Hart's rule of recognition doesn't travel to India

Every legal positivist worth reading is read in India sooner or later, and Herbert Hart is read sooner. His Concept of Law is on every jurisprudence syllabus in every law school in the country. His union of primary and secondary rules sits in the lecture notes of every student who will eventually appear for a judicial service examination. The rule of recognition is taught as if it were a settled instrument, as natural to legal thought as the parts of speech are to grammar. What is rarely asked, in those classrooms or in the literature beyond them, is whether the instrument actually fits the system the students will go on to work in. The Indian Supreme Court, since at least 1973, has been making rulings that a Hartian observer can describe but cannot fully account for. The basic-structure doctrine is the most visible of these. It is also the clearest sign that the Hartian framework, when imported wholesale, leaves something important about Indian constitutionalism unexplained.

The position of this essay is narrow. I am not arguing that Hart was wrong about English law, or that positivism as a general theory of law fails. I am arguing something more local: that the rule of recognition, as Hart described it in 1961, does not travel cleanly to a constitutional order whose Supreme Court treats the founding document as both source and contested object. India has built, over fifty-three years of basic-structure jurisprudence, a constitutional grammar that resists the Hartian description not at its edges but at its centre. The grammar can be re-described, with effort, in Hartian terms. But the re-description is so heavy that it ceases to be illuminating.

What Hart actually said

Hart's argument is best approached from the gap he was trying to close. Austin had said that law was the command of a sovereign habitually obeyed and habitually obeying no one. Hart showed, across the early chapters of Concept of Law, that this account could not explain power-conferring rules, could not explain the continuity of legislative authority across changes of personnel, could not even explain the fact that a statute binds the legislators who enacted it.¹ The remedy Hart proposed was a union of two kinds of rule: primary rules of obligation, which tell people what they must or must not do, and secondary rules, which are rules about the primary rules — rules that say how primary rules are made, changed, adjudicated, and identified.

The most important of the secondary rules is the rule of recognition. It is the rule by which officials of a legal system pick out which primary rules count as rules of that system. In a simple society, the rule of recognition may be no more than the acknowledgement that what is written on a particular stone tablet is law. In a developed legal system, it is more complex: a hierarchy of criteria, often including a written constitution, legislative enactment, judicial precedent, custom, with one criterion treated as supreme when the criteria conflict. The supremacy of statute over common law in England is, on Hart's reading, a feature of the English rule of recognition.

Two features of the rule of recognition are essential for the argument that follows. First, the rule of recognition is not itself valid or invalid. It is the test of validity for everything else in the system. To ask whether the rule of recognition is valid is, Hart says, like asking whether the standard metre bar in Paris is correctly one metre long. The question does not arise within the system; the rule is presupposed.² Second, the rule of recognition exists as a matter of fact — specifically, as the convergent practice of officials in identifying the law. The rule does not float somewhere above the practice. It is the practice. If officials change what they treat as law, the rule changes; there is no separate Platonic rule to which the practice could fail to conform.

Hart was alive to the possibility of a system, like the United States, where the supreme criterion identified by the rule of recognition is itself a constitution rather than a legislature.³ The American picture, as he describes it, is one in which the ultimate rule of recognition picks out the federal Constitution as the supreme criterion, and the criterion is supreme in the sense that statutes inconsistent with it are not law. The criterion is contained inside the document. The document does its work as supreme criterion and stops doing work beyond that. Nothing in the rule of recognition allows it to be policed against something the document does not name.

How the Indian Constitution behaves

The first thirty years of Indian constitutional adjudication can, with some loss of nuance, be told as a Hartian story. The Court read the document, identified by reference to the rule of recognition that the Constitution of India 1950 was the supreme criterion. Where the text was clear, the text governed. A.K. Gopalan v. State of Madras (1950) is the paradigm. The majority refused to read 'procedure established by law' in Article 21 to import substantive due process; refused to find an overlap between Articles 19, 21 and 22; and gave plain effect to a text the framers had deliberately written in narrower terms than the American Bill of Rights.⁹ This is textualism, and the textualism is comfortably Hartian. The Constitution is treated as the supreme criterion. Validity is tested by reference to the words.

Even the early amendment-power cases stay inside the Hartian frame. In Sankari Prasad v. Union of India (1951) and Sajjan Singh v. State of Rajasthan (1965), the Court asked whether Article 13(2) restricted the amending power in Article 368 and answered no, because the text of Article 368 was, in Patanjali Sastri J's phrase, 'perfectly general' and admitted no exception. Golak Nath v. State of Punjab (1967) reversed the answer but did not abandon the method. The judges in the majority and the minority both relied on textual readings; they disagreed about which reading the text supported. The supreme criterion remained internal to the document.

Kesavananda Bharati v. State of Kerala (1973) is where the picture changes.⁵ The thirteen-judge bench produced eleven opinions, the majority holding that Parliament's amending power, however broad, cannot alter, abrogate, or destroy the basic structure of the Constitution. The list of basic features was not closed. Federalism, secularism, the supremacy of the Constitution, the rule of law, judicial review, the parliamentary form of government, the separation of powers — different opinions named different items, and later benches added more. What unites the list is not its content but its constitutional location: nowhere in the text.

This is the move a Hartian observer cannot easily describe. The Constitution is still treated as the supreme criterion in the rule of recognition; constitutional amendments are still required to pass under Article 368; the procedural rule still says what it always said. But the Court has identified a further test that an amendment must pass to count as a valid amendment. The test refers to features the document does not enumerate. The features are derived, on different theories at different times, from the structure of the document, from its scheme, from its preambular commitments, from natural-law intimations, from what the framers must have intended even though they did not say. The features are policed by the Court. The amendment that violates them is struck down — not because it failed to satisfy any rule the document names, but because it failed to satisfy a rule the Court has read out of the document's shape.

Indira Nehru Gandhi v. Raj Narain (1975) gave the doctrine its first decisive use. The Court struck down clause (4) of Article 329A, inserted by the 39th Amendment during the Emergency to insulate the Prime Minister's election from judicial review, on the ground that it breached the basic structure.⁶ Minerva Mills Ltd. v. Union of India (1980) struck down clauses (4) and (5) of Article 368 itself, inserted by the 42nd Amendment, on the ground that the limited power to amend was a basic feature.⁷ The 42nd Amendment had said, in terms, that there was 'no limitation whatever' on Parliament's amending power. The Court replied that the limitation was a basic feature, and a basic feature cannot be amended away, even by an amendment that purports to remove the limitation. The reasoning is impossible to fit into a Hartian rule of recognition that picks out Article 368 as the supreme criterion. The criterion has been re-described, by judicial action, to include something the criterion does not contain.

The Hartian re-description, and why it does not work

A determined Hartian will reply that nothing in this story is impossible to describe in Hart's terms. The rule of recognition, the reply runs, simply is whatever criteria the officials of the system actually use to identify primary rules of the Indian legal order. If, as a matter of practice, the Indian Supreme Court tests amendments against basic-structure grounds, then the rule of recognition in India just is: 'enactment under Article 368, provided it does not violate the basic structure'. The basic-structure proviso has become part of the rule itself. The system has a more complicated rule of recognition than the framers wrote down, but it has a determinate rule, and Hart's framework accommodates it.

This reply is available but it is not satisfying, and the dissatisfaction is instructive. Three things go wrong.

The first is the indeterminacy of the content. Hart's rule of recognition, on his own account, exists as a matter of fact — the convergent practice of officials in identifying the law. The convergence is what makes the rule a rule. The basic structure has never converged in this sense. The list of basic features is open-ended; different judges name different items in different cases; the Court has never adopted a closed enumeration. I.R. Coelho v. State of Tamil Nadu (2007) extended basic-structure review to laws inserted into the Ninth Schedule.¹¹ The NJAC case in 2015 added the primacy of the Chief Justice in judicial appointments to the list.¹² A rule of recognition that grows by judicial pronouncement, with no convergent identifying criterion outside the conclusions of particular cases, is doing different work from the rule Hart described. It is functioning less as a test the document supplies than as an authorisation the Court has issued to itself.

The second is the direction of policing. Hart's rule of recognition is what officials use to identify the law. The basic-structure doctrine, by contrast, is what the Court uses to invalidate the law — including law produced by the rule's own supreme criterion, namely Article 368. The Court treats Article 368 as both a source of valid amendments and a procedure whose products are vulnerable to overturning on grounds the Article does not state. To say that the rule of recognition has simply been complicated is to elide the direction of force. The rule has not been complicated; the rule has been turned back on itself. The Court is using the supreme criterion to constrain the supreme criterion. Nothing in Hart prepares the reader for this manoeuvre.

The third is the role of the document itself. In a Hartian system, the document does its work as supreme criterion and stops doing work beyond that. The American constitutional order, on Hart's reading, treats the document this way; the courts identify statutes as invalid by reference to provisions of the document, not by reference to some further unwritten test the document is supposed to embody. The Indian constitutional order does not treat the document this way. The Court treats the document as containing two layers — the visible text and an invisible structure derived from but not reducible to the text — and tests provisions of the document against the second layer. The document is, in Indian constitutional practice, both rule and the thing the rule polices. A rule of recognition that does both is not a Hartian rule of recognition.

The Kelsenian alternative, and why it does not work either

If Hart cannot describe basic-structure jurisprudence, perhaps Kelsen can. Hans Kelsen's Pure Theory of Law gives a different account of the foundation of a legal system. The validity of any norm in the system, on Kelsen's account, derives from a higher norm; the chain ends at the Grundnorm, a presupposed basic norm that says we ought to behave as the historically first constitution prescribes.⁴ The basic norm is not posited by any human act; it is presupposed in legal thinking. It is the regress-stopper, the foundation of the system's normative order.

The Indian Supreme Court has sometimes been read as having identified the basic structure with something like a Kelsenian basic norm. The reading is not absurd. If the basic structure is the set of presupposed features without which the Constitution would not be the Constitution it is, then it functions like a Kelsenian regress-stopper — the norm beyond which the system cannot be amended without ceasing to be the same system. The constitutional identity language in IR Coelho and elsewhere lends itself to this reading.

Kelsen does better than Hart on parts of this terrain, but he fails on a different part. The Kelsenian basic norm is presupposed; it is not articulated by the courts of the system. Kelsen is explicit that the basic norm is not itself a positive norm; it does not appear in the document; it cannot be amended; it is a transcendental-logical presupposition of legal thinking. The Indian basic structure, by contrast, is not presupposed silently. It is articulated by the Court, expanded by the Court, applied by the Court to invalidate specific amendments. The basic structure is a positive doctrine — not a presupposition. To call it a Grundnorm flatters it and misdescribes it. A Grundnorm that the apex court keeps enlarging case by case is not a Grundnorm any Kelsenian would recognise.

Both Hart and Kelsen, that is, supply tools that catch part of the Indian phenomenon and miss the rest. Hart captures the surface — the Court treats Article 368 as the supreme procedural criterion. Kelsen captures the depth — the Court treats the Constitution as having a presupposed essence that cannot be amended away. Neither captures the manoeuvre that distinguishes Indian constitutionalism from both English and Continental positivism: that the supreme criterion of the system and the presupposed essence of the system are located in the same document, and that the second is policed against the first by the institution the first has constituted.

How the Court got there

The story is not the story of a single doctrinal mistake. It is the story of an interpretive evolution that Chintan Chandrachud, writing in the Oxford Handbook of the Indian Constitution, traces across three phases — textualism in the 1950s and 1960s, structuralism from Kesavananda Bharati through Maneka Gandhi v. Union of India (1978) and Minerva Mills, and what he calls panchayati eclecticism since the late 1990s.⁸ The structuralism is the part that matters here. The Court did not arrive at basic structure by accident or by a single leap; it arrived by gradually treating the Constitution as a structured organism whose parts illuminated each other, and whose scheme was as much a source of constitutional law as its individual provisions.

Maneka Gandhi v. Union of India (1978) is the case in which the structuralist method is articulated most clearly outside the basic-structure cases.¹⁰ Beg CJ rejected the clause-by-clause reading of Part III that Gopalan had imposed, holding that the fundamental rights are 'parts of an integrated scheme' whose 'waters must mix to constitute that grand flow of unimpeded and impartial justice'. The metaphor matters. The Court is no longer asking what each provision says; it is asking what the provisions, taken together, mean. The shift is from the document as a series of clauses to the document as a structure. Once that shift has been made, the move to basic-structure review of constitutional amendments becomes available. If the Court can read scheme into a fundamental rights chapter to override the framers' textual compartmentalisation, it can read structure into the whole document to constrain Parliament's textual amendment power.

The trajectory was not inevitable. The framers had inserted Article 368 into the document with the express purpose of giving Parliament the power to amend any provision by a two-thirds majority. Pratap Bhanu Mehta and others have argued that the framers wanted a constitution that could change with the country, and that the basic-structure doctrine is, on the framers' theory, an act of judicial usurpation. The point is not that this argument is right; it is that the doctrine can only be defended on a theory of constitutional interpretation that goes well beyond what the framers wrote, and that the theory is not the positivist theory either Hart or Kelsen described.

The honest case for the Hartian description

Before I close, let me put the strongest version of the contrary view — that the Hartian frame can in fact accommodate basic structure — back on the table. A reader inclined to defend Hart will say something like this: in any developed legal system, the rule of recognition is what the officials of the system actually use. Hart was clear that the rule exists only in the practice of officials. The practice in India includes basic-structure review. That is just to say that the Indian rule of recognition, as a matter of fact, is more elaborate than the rule of recognition Hart was describing for England. The framework does not break; it stretches to include what Indian officials actually do.

This reading is not wrong. It is what is available, on Hart's own terms, if you want to keep Hart. The reason it does not satisfy is methodological rather than logical. The whole point of describing a legal system through its rule of recognition is to identify, with some precision, what officials of the system are doing when they say a particular norm is or is not law. If the rule of recognition is whatever officials in fact do, the description loses traction. It catches everything because it picks out nothing. The basic-structure proviso is real; it is not arbitrary; it has internal logic that distinguishes valid from invalid amendments. But the logic is not Hartian. It is a logic of constitutional self-binding that Indian judges have developed and continue to develop. Calling that logic part of the rule of recognition is to use 'rule of recognition' to mean 'whatever Indian courts do', which is true but vacuous.

The honest reading is the more limited one. Hart's apparatus describes part of the Indian system well — the part where the Court treats the Constitution as supreme over statutes, and Article 368 as the procedural criterion for amendments. The apparatus does not describe the part where the Court treats Article 368 as itself constrained by something the document does not name. For that part, the apparatus is not adequate, and pretending otherwise does Hart no favours.

Where this leaves jurisprudence in Indian law schools

Two consequences follow, neither of them dramatic, both of them worth taking seriously.

The first is for legal education. Hart is taught in Indian law schools as if his framework needs no adjustment to make sense of Indian constitutional practice. The framework is presented, the doctrine is presented, and the relationship between the two is left to the student to work out — which usually means it is not worked out at all. A more honest pedagogy would teach Hart and then teach the gap. The gap is not a failure of Hart; it is the place where Indian constitutional theory has to do its own work. Indian jurisprudence has not done that work systematically. The legal-philosophy literature on basic structure is large, but most of it argues about whether the doctrine is constitutionally legitimate, not about whether the doctrine fits any standard positivist account of legal validity. The questions are different. A doctrine can be legitimate and still sit awkwardly in the available theoretical apparatus.

The second consequence is for the Court itself. The Court has, over five decades, built a constitutional architecture that no imported jurisprudence quite describes. The basic-structure doctrine is its most ambitious move and its most visible vulnerability. The vulnerability is not that the doctrine lacks textual warrant — many Indian constitutional doctrines lack textual warrant and survive. The vulnerability is that the doctrine has no theoretical account of itself that satisfies the standards of legal-philosophical rigour the Court invokes when convenient. Each new addition to the list of basic features — federalism in Bommai, free and fair elections in Indira Nehru Gandhi, the primacy of the Chief Justice in the NJAC case — is announced rather than derived. The doctrine grows the way common law grows, by accretion across cases. But it claims a status common law does not claim: the status of a constraint on the constituent power itself.

This is the place where Hart's framework, taken seriously, would force a question that Indian constitutional law has avoided. If the rule of recognition is, as Hart says, a matter of officials' convergent practice, and if the basic-structure doctrine has now become part of that practice, then the rule of recognition is being shaped by the Court's pronouncements rather than the other way around. The Court is no longer applying the rule; the Court is making it. Whether that is a tolerable arrangement is a question of constitutional theory that the Indian academy has not engaged with as directly as the practice now demands. The basic-structure doctrine has the virtue of having checked, on several occasions, attempts to dismantle features of the Constitution that most Indians would not want dismantled. It has the cost of having made the rule of recognition in Indian law a thing the Court controls, expands, and revises, with no settled criterion outside its own decisions for whether a particular expansion is valid.

Hart's apparatus does not condemn this arrangement; it cannot describe it. Kelsen's apparatus does not condemn it either; it misnames it. What Indian constitutional jurisprudence needs is an apparatus that is not imported. The work of building one has been left to the Court, which has done it case by case, doctrinally rather than philosophically. That work is incomplete. The basic-structure doctrine is the central piece of Indian constitutional theory that has not been theorised by Indian jurisprudence with the seriousness the doctrine itself demands. Until it is, the rule of recognition for the Indian legal system will be what the Court most recently said it is — which is not what Hart meant by a rule of recognition, and not what any positivist account of legal validity is meant to permit.

Get the week's biggest decisions, distilled.

A weekly digest for lawyers, students and the legally curious. Every Friday morning. Free.