questions outlived the compensation case, and both concern what followed the corporation rather than the disaster. Union Carbide India Limited did not remain Union Carbide’s. In 1994 the parent sold its 50.9 per cent holding in the Indian company, which was renamed Eveready Industries India Limited and continued in businesses unconnected with the Bhopal plant, whose lease the Madhya Pradesh government cancelled, taking possession of the site in 1998. In 2001 the Dow Chemical Company acquired Union Carbide Corporation itself, which continued as a wholly owned subsidiary. The dispute is easily stated and has never been resolved on its merits. Dow’s position is that it never owned, operated or managed the Bhopal plant, that it acquired the shares of a company whose Bhopal liabilities had been fully and finally settled in 1989, eleven years before the acquisition, and that a share purchase does not transfer to the purchaser a personal liability that the acquired company itself no longer bears. The position of the victims’ organisations, and the premise of the Union’s curative petition, which was directed against Union Carbide and its successor firms, is that corporate restructuring cannot be allowed to strand the obligations of a disaster, that Union Carbide Corporation remains a live accused and a live judgment-debtor in Indian proceedings, and that the company that owns it cannot treat the Bhopal chapter as closed while its subsidiary declines Indian process. The article records the dispute rather than resolving it; what is not contested is that no court has held Dow liable for Bhopal, and that Union Carbide Corporation, an absconder in the criminal case, has never submitted to trial in India.

What happened to Union Carbide’s Indian company after the disaster?

It was sold. Union Carbide Corporation disposed of its majority holding in Union Carbide India Limited in 1994, and the company was renamed Eveready Industries India Limited, continuing in batteries and other businesses. The Bhopal plant closed with the disaster; its lease was cancelled, and the site passed into government possession in 1998, its stored chemical waste still on it.

The site is the physical remainder of the legal history. Several hundred tonnes of toxic waste from the plant’s operations remained stored at the abandoned works, and campaigners and residents of the surrounding settlements have maintained, across the whole period, that contamination from the site has entered the local groundwater; the companies have disputed responsibility for a site they ceased to hold, successive governments have disputed the allocation of the cost, and remediation has been the subject of continuing proceedings and directions rather than completion. Residents of the settlements around the plant carried the contamination question to the United States as well, suing Union Carbide there for the cleanup of the site; those proceedings ended in dismissal, the federal appellate court holding that the evidence did not establish the parent company’s direct responsibility for the plant operations from which the contamination was said to come. Consistently with the discipline this article applies to contested technical questions, no finding on the extent or the health effect of any contamination is asserted here. What belongs to the legislative record is narrower and undisputed: the disaster produced no statute imposing a funded, enforceable duty of site remediation on the operator of a hazardous plant for the decades after closure, and the argument about who pays for Bhopal’s ground has therefore been conducted without a statutory answer, in the same improvised way as the argument about who paid for its victims.

The curative petition and the closing of the compensation question

The route back to the Supreme Court ran through the trial verdict. The public and political reaction to the two-year sentences of June 2010 was immediate, and the Union government constituted a Group of Ministers to examine the whole Bhopal question afresh, compensation, remediation, medical care and the criminal appeals together. The Group’s recommendations produced an enhanced ex gratia package, funded by the Union, for the worst-affected categories of victims, governmental commitments on the site’s cleanup, and the decision that produced the final legal act of the compensation history: the Union would ask the Supreme Court, in the rarest of its jurisdictions, to cure its own settlement. The curative petition of 2010 asked the Court to reopen the 1989 settlement and direct Union Carbide’s successor firms to pay an additional Rs 7,844 crore. The Union’s case was that the settlement had been computed on factual assumptions, about 3,000 deaths and about 105,000 injured, that the adjudicated reality had falsified, that environmental damage had never been priced into the bargain at all, and that a settlement procured on a radically false factual basis should not bind the victims it undercounted. The successor firms, represented in the proceedings, answered that a settlement can be reopened only on the ground of fraud, that no fraud was pleaded, that the Union had itself certified the fund’s sufficiency when claims were adjudicated, and that the depreciation of the rupee and the passage of time were not legal grounds for a top-up.

On 14 March 2023, a Constitution Bench of five judges, headed by Justice Sanjay Kishan Kaul and sitting with Justices Sanjiv Khanna, Abhay S. Oka, Vikram Nath and J. K. Maheshwari, dismissed the petition. The Bench held that a settlement stands unless fraud is shown, that the Union had furnished no rationale for raising the matter two decades after it had attested the fund’s adequacy, and that reopening the bargain would open a Pandora’s box detrimental to the claimants themselves. It directed that Rs 50 crore of the settlement money still lying with the Reserve Bank of India be used to satisfy pending claims. And it turned one part of the 1991 review judgment back on the Union: the review bench had directed the government to take out insurance for the victims against precisely this contingency of a shortfall, the direction had never been implemented, and the Bench described that failure as gross negligence on the Union’s part, holding that the responsibility for making good any deficiency lay where the settlement and the review judgment had placed it, on the Union itself. The judgment of March 2023 is the end of the compensation litigation that began in American courtrooms in 1984.

The hearing had exposed the strangest feature of the fund’s long life, the surplus paradox. Money from the settlement still lay undistributed with the Reserve Bank decades after the disaster, the residue of claims never pressed, claimants never traced and categories never exhausted, while the recognised victims maintained, with justice, that what they had received was inadequate to the harm. Both facts were true at once, and the Bench’s direction that Rs 50 crore of the residue be applied to pending claims was an administrative answer to an administrative residue, not a repricing of the disaster. The deeper exchange of the hearing ran the other way. The Bench pressed the Union on why, if the victims’ need was as the Union described it, the Union had not met the deficiency from its own resources in the decades since it had certified the fund sufficient, and on the unimplemented insurance direction of 1991, which had been designed for exactly the shortfall lamented in the Union’s petition. The curative petition thus ended by reallocating the grievance: the Court refused to move the settlement, and placed the moral residue of the undercount where the settlement’s own structure had always put it, on the parens patriae that had struck the bargain. For the legislative historian, that is the fitting close of the 1985 Act’s story. The statute made the state the victims’ only voice; the state’s voice, in 1989, said the number was enough; and when the state returned in 2010 to say the number had been wrong, the Court held it to the consequences of having spoken for everyone. The criminal appeals and the remediation proceedings continue on their own tracks, but the question whether Union Carbide or its successors will pay more for Bhopal has been answered, by the only court that could reopen it, in the negative.

What the Bhopal sequence shows about Indian legislation

The verdict this history supports is not that the legislative response to Bhopal failed. Four of its five instruments operated, and two of them, the umbrella Act and the hazardous processes chapter, restructured the everyday law of Indian industry in ways that have nothing to do with commemoration: standards, siting, disclosure, health surveillance and closure directions are the working law of every chemical plant in the country. The absolute liability doctrine, though born in another case, is the liability principle Indian courts reach for. The failure is specific, and it is located where this series has located it before, in the gap between enactment and commencement. The one instrument in the sequence designed exclusively for the victims of the next Bhopal, a specialist compensation tribunal applying strict liability, was the one instrument the executive never switched on, and it sat unnotified for fifteen years until a broader statute repealed it without debate about what had been lost. The notification gap is usually a story about rights delayed or regulators unborn. In this setting it is a story about a forum that did not exist when the accidents of the 1990s and 2000s happened, so that the victims of each of them began where Bhopal’s victims began, in the ordinary courts, under the ordinary law, waiting. The compensation tribunal was the one response that never arrived, and its absence, not the settlement’s arithmetic, is the legislative fact about Bhopal that generalises.

The successor forum has been tested, and the test is the fairest measure of what the 1995 Act’s absence cost. When styrene gas leaked from a polymer plant at Visakhapatnam in May 2020, killing and injuring residents of the surrounding villages, the National Green Tribunal took cognisance of the accident within days, applied the liability principles the Bhopal generation had articulated, and directed an interim deposit of Rs 50 crore toward compensation while the claims were examined. Whatever the later history of that case, the institutional reflex was the one the 1995 Parliament had intended and the uncommenced Act could not supply: a specialist forum, moving at the speed of the emergency, putting money on account before liability had finished being argued. Between Bhopal in 1984 and Visakhapatnam in 2020 lie thirty-six years in which that reflex existed, for the first fifteen, only in a statute that had never been switched on.

Two absences complete the audit. The sequence produced no reform of corporate criminal liability: the Penal Code the prosecutors inherited offered a choice between a homicide charge the evidence meetings of 1996 would not sustain and a negligence charge capped at two years, and no legislature, in the decades when Bhopal’s verdicts were front-page grievances, created the calibrated offence for industrial catastrophe, death caused by the operation of a hazardous enterprise in breach of its safety duties, that the gap invited. The criminal law of the next disaster remains the criminal law Bhopal found. And the sequence produced no remediation code: the duty to restore a poisoned site, funded in advance by the operator and enforceable after closure, exists in Indian law only in fragments of the umbrella Act’s powers and the Green Tribunal’s directions, which is why the Bhopal site’s cleanup has been an argument among governments and companies for a generation rather than an obligation with a debtor. A reader tracing the whole response should hold the shape of it clearly. Prevention was legislated thoroughly and early. Liability was articulated completely, by a court, in another case. Compensation was improvised, collectivised, settled, and defended with diminishing judicial enthusiasm until a Constitution Bench closed it in 2023 on the ground that settlements end. The forum built for the purpose was never opened. That is the legislative history of the Bhopal disaster, and its lesson for the next one is written in the order of the statutes themselves: India builds the regulatory answer first and the victim’s answer last, and once, in 1995, it built the victim’s answer and left it in the Gazette, unread by any bench, unrepealed for fifteen years, the most eloquent statute in the series, and the one whose empty commencement clause every student of this history should read first.

Frequently Asked Questions

Q: Which laws were enacted after the Bhopal disaster?

Four statutes and one doctrine make up the legislative response. The Bhopal Gas Leak Disaster (Processing of Claims) Act, 1985 gave the Central Government the exclusive right to represent every claimant and created the scheme for registering, categorising and paying claims. The Environment (Protection) Act, 1986 created the umbrella framework of central powers over standards, siting, hazardous substances and closure of industry. The Factories (Amendment) Act, 1987 inserted Chapter IVA on hazardous processes into the Factories Act, 1948. The Public Liability Insurance Act, 1991 imposed compulsory insurance and no-fault immediate relief for accidents involving hazardous substances. The National Environment Tribunal Act, 1995 created a compensation tribunal on paper but was never brought into force and was repealed in 2010. Alongside the statutes, the Supreme Court’s oleum judgment of 1986 produced the doctrine of absolute liability.

Q: What did the Bhopal Gas Leak Claims Act do?

The Bhopal Gas Leak Disaster (Processing of Claims) Act, 1985, Act 21 of 1985, transferred every victim’s claim to the state. Section 3 gave the Central Government the exclusive right to represent, and act in place of, every person entitled to make a claim, in India or abroad, including the power to sue, withdraw proceedings and enter into a compromise. Individual lawsuits by victims were displaced. Section 4 required the government to have regard to a claimant’s wishes and allowed a claimant to associate a personal legal practitioner at personal expense. Under Section 9 the government framed the Registration and Processing of Claims Scheme, 1985, which set up registration, categorisation of claims by class of harm, determination of compensation and appeals within the machinery, administered through the Welfare Commissioner and the claims courts. The Supreme Court upheld the Act in Charan Lal Sahu in 1989.

Q: What was the 1989 Bhopal settlement and why was it criticised?

On 14 February 1989 a five-judge bench of the Supreme Court recorded a settlement under which Union Carbide Corporation paid US$470 million in full settlement of all claims arising out of the disaster, all civil proceedings were concluded, and all criminal proceedings were quashed. The Court’s reasons, given on 4 May 1989, show the figure was computed on assumptions of about 3,000 deaths and about 105,000 injured. The criticism has three heads. The census of harm proved wrong: the claims machinery eventually awarded compensation in 5,295 death cases and 5,74,376 cases in all. The victims had no voice, because the 1985 Act had taken their claims and no hearing preceded the bargain. And the quashing of criminal prosecutions as a term of a civil settlement was widely condemned; the Supreme Court itself, in review in 1991, upheld the money but restored the criminal proceedings.

Q: What is absolute liability and where did it come from?

Absolute liability is the Indian doctrine that an enterprise engaged in a hazardous or inherently dangerous activity is liable for harm caused by an accident in that activity without any of the exceptions that qualify the English rule of strict liability, and without proof of negligence. It was laid down by a Constitution Bench of the Supreme Court in M. C. Mehta v Union of India, reported at (1987) 1 SCC 395, arising from the oleum gas leak at the Shriram plant in Delhi in December 1985, while the Bhopal proceedings were pending. Chief Justice Bhagwati’s judgment held that the enterprise owes an absolute and non-delegable duty to the community, and that the measure of compensation should be correlated to the magnitude and capacity of the enterprise. The doctrine was provoked by Bhopal but never applied to it, because the Bhopal claims were settled rather than adjudicated.

Q: What does the Public Liability Insurance Act require of hazardous industries?

The Public Liability Insurance Act, 1991 requires every owner who handles a hazardous substance above the quantities notified under the Act to take out insurance, before starting to handle it, against the liability the Act creates. That liability, under Section 3, is to pay the relief specified in the Schedule where death, injury or damage to property results from an accident involving the substance, and it arises on the principle of no fault: the claimant does not have to prove any wrongful act, neglect or default. The owner must also contribute to the Environmental Relief Fund a sum equal to the insurance premium. The Schedule relief is modest and fixed: Rs 25,000 for a death, medical expenses up to Rs 12,500, disability relief scaled to Rs 25,000 for total permanent disability, limited wage relief and up to Rs 6,000 for property damage. Claims are made to the district Collector, who inquires and awards.

Q: Why was the National Environment Tribunal Act never brought into force?

The honest answer is that the commencement notification was simply never issued, and the reasons are contested. The Act’s commencement depended on a notification by the Central Government under its appointed-day clause. Commentators have alleged pressure from industries handling hazardous substances, which faced a strict liability forum able to award compensation for environmental damage. The institutional record suggests a second cause: reform policy moved toward a broader environmental tribunal of general jurisdiction, a direction the Law Commission’s 186th Report of 2003 endorsed while recording that the 1995 Act had stood unnotified for eight years, and the narrower accident tribunal lost its sponsors. No government explained the omission in Parliament, because none had to: an unissued notification attracts no debate. The Act remained uncommenced for fifteen years until the National Green Tribunal Act, 2010 repealed it.

Q: What happened to the criminal case over Bhopal?

The criminal proceedings were quashed by the 1989 settlement, restored by the Supreme Court’s review judgment of 3 October 1991, and then reshaped by the decision of 13 September 1996 in Keshub Mahindra v State of Madhya Pradesh, which quashed the charge of culpable homicide not amounting to murder under Section 304 Part II and directed trial under Section 304A, causing death by a rash or negligent act. The trial ended on 7 June 2010 in the Chief Judicial Magistrate’s court at Bhopal: seven former officials of Union Carbide India Limited and the company were convicted under Sections 304A, 336, 337 and 338 of the Penal Code, each individual receiving two years’ imprisonment, the maximum the negligence provision allows, and a fine of Rs 1 lakh. Warren Anderson, the parent company’s chairman, was declared an absconder and never tried. A Central Bureau of Investigation curative petition against the 1996 reduction was dismissed in 2011.

Q: Did the Bhopal disaster change factory safety law?

Yes, directly and in terms. The Factories (Amendment) Act, 1987 inserted Chapter IVA, Sections 41A to 41H, into the Factories Act, 1948 with effect from 1 December 1987, creating the first statutory regime for hazardous processes. Section 41A requires site appraisal committees for the location or expansion of hazardous factories. Section 41B compels the occupier to disclose information about dangers and health hazards to workers, the Chief Inspector, the local authority and the general public in the vicinity, to publish a health and safety policy, and to maintain an on-site emergency plan. Section 41C fixes specific responsibility on the occupier, including health records and medical examination of workers. Section 41F prescribes permissible exposure limits, Section 41G requires safety committees with equal worker representation, and Section 41H gives workers a right to warn of imminent danger. Contravention of the disclosure, occupier-responsibility and warning provisions carries up to seven years’ imprisonment under Section 96A.

Q: What is forum non conveniens, and why did the American courts send the Bhopal claims to India?

Forum non conveniens is the doctrine under which a court that has jurisdiction over a case declines to hear it because another forum is clearly more appropriate for the trial. The Bhopal claims were first filed in the United States, where the parent company, its assets and its records were. Judge Keenan of the District Court for the Southern District of New York dismissed the consolidated action in 1986 on that ground: the plant, the evidence, the witnesses and the victims were in India, Union Carbide undertook to submit to Indian jurisdiction and to waive limitation defences, and the judge held that retaining the case would be a form of legal imperialism against the courts of a sovereign nation. The dismissal was substantially upheld on appeal in 1987. The Indian government had itself argued that its courts lacked the machinery for such litigation; the argument failed, and the Union then sued at Bhopal under the 1985 Claims Act.

Q: What did the Supreme Court hold in Charan Lal Sahu about the Bhopal Claims Act?

In Charan Lal Sahu v Union of India, decided in December 1989 and reported at (1990) 1 SCC 613, a Constitution Bench upheld the Bhopal Gas Leak Disaster (Processing of Claims) Act, 1985 against the challenge that it unconstitutionally extinguished the victims’ own right to sue. The Court held that the state may act as parens patriae, parent of the nation, for victims of a mass disaster who are no match, physically, mentally, financially or evidentially, for a multinational defendant, and that Parliament could therefore vest the exclusive conduct of the claims in the Central Government. The Bench also confronted the complaint that the victims had not been heard before the February 1989 settlement. It held that a settlement without notice to the victims was not quite proper, but that a post-decisional hearing would serve no useful purpose because the victims could show no further material capable of impeaching the settlement’s basis.

Q: How were individual Bhopal claims registered, categorised and paid under the 1985 Scheme?

The Bhopal Gas Leak Disaster (Registration and Processing of Claims) Scheme, 1985, framed under Section 9 of the Claims Act, built an administrative adjudication system under the Welfare Commissioner. A claimant registered a claim, supported by the available evidence of death, injury or loss, which in many cases meant hospital records, death certificates or the testimony of neighbours rather than documents. Deputy commissioners inquired into and categorised each claim into classes running from death and permanent disability through temporary disability and minor injury to loss of property and livestock, each class carrying its compensation range, and determined the amount payable. A claimant dissatisfied with the categorisation or the award could appeal within the machinery to the Additional Commissioner, with the claims courts handling the caseload. On the official figures, 10,29,517 claims were filed and compensation was awarded in 5,74,376 cases, including 5,295 death cases, the adjudication running through forty claims courts over some fourteen years.

Q: What interim compensation did the courts order Union Carbide to pay before the 1989 settlement?

The first money ordered in the litigation was interim, not final. In the Union of India’s damages suit at Bhopal, the District Judge, by an order dated 17 December 1987, directed Union Carbide Corporation to deposit Rs 350 crore as interim compensation for the victims, invoking the court’s inherent powers under the Code of Civil Procedure. Union Carbide challenged both the amount and the very permissibility of interim compensation in a suit where liability was denied. The Madhya Pradesh High Court, by its order of 4 April 1988, reduced the amount to Rs 250 crore while sustaining the power. Both sides appealed to the Supreme Court, the Union against the reduction and the company against the existence of the power. Those appeals were pending, with the interim question undecided, when the Supreme Court recorded the overall settlement of 14 February 1989, so the interim compensation question was never finally answered as doctrine.

Q: Why did the Supreme Court reduce the Bhopal charges to Section 304A in 1996?

In Keshub Mahindra v State of Madhya Pradesh, decided on 13 September 1996 and reported at (1996) 6 SCC 129, a Division Bench examined the prosecution material at the stage of charge. The Sessions Court had framed charges including culpable homicide not amounting to murder under Section 304 Part II of the Penal Code, which requires that the act be done with knowledge that it is likely to cause death. The Bench held that the material, directed at failures of maintenance, staffing, supervision and safety systems, did not make out the knowledge the graver charge required against the accused, and quashed the charges under Sections 304 Part II, 324, 326 and 429. It directed that the trial proceed under Section 304A, causing death by a rash or negligent act, read with the allied endangerment provisions. The decision has been criticised ever since for translating documented, long-running safety failures into negligence before any evidence was tested at trial.

Q: Did Warren Anderson ever stand trial in India for the Bhopal disaster?

No. Warren Anderson, chairman of Union Carbide Corporation, was the first-named accused in the criminal case. He arrived in Bhopal on 7 December 1984, was arrested, was released on bail the same day, and left India. He never returned. The Bhopal court declared him an absconder and later a proclaimed offender, and his case was separated from that of the accused who faced trial. The Indian government pursued his extradition from the United States over many years; the requests did not succeed, and extradition was never effected. Anderson died in the United States in 2014, an accused person who had never been tried. The two corporate accused, Union Carbide Corporation and Union Carbide Eastern, likewise never stood trial. The only completed trial was of the Indian subsidiary’s officials and the subsidiary company itself, ending in the convictions of 7 June 2010.

Q: Why did the Supreme Court dismiss the Centre’s curative petition for more Bhopal compensation in 2023?

The Union’s curative petition, filed in 2010, sought an additional Rs 7,844 crore from Union Carbide’s successor firms, arguing that the 1989 settlement had been computed on wrong assumptions about deaths and injuries and had never priced environmental damage. On 14 March 2023 a Constitution Bench of five judges, headed by Justice Sanjay Kishan Kaul, dismissed it. The Bench held that a settlement can be set aside only on the ground of fraud, that no fraud had been pleaded, that the Union had furnished no rationale for raising the matter two decades after attesting that the fund was sufficient, and that reopening would open a Pandora’s box detrimental to the claimants. It directed that Rs 50 crore of settlement money still lying with the Reserve Bank of India be used for pending claims, and it faulted the Union for never implementing the 1991 review judgment’s direction to take out insurance for the victims, holding the Union responsible for making good any deficiency.

Q: How is absolute liability different from the strict liability rule in Rylands v Fletcher?

The rule in Rylands v Fletcher, from 1868, makes a person who brings a dangerous thing onto land liable for its escape without proof of negligence, but the liability is qualified by exceptions: act of God, the act of a stranger, the plaintiff’s own default, consent and statutory authority. Absolute liability, laid down in M. C. Mehta v Union of India in 1986, removes the exceptions. An enterprise engaged in a hazardous or inherently dangerous activity is liable for accident harm simply because the harm resulted from the activity, and no act of God, no third party’s intervention and no statutory permission is a defence. The difference was not academic for Bhopal: Union Carbide’s own account blamed deliberate sabotage by a disgruntled employee, which under the English rule could have been run as a stranger’s-act defence through years of litigation. Absolute liability also adds a measure the English rule lacks: compensation correlated to the magnitude and capacity of the enterprise.

Q: Does Dow Chemical accept liability for Bhopal as Union Carbide’s owner?

No, and the question is contested rather than settled on any merits determination. The Dow Chemical Company acquired Union Carbide Corporation in 2001, and Union Carbide continues as its wholly owned subsidiary. Dow’s position is that it never owned, operated or managed the Bhopal plant, that Union Carbide’s Bhopal liabilities were fully and finally discharged by the 1989 settlement, eleven years before the acquisition, and that purchasing the shares of a company does not transfer to the purchaser a liability the acquired company itself no longer bears. The victims’ organisations answer that corporate succession cannot strand a disaster’s obligations, that Union Carbide Corporation remains an accused in the Indian criminal case and has never submitted to it, and that the company owning it cannot treat the matter as closed. The Union of India’s curative petition was directed against Union Carbide and its successor firms; its dismissal in March 2023 closed the compensation claim without any court holding Dow liable.

Q: What is the Environmental Relief Fund and how is it financed?

The Environmental Relief Fund is the fund provided for by Section 7A of the Public Liability Insurance Act, 1991. It stands behind the compulsory insurance the Act requires of every owner handling hazardous substances. Under the Public Liability Insurance Rules, every owner contributes to the fund a sum equal to the premium payable to the insurer, paid together with the premium and remitted through the insurer. When a Collector’s award of no-fault relief in an accident exceeds the amount payable under the insurance policy, the excess is met from the fund, and any remaining shortfall falls on the owner, so the claimant is not left to chase the gap. The fund socialises the risk across the whole hazardous sector, each owner’s premium-sized contribution standing behind every other owner’s accident. Although the statutory provision dates from the early 1990s, the Environment Relief Fund Scheme that made the fund operational was notified only on 4 November 2008.

Q: How long did the Bhopal claims process take from registration to final payment?

Far longer than the legislation that created it. The Claims Act was law within four months of the disaster, but the adjudication of claims under the 1985 Scheme occupied forty claims courts for some fourteen years, working through 10,29,517 filed claims, with the bulk of categorisation and award complete by about 2003 and compensation awarded in 5,74,376 cases. Disbursement ran behind adjudication, and the fund’s growth through interest and exchange movement produced a second round: in 2004 the Supreme Court directed that the accumulated balance be distributed pro rata among the claimants already categorised, roughly doubling the first payments, and official replies record total disbursement of approximately Rs 3,058 crore by July 2011. Money from the settlement was still being applied as late as 2023, when the Constitution Bench directed that Rs 50 crore lying with the Reserve Bank of India be used to satisfy pending claims. For an individual claimant, the interval between injury and completed payment was commonly measured in decades.

Q: Who is responsible for cleaning up the abandoned Bhopal factory site?

That question has never received a settled legal answer, which is itself the finding. The plant closed with the disaster. Union Carbide Corporation sold its majority holding in the Indian company in 1994, the company was renamed Eveready Industries India Limited, and the Madhya Pradesh government cancelled the plant’s lease and took possession of the site in 1998, with several hundred tonnes of stored toxic waste still on it. Campaigners and residents maintain that contamination from the site has entered local groundwater and continue to press for remediation at the companies’ expense. Union Carbide and Dow maintain that liability for the site does not rest with them, Dow pointing to the 1989 settlement and to its acquisition of Union Carbide only in 2001. Successive governments have disputed the allocation of cost, and remediation has proceeded through proceedings, directions and official committees rather than under any statute that assigns a funded duty of cleanup to the operator of a closed hazardous plant.