Strict liability after Bhopal: what Rylands couldn't hold

The most-cited tort rule in modern Indian law is a sentence. Bhagwati C.J., writing for a five-judge bench in M.C. Mehta v UOI in late 1986, said that an enterprise carrying on a hazardous or inherently dangerous activity is absolutely liable for any harm its activity causes, and that this liability is not subject to any of the exceptions which operate vis-a-vis the rule in Rylands v Fletcher. That sentence has done a lot of work over the last forty years. It has been treated as the foundation of the polluter-pays principle, of the public-trust doctrine in environmental cases, of damages claims against chemical plants, of writ jurisdiction over Bichhri-style sludge, and of every routine appellate observation that India does not follow the English law of strict liability anymore. I am going to argue, in this essay, that the sentence is not a doctrine. It is a holding that has never been theorised, never been codified, and never been given a limiting principle that a lower court could apply without consulting the rhetorical context of M.C. Mehta itself.

What Bhagwati actually walked away from

To see what M.C. Mehta did, you have to see what Rylands v Fletcher actually was. The 1868 judgment was a careful piece of common-law work. Blackburn J, in the Court of Exchequer Chamber, put it this way: a person who for his own purposes brings on his land and keeps there anything likely to do mischief if it escapes, must keep it in at his peril, and is prima facie answerable for all the damage which is the natural consequence of its escape. The House of Lords added a qualification — the use of land must be non-natural — and bedded the rule down with a set of exceptions. Plaintiff's own default. Act of God. Consent of the plaintiff. Act of a third party. Statutory authority. Each exception was the product of a real case and reflected a real moral intuition about when a person who had built a reservoir, kept gas, or stored explosives could fairly be made to pay even though he had been careful.¹

The rule had three working parts: a dangerous thing, an escape, a non-natural use of land. Read v J Lyons, the shell-explosion case from 1947, made the second element bite hard — a worker injured on the defendant's own premises, by the defendant's own munitions, could not invoke the rule because nothing had escaped.³ Indian courts had absorbed all of this. T.C. Balakrishnan Menon v T.R. Subramanian, the Kerala High Court's 'Minnal Gundu' case, brought the rule to bear on a coconut-shell explosive that fired into a festival crowd, treated it as ultrahazardous, and made the duty non-delegable.⁷ The Privy Council in Madras Railway Co v Zamindar had earlier carved out the agricultural-tank exception for Indian conditions. By 1985 the Indian law of strict liability was a doctrine: a rule with named elements, named exceptions, and a body of cases that had worked through the edges.

Then came Bhopal, and a year after Bhopal, the oleum-gas leak from the Shriram Foods unit in Delhi. The Court was being asked to decide whether the rule in Rylands was adequate to a constitutional question — what the Indian state owed to citizens harmed by an industrial mass disaster — and Bhagwati C.J. answered the question by jettisoning Rylands. The new rule, he said, would be the rule of absolute liability. It would have no exceptions. It would not be encumbered by the 19th-century English law of land use. It would track the deterrent logic of large enterprises which, having undertaken hazardous activity for profit, must bear the social cost as an item of overheads.² The passage is fluent, brave, and unobjectionable as moral argument. It is also, viewed strictly as doctrine, a single paragraph long.

Why this matters now, not just then

It would be one thing if the M.C. Mehta sentence had been allowed to remain a moral peroration about the Bhopal moment. It is not. The Indian environmental regime, almost in its entirety, rides on it as if it were settled doctrine. Every PIL that asks the writ court to shut down a polluting unit, every appellate observation about polluter-pays, every High Court order requiring an industrial respondent to remediate soil or water, traces its ground of liability — when it bothers to articulate one — back to the absolute-liability passage in M.C. Mehta. The Supreme Court in Indian Council for Enviro-Legal Action (Bichhri) explicitly invoked it to hold the H-acid manufacturers liable for the contamination of the village's groundwater and to direct the Central Government to recover remediation costs from them.⁵ Vellore Citizens' Welfare Forum wove the principle into Article 21 jurisprudence, alongside the polluter-pays and precautionary principles, treating it as part of the constitutional minimum that the state owed citizens whose right to life was being eroded by environmental degradation.⁶

This is a real edifice. It is also an edifice resting on a foundation that has not been examined in forty years. No subsequent judgment has tried to specify what counts as a hazardous or inherently dangerous activity, in advance and with reasons, rather than as a post-hoc characterisation of the activity that has just gone wrong. No subsequent judgment has explained whether the absolute-liability rule applies to harm caused otherwise than by 'escape' — whether, for example, an injury to a worker inside a chemical plant invokes the same rule, or whether the Read v J Lyons requirement of escape survives in some form. No judgment has told us whether the absolute-liability rule applies to damage that is not physical — pure economic loss to neighbouring businesses, for example, when a refinery shuts down a port for a week. We have a rule that everyone cites and that nobody has limited.

The first leg: the divergence from Rylands was rhetorical, not doctrinal

Read carefully, the M.C. Mehta passage does not so much replace Rylands v Fletcher as repudiate it on grounds of vintage. Bhagwati's reasoning is that Rylands was evolved in the 19th century at a time when these developments of science and technology had not taken place; that India cannot hold its hands back; that we should evolve a new principle of liability which English courts have not done. Each of these is a true sentence. None of them is a doctrinal claim. They are claims about the source of authority — that the Court is free of Rylands — not claims about what the new rule is or what its outer limits are.

What follows the repudiation is the famous formulation: an enterprise engaged in a hazardous or inherently dangerous activity owes an absolute and non-delegable duty; if any harm results, the enterprise is absolutely liable; and the measure of compensation must be correlated to the magnitude and capacity of the enterprise so that it has a deterrent effect. Each phrase here is doing work. 'Hazardous or inherently dangerous activity' replaces 'non-natural use of land'. 'Absolute' replaces 'strict'. 'Magnitude and capacity of the enterprise' replaces 'compensation for actual harm suffered'. But none of these substitutions has been carried through into a body of working law. We have a substitution at the level of vocabulary; we do not yet have a substitution at the level of doctrine.

Compare what would have been required for a real replacement. A genuine doctrine would have specified, even if only by example, what kinds of activities count as hazardous — toxic-chemical manufacture, certainly, but radioactive material, biotech, long-haul fuel transport, mineral extraction, what about thermal power, what about a high-voltage substation? A doctrine would have explained why escape is no longer a necessary element — and what replaces it as the trigger of liability. A doctrine would have addressed whether the older Indian carve-outs (agricultural-tank water, common-benefit cases) survive. A doctrine would have given trial courts and tribunals a way to apply the rule without having to perform the rhetorical move that M.C. Mehta performed in the first place. None of that exists. We have one paragraph in M.C. Mehta, a few approving repetitions in Bichhri and Vellore, and an entire field of practice riding on the sentence's authority.

The second leg: a statutory vacuum the Court never asked Parliament to fill

The deeper failure is not the Court's. It is Parliament's. M.C. Mehta was 1986. The Bhopal settlement was 1989. The Public Liability Insurance Act came in 1991, and it is the only piece of post-Bhopal legislation that even tries to give absolute liability a statutory home.⁹ It does not, in fact, do that. The PLI Act creates a small insurance floor for accidents involving hazardous substances. It defines hazardous substance broadly enough to be useful and narrowly enough to be enforceable. It says the liability is on the principle of no-fault. But the relief it provides is on a tariff: a few lakhs per fatality, scaled compensation for various categories of injury. It is workmen's compensation for industrial-disaster victims. It is not the deterrent-quantum logic of M.C. Mehta.

The gap between the two regimes — the constitutional rule announced in M.C. Mehta and the statutory floor laid down in the PLI Act — has never been closed. The PLI Act's compensation scheme was not updated to match the magnitude-and-capacity yardstick the Court had announced four years earlier. The Bhopal settlement itself, when the Court finally approved it in Union Carbide Corporation v UOI, did not work out, in any extended way, whether the absolute-liability principle in M.C. Mehta applied to UCC or to UCIL.⁴ The five-judge bench that had given the world the absolute-liability sentence in 1986 sat in 1989 to approve a settlement of USD 470 million on the basis of an estimate of fatal cases at 3,000 and permanent-disability cases at 30,000 — figures that were later shown to be cosmically off — and the absolute-liability principle barely figures in the Court's reasoning. The settlement was a settlement: a compromise number reached on a balance of considerations that included the urgency of relief and the litigation cost of pressing the doctrinal claim through to a contested judgment.

The point is not that the Court should have refused the settlement. The point is that the absolute-liability principle, in the case to which it was most obviously applicable, was not used to do the work it was announced to do. The deterrent-quantum logic — that compensation must be correlated to the magnitude and capacity of the enterprise — would have generated a settlement number with one or two more zeroes attached. It generated nothing of the kind. The Court announced the principle in 1986 and approved a settlement in 1989 that was reached without the principle's discipline. That is not a doctrine in operation. That is a slogan that the Court was unwilling to enforce against the body that had benefited most clearly from its absence.

The case for the absolute-liability rule, addressed honestly

I want to set out, before going further, the strongest version of what an opponent of this essay would say. It runs roughly as follows.

Bhopal was the largest industrial mass-disaster in history. Three thousand dead the first night, six lakh injured, a multinational respondent that had structured its Indian subsidiary at arm's length, an American court that refused jurisdiction on grounds of forum non conveniens, an Indian Government that had to legislate to give itself the standing to sue. In that environment, the common-law architecture of fault-based liability is not just inadequate; it is morally indecent. The five exceptions to Rylands v Fletcher were designed for a Victorian world in which the largest industrial accident was a flooded coal mine. 

To apply that architecture, with its 'act of stranger' carve-out, to a multinational chemical plant whose Indian subsidiary was now claiming sabotage by a disgruntled employee — the very defence that Rylands would have permitted — was to give the entire post-industrial legal apparatus to the defendant. M.C. Mehta's absolute-liability rule was the only morally serious response. The rule's lack of an internal limiting principle was not an oversight; it was a feature. Doctrinal tidiness is a luxury an environmental regime designed for industrial-disaster scale cannot afford.

And the corollary: the absolute-liability rule has, on the whole, worked. It changed the negotiating posture of every industrial defendant in India. It gave PIL petitioners a coherent ground of liability that they could plead. It allowed the Court in Bichhri to do what it did. It supplied the moral architecture for the polluter-pays principle that Vellore imported into Article 21. The lack of statutory codification is not a defect of the rule but a defect of Parliament, which has had forty years to write a Hazardous Industries Liability Act and has not done so. The lack of a worked-out limiting principle is the price of having a rule that can be deployed against well-resourced defendants who would otherwise drag the proceedings out for a generation. The rule is, by design, an anti-litigation rule.

I take all of this seriously. Bhopal is not a hypothetical to me. The settlement number was scandalously low; the relief reached the victims slowly and partially; the criminal proceedings ended in a fine on Indian managers and the quiet exit of the multinational parent. Anyone who looks at that record and concludes that the law of torts had to be rewritten will be reaching a conclusion that is not foolish. The absolute-liability rule is the law's apology for what happened in Bhopal, and it is a serious apology. I do not want to argue otherwise.

Where the steel-manned defence falls short

What I want to argue is that the apology is not a doctrine, and that treating it as one has costs the defenders of the rule do not face up to.

The first cost is enforceability. A rule with no specified outer edges is a rule that any sufficiently determined defendant can litigate around. The PLI Act gives such a defendant a tariff to point to: my liability is what the Act says it is. The constitutional rule in M.C. Mehta, lacking a statutory body, gives the defendant a different opening: the rule is rhetorical, the application is discretionary, the quantum is at the writ court's mercy, the remedy is not res judicata if the same plant is sued in a fresh forum, and the calculation of magnitude-and-capacity is essentially political. 

Indian environmental defendants have learned, slowly, to live with this. They settle on the tariff floor where they can, and they litigate the constitutional residue when they cannot. The result is a regime in which absolute liability is, in operational terms, the price of a few lakhs per fatality plus the negotiating cost of a writ proceeding. That is not what Bhagwati C.J. announced.

The second cost is principled. A rule announced in constitutional rhetoric, never carried into statutory form, never bounded by a worked-out limiting principle, becomes available for uses its authors did not contemplate. S. Vedant Acharya v Highways Department is a useful counterpoint here.⁸ A culvert collapsed in a flood after fifteen days of rain, a bus plunged in, a passenger died. The Highways Department pleaded act of God; the Supreme Court reversed the High Court and held that foreseeable rain and flood were not act of God, that the Department had a duty to anticipate, and that the failure to take preventive measures was negligence. That is a fault-based finding. It is also, in its own quiet way, doing the same moral work that the absolute-liability rule does in Bhopal — making a powerful institution answer for damage caused by foreseeable risks it failed to provide for. We have a fault-based law of negligence that is perfectly capable of holding institutional defendants to account where the activity is foreseeable-dangerous rather than ultrahazardous. We do not need to drag M.C. Mehta into every claim against a state highway authority or a public utility to make a serious case.

The third cost is the most important, and it touches the constitutional architecture. Vellore read absolute liability into Article 21, alongside the polluter-pays and precautionary principles, as part of the constitutional minimum the state owes citizens. That is a major move. It converts a tort rule announced in a writ proceeding into a fundamental-rights claim. It allows the writ court to adjudicate, under Article 32 or Article 226, the same questions that would otherwise be civil suits requiring proof of causation, damages, and the elements of the cause of action. The procedural protections that go with civil litigation — pleadings, discovery, oral evidence, the rules of evidence themselves — drop away. We are running an environmental enforcement regime, in respect of large industrial defendants, on a constitutional fast-track that was designed for emergency vindication of fundamental rights, not for routine compensation of pollution victims. The fast-track is fast because the doctrinal limits are loose. The doctrinal limits are loose because the rule was never theorised in the first place.

What an honest reading of M.C. Mehta would require

An honest reading of M.C. Mehta would do three things.

It would treat the absolute-liability rule as a constitutional command to Parliament, not as a free-standing tort doctrine. The Court in 1986 was telling the political branches that the existing law was inadequate. That command has been ignored. It needs to be reissued, more sharply, as the source of an obligation to legislate — a Hazardous Industries Liability Act with named industries, scheduled categories of harm, statutory minima for compensation that track the deterrent-quantum logic Bhagwati C.J. announced, and a procedure for invoking the rule that does not depend on the writ jurisdiction. The PLI Act is the start of such a regime, not a substitute for it.

It would limit the use of the absolute-liability rule, in the meantime, to the kind of industrial-disaster cases M.C. Mehta itself contemplated — escape of toxic substance from a chemical, biological, or radiological facility, on a scale that produces multiple-fatality harm. 

Smaller cases — a single worker injured on a plant site, a single neighbour whose property is damaged by a routine industrial discharge, a public-utility failure of the Vedant Acharya kind — should be litigated under the law of negligence and the law of nuisance, where the doctrines have edges and the trial courts know how to apply them.

It would acknowledge that the absolute-liability rule, as currently used, is a constitutional shortcut. The shortcut was justified at the moment it was announced. It has not been justified since. Forty years is long enough for a rule of liability to have either been theorised into a working doctrine or codified into a working statute. M.C. Mehta has been neither. The result is that the rule does what it does because the writ court chooses to apply it, not because a body of law tells the court to. The standard tort treatments make this point in their own quiet way: they still treat M.C. Mehta as a special-application case, an Indian extension of the Rylands tradition rather than a stable replacement for it.¹⁰ A doctrine that the leading textbooks cannot place is a doctrine that has not yet finished being written.

The textbook ambiguity matters because trial courts in India read textbooks more reliably than they read the Supreme Court Reports. A District Judge hearing a tort claim against a chemical plant in a tier-three town is going to reach for Ratanlal & Dhirajlal before reaching for the SCC Online printout of M.C. Mehta; if the textbook frames the rule as an Indian extension of Rylands rather than as a free-standing doctrine, that is the framing the District Judge will apply. The bench in Delhi can announce as much absolute liability as it likes; what reaches the trial court is the version the textbook has been able to render with sufficient confidence to print. Forty years on, that version is still 'an Indian variant of strict liability without the exceptions' — which is to say, it is still defined by what it is not, rather than by what it is.

None of this should be read as nostalgia for Rylands v Fletcher. The English rule is a Victorian artefact and the criticism of it that runs through Bhagwati's judgment is fair. Indian conditions, Indian industrial scale, Indian regulatory capacity were and are different. The right response was not to keep Rylands; the right response would have been to write a new rule with the same care that Rylands itself was written with. Rylands took fifty years of subsequent litigation to acquire the body of edges it has — the elements, the exceptions, the carve-outs for common-benefit and statutory authority — and it acquired them through the patient case-by-case work of trial and appellate courts. M.C. Mehta has had forty years and acquired none of them. The Court announced the rule and walked away from the writing. The position I want to land is plain enough. The absolute-liability rule was the right answer to the wrong question. It answered the moral question of what Bhopal demanded; it did not answer the doctrinal question of what should replace Rylands in Indian tort law. Treating one answer as if it were the other is the mistake the last forty years of environmental jurisprudence has been built on, and it is the mistake the next forty years should not repeat. The rule is necessary, in the narrow band of cases for which it was announced. The rule has been borrowed for far more work than it can carry, and the borrowing has gone unchecked because the alternative — a worked-out statute, a properly-articulated doctrine — has felt like more work than the writ-court rhetoric is worth. It is not. The work is the doctrine. There is no other way to have one.

I disagree with the standard reading of M.C. Mehta v Union of India. The standard reading treats the absolute-liability passage as a doctrinal achievement on a par with Kesavananda or Maneka Gandhi. I think that reading flatters the judgment. M.C. Mehta is a moral statement about Bhopal, dressed in tort vocabulary, that has been doing the work of a doctrine without ever being one. The body of cases that has grown around it — Bichhri, Vellore, the polluter-pays jurisprudence — is at its best a careful application of the moral statement and at its worst a free-floating writ-court power to do whatever the bench thinks the polluter should do.

The honest position is to say that the absolute-liability rule, as currently constituted, is a placeholder for a statute that Parliament has refused to write and a doctrine that the Supreme Court has refused to develop. It has bought Indian environmental enforcement forty years of authority on the back of a single paragraph. The bill for that comes due whenever a defendant with the means and the patience decides to litigate the rule's edges, and there is no body of law to point to in reply. We have been lucky that the bill has not yet been called in by a defendant willing to spend a decade in the Supreme Court contesting whether the rule even applies to its activity. The next defendant may be less obliging, and when that happens the absence of doctrinal scaffolding will be felt as something other than a forgiving silence.

The remedy is not to return to Rylands. The remedy is to write what was never written: the limiting principle, the statutory regime, the body of trial-level precedent that gives the rule edges a court can hold to. Until that work is done, the absolute-liability rule will continue to do what it does — and the next industrial mass-disaster in India will be litigated, like Bhopal was, on the strength of a sentence rather than the strength of a doctrine.