Article 31C's Ouster Clause Fell By Three Different Routes, Not One

The standard headnote summary of Kesavananda Bharati on Article 31C runs along these lines: the first part of Article 31C — the immunity of laws giving effect to Articles 39(b) and 39(c) from challenge under Articles 14, 19 and 31 — was upheld; the second part — the ouster clause that excluded judicial review of whether the law in fact gave effect to such policy — was struck down. The headnote is correct on the result, and the result is, in shorthand, an 8:5 holding against the ouster.¹ What the headnote does not say is that the eight came together for three different reasons. The reasons matter, because the consequence of compressing them into one ratio is that the protected status of judicial review — which everyone now reads Kesavananda as having settled — was not settled. Minerva Mills, seven years later, had to settle it; and the work it did was the work an honest reading of Kesavananda’s reasons should have done in 1973.

The three routes

The 13-judge bench in Kesavananda divided on the validity of Article 31C in two stages. The first stage was the question whether the first part — the immunity itself — was valid. On that question the result was 8:5 in favour of validity, with Sikri C.J., Shelat & Grover JJ., and Hegde & Mukherjea JJ. dissenting. The second stage was the question whether the second part — the ouster of judicial scrutiny on the question of whether the law in fact furthered Article 39(b) or (c) — was valid. On that question the result was 8:5 against validity, with the five dissenters on the first part joined by Jaganmohan Reddy, Khanna and Chandrachud JJ.²

That second 8:5 majority is the operative holding the standard summary captures. Within that eight, the reasons split three ways.

Route one: the whole clause is a delegation of constituent power (5 judges)

Sikri C.J. is the cleanest statement of the first route. His Part VI on Article 31C reads the impugned section as, in effect, a transfer of the amending power from Parliament to the State Legislatures. The reasoning runs as follows. Article 31C lifts the ban imposed by Articles 14, 19 and 31 in respect of any State law giving effect to Articles 39(b) or (c). The lifting is contingent — it operates only when the State legislates. The amendment, therefore, is not complete until the State law is passed. “The amendment is then made not by Parliament,” he writes, “as the extent of the amendment is not known till the State legislates.” In substance, the State law is what amends the Constitution. But the State Legislature is not the constituent body which Article 368 contemplates: Article 368 provides that an amendment must be initiated only by the introduction of a bill in either House of Parliament, and must be passed by a two-thirds majority of Parliament. To allow a State Legislature, by a simple majority, to do what Article 368 reserves to Parliament is to permit Parliament to delegate its constituent function. Parliament has no such delegation power. Therefore Article 31C, in toto, is void as a delegation of the amending power.³

Shelat & Grover JJ. and Hegde & Mukherjea JJ. join Sikri on the same reasoning.⁹ Each pair of opinions adds its own emphasis. Hegde & Mukherjea draw out the “totalitarianism” point — that Article 31C, considered as a whole, would permit a State Legislature to abrogate freedom of speech, freedom of assembly, equality before law, and the right to property at will, so long as the State could plausibly describe its law as serving Article 39(b) or (c). Shelat & Grover add the federalism point — that the relationship between Article 368 and the State Legislatures is fixed by Article 368 itself, which requires ratification by half the States only in specified cases, and that Article 31C circumvents the ratification requirement by making each State Legislature its own amending authority. The line is the same: the whole clause fails because the whole clause is a delegation, and the delegation is not Parliament’s to make.

What is important for our purposes is that, on this route, the ouster clause is not a separable problem. The ouster clause matters because it makes the State’s declaration self-validating, and that is what gives the delegation its operative effect — without the ouster, a Court could check whether the State law actually furthered the policy, and the delegation would be reined in by judicial nexus-review. But the delegation is the underlying vice, and the whole clause is its expression. Strike the delegation, and you strike both parts of 31C.

Route two: the declaration makes the legislature a judge in its own cause (2 judges)

Jaganmohan Reddy J. and Khanna J. take a different route. They accept the first part of Article 31C — the immunity itself — as a permissible exercise of the amending power. Their reasoning is that Parliament may, in principle, narrow the scope of Articles 14, 19 and 31 in respect of a specified category of legislation (laws giving effect to Articles 39(b) or (c)). That narrowing does not, by itself, damage the basic structure or framework of the Constitution; it simply reorders the relationship between Part III and Part IV in a defined area. The amending power, on Khanna’s reading, is plenary subject only to the limit of identity. Reordering Part III against Part IV in a defined area is well within the limit.

The second part is a different matter. The ouster operates this way: any law containing a declaration that it is for giving effect to Articles 39(b) or (c) shall not be called in question in any court on the ground that it does not give effect to such policy. The declaration is, in effect, a self-validating certificate. By means of the certificate the State Legislature both legislates and pronounces, conclusively, on the legislation’s constitutional fit. That, Khanna writes, has two consequences. The first is that, with the certificate, even a law that does not in fact further Article 39(b) or (c) is immune from challenge — and that converts the immunity from a narrow, defined immunity (laws actually furthering Article 39(b) or (c)) into an open immunity (laws declared to further Article 39(b) or (c)). The defined category is no longer a category. The second is that the Court is excluded even from the threshold question of whether a piece of legislation is within the class of laws that Article 31C protects. That exclusion strikes at basic structure.⁴

The point about basic structure here is doing narrower work than it is sometimes thought to do. Khanna is not saying that judicial review of all legislation is basic structure. He is saying that judicial review of whether a legislation is within the protected class under Article 31C cannot be excluded, because, if it is excluded, the State Legislature becomes the final authority on what Article 31C means — and that finality is itself an amendment to Article 31C, accomplished by the State, in contravention of Article 368. The basic-structure point is being made in service of the textual point about what “amendment” can mean. Reddy J. agrees with the result but reaches it by way of a narrower observation about severability — that the “conclusiveness” clause is severable from the substantive immunity, and that the substantive immunity stands without it.

So: on this route, only the ouster falls, and it falls because the self-validating declaration converts the State Legislature into a constituent body. The first part is fine because it is a clean substantive immunity that the Court can police on nexus.

Route three: the ouster is valid, but read down to preserve nexus review (1 judge)

Chandrachud J. completes the eight by a route that is, strictly, separate from the seven who held the second part invalid. His Conclusion 9 says that Section 3 of the 25th Amendment, which introduced Article 31C, is valid. But it adds: “in spite of the purported conclusiveness of the declaration therein mentioned, the Court has the power and the jurisdiction to ascertain whether the law is for giving effect to the policy of the State towards securing the principles specified in Article 39(b) or (c).” And then a sharper sentence: “If there is no direct and reasonable nexus between such a law and the provisions of Article 39(b) or (c), the law will not, as stated in Article 31C, receive immunity from a challenge under Articles 14, 19 or 31.”⁵

This is reading down. Chandrachud does not say the ouster clause is void. He says it is valid, and adds that despite the clause the Court retains the power to apply a nexus test. The reading-down has the same operational effect as a striking-down — the Court remains in the loop on the threshold question — but the doctrinal frame is different. On Chandrachud’s view the words “shall be called in question in any court on the ground that it does not give effect to such policy” are not stripped out of Article 31C; they are simply confined to a residual question (whether the policy, once nexus is shown, has in fact been advanced by the law) that no court was ever going to police. The substantive nexus question — whether there is a real link between the law and Article 39(b) or (c) — is, on Chandrachud’s reading, untouched by the ouster, because the ouster only operates after the threshold has been crossed.

The “View by the Majority” signed by nine judges, however, does not register Chandrachud’s reading-down. It says, flatly, that the second part of Article 31C is invalid.⁶ The signatory list includes Chandrachud J. He thus joined an order that struck down a clause he had, in his own opinion, upheld. The signatory list is a settlement document — what nine judges could agree to write down as the result — and it papers over a methodological disagreement that does not paper over comfortably.

What the three routes share and where they part company

All three routes share the operational conclusion that the State Legislature cannot, by means of a self-validating declaration, immunise its law from threshold judicial scrutiny. The routes part company on what that conclusion rests on. Route one rests it on a structural prohibition against Parliament delegating its constituent function. Route two rests it on what “amendment” can mean within Article 368 — combined with the consequence that judicial nexus-review of a State law claiming to act under Article 31C is itself part of the architecture that any valid amendment must preserve. Route three rests it on a reading of Article 31C that confines the ouster to a residual question and protects the threshold by interpretation rather than invalidation.

The three routes have different implications for cases that come next. On route one, any future amendment that delegates the amending power to a body other than Parliament — to a state government, to an executive authority, to a tribunal — is liable to fall whether or not it contains an ouster clause, because delegation is the vice. On route two, any future amendment that contains a self-validating declaration is liable to fall, because finality-by-declaration is the vice — but a substantive immunity without a declaration would survive. On route three, no amendment is in principle vulnerable; what is vulnerable is an overbroad reading of an ouster clause, which the Court can always avoid by reading the clause down. Different futures, different doctrines.

What the standard summary buries

The standard summary treats the second-part-invalid result as a ratio. A ratio is a reason a court has agreed on for a result. The eight judges who held the second part invalid did not agree on a reason. They agreed on a result. That distinction is not pedantic. It is the difference between a holding that a future Court can apply and a result that a future Court has to redo. The future Court — sitting as Minerva Mills in 1980 — had to redo it. The reason it had to redo it is that the question put to it was whether Section 4 of the 42nd Amendment, which had enlarged Article 31C from a narrow immunity for Articles 39(b) and (c) into a broad immunity for all directive principles, was valid.⁷ A Court applying route one would have answered the question by saying that the enlargement is a further extension of the delegation of constituent power and falls for the same reason. A Court applying route two would have answered the question by asking whether the enlargement caused the document to lose its identity. A Court applying route three would have asked whether the enlargement, properly read down, leaves the nexus test in place. The actual Minerva Mills court did none of these things. It held that the enlarged Article 31C upset the balance between Part III and Part IV, that the balance was itself part of the basic structure of the Constitution, and that any amendment which damaged the balance fell.

That holding is independent of all three routes from Kesavananda. It is, properly speaking, a fourth route — the basic-structure-as-balance route — that Kesavananda had not opened. Minerva Mills opened it. And it opened it because the three routes from Kesavananda on Article 31C had not given the 1980 Court a usable rule of decision. Each route was tied to a specific feature of the original 31C (the contingent State-trigger; the self-validating declaration; the breadth of the ouster), and the 42nd Amendment had altered the original 31C in ways that did not engage those features. The Court had to find another reason. The reason it found has become the dominant strand of basic-structure jurisprudence. Route one, route two and route three have receded.

Why the question of judicial review’s protected status is the one that paid the cost

The deepest cost of the underdetermination, however, is on the related question of whether judicial review is part of the basic structure. The standard reading of Kesavananda is that the case settles this question affirmatively — the ouster fell, judicial review was protected, the Court drew a line. But the question is not settled by route one, which treats the vice of Article 31C as delegation; not by route two, which treats the vice as the conversion of the State Legislature into a constituent body via the declaration; and only obliquely by route three, which treats nexus-review as immune to ouster by interpretation. None of the three routes endorses the proposition that judicial review, as such, is part of the basic structure of the Constitution. That proposition was first stated cleanly in Minerva Mills — not in Kesavananda. The earlier case had given the proposition the benefit of an 8:5 result. It had not given it the benefit of an 8:5 reason.

The unsettlement matters because it explains why the Court has had to keep coming back to the question. In Indira Nehru Gandhi (1975), the Court treated the entrenchment of judicial review as settled and applied it to strike down Article 329A. In Minerva Mills, it had to settle the question that Indira Nehru Gandhi had assumed. In L. Chandra Kumar (1997), the question returned for a third time — this time whether the exclusion of High Court Article 226 jurisdiction by the Administrative Tribunals Act was constitutional. In I.R. Coelho (2007), the question returned for a fourth time — whether laws placed in the Ninth Schedule after 1973 could be reviewed for basic-structure violations. Each iteration extends judicial review’s protected status to a new context. None of them rests cleanly on what Kesavananda decided about Article 31C, because what Kesavananda decided about Article 31C was, on the question of judicial review’s protected status, underdetermined.¹⁰

It would be tempting to read the underdetermination as a feature of the case rather than a defect. A 13-judge bench that has gathered for a constitutional moment is, on this view, a bench whose composite holdings cumulatively close more questions than any one judge’s reasoning could. The arc from Kesavananda through Indira Nehru Gandhi, Minerva Mills, L. Chandra Kumar, Coelho would then be the steady, lawful working-out of what the 13 had collectively meant. I find this reading unconvincing, because the steady working-out has not been steady. It has been jerky, often correcting earlier propositions, with each later bench writing as if the question were being settled for the first time. The jerks are the trace of the underdetermination.

Ambedkar called Article 32 the very soul of the Constitution and the very heart of it: “an article without which this Constitution would be a nullity.”⁸ The Constituent Assembly was thus on notice, in 1948, that judicial review of state action against Part III was the legitimating instrument of the entire fundamental-rights project. One would have expected the Court in 1973, faced with the most direct attempt that had yet been made to oust that review, to deliver a single reasoned answer on whether the review was constitutionally protected. The 13-judge bench did not. It delivered three answers and a signed order that the three could agree to. The unanimity of the order has, since, been read as the unanimity of the reasoning. It was not.

What this reading clarifies and what it leaves open

The point is not that the eight judges who held the second part of Article 31C invalid were wrong. Each of the three routes is defensible on its own terms. The point is that pretending the eight agreed on a reason has costs. It buries route one’s federalism point, which would have given the doctrine a non-substantive ground for striking delegations of the amending power — a ground that would have been useful in the Ninety-ninth Amendment case and would be useful in any future attempt to constitute alternative amending bodies. It buries route two’s textual point about what an “amendment” can mean. And it buries route three’s reading-down move, which is, in many ways, the most parsimonious solution and the one most consistent with the Court’s normal approach to ouster clauses.

What does the recovery of the three routes look like in practice? It looks like writing, in each future basic-structure case, which of the three routes is being relied on, and saying why. It looks like reading Article 31B, after Coelho, against route one — because the Ninth Schedule mechanism is structurally a delegation, and route one tells us why that is the vice. It looks like reading the various ouster clauses that have proliferated in modern Indian statutes against route three — because reading down ousters to preserve threshold review is the move route three teaches. And it looks like reading any future attempt to immunise state action by self-validating declaration against route two — because the conversion of a body into a final authority on its own competence is the vice route two identifies.

The eight judges agreed that the ouster fell. They agreed that judicial review survived the assault of the 25th Amendment. They did not agree on how. The Court has, since, used the agreement and forgotten the disagreement. Minerva Mills had to do the work the disagreement was hiding. There is more work of that kind ahead, and the way to be ready for it is to remember that the eight took three routes.

Get the week's biggest decisions, distilled.

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