Khanna Did Not Sign the Basic Structure Theory You Think He Signed

There is a story the legal profession in India tells itself about the seventh vote in Kesavananda Bharati. The story is that thirteen judges sat, six held that Parliament could amend any provision of the Constitution, six held that there were inherent limits on Article 368, and Justice H.R. Khanna swung the result by joining the second six. The seven-judge majority then gave us the basic-structure doctrine. This is the story repeated in undergraduate textbooks, in the Supreme Court’s own later judgments, and in the headnote anyone consults when they want a summary of the case. The story is wrong on a small but consequential point. Khanna J. did not join the second six. He arrived at the same destination by a different road, and he was emphatic that the road mattered. The doctrine we live under has inherited the destination and lost the road.¹

I want to argue something specific. The basic-structure majority in Kesavananda contained two distinct theories of why Parliament cannot use Article 368 to abrogate the Constitution. The first theory — held by Sikri C.J., Shelat, Hegde, Grover, Mukherjea and Jaganmohan Reddy JJ. — was that there are inherent or implied limits on Article 368 itself, flowing from the structure of the Constitution and from extra-textual values that pre-exist the amending power. The second theory — Khanna’s alone among the majority — was that there are no inherent limits on Article 368 at all; what there is, instead, is a limit on what the word “amendment” in Article 368 can be taken to mean. These two theories produce the same operative holding in 1973. They produce different doctrines downstream. The doctrine we have built since 1980 — in Minerva Mills, in L. Chandra Kumar, and at full extension in I.R. Coelho — is the doctrine the first six judges would have written. It is not the doctrine Khanna wrote. And without Khanna there was no seven.

The seam in the seventh opinion

Read Khanna’s opinion alongside the doctrine map of the thirteen-judge bench and the seam shows in two places. The first is his treatment of implied limitations on the amending power, which the judgment frames as Issue 7 of the case: are there inherent or implied limits on Article 368, in addition to whatever flows from the article itself? Sikri C.J., Shelat, Grover, Hegde and Mukherjea JJ. answered yes. Jaganmohan Reddy J. left the question open. Khanna answered no, joining Ray, Palekar, Mathew, Beg, Dwivedi and Chandrachud JJ. on that issue.⁴ Of the thirteen judges, only Khanna sits on the no-side of Issue 7 and the yes-side of Issue 2 (does Article 368 enable Parliament to alter the basic structure?). He is the sole judge whose two answers, if read together, look contradictory — and they are not contradictory only because of what he meant by Issue 2.

His Conclusion (ix), stated in his own summary at the end of the opinion, is the cleanest statement of the seam: “There are no implied or inherent limitations on the power of amendment apart from those which inhere and are implicit in the word ‘amendment’.”² That sentence is, on its face, an answer to two questions at once. To the first — are there implied or inherent limits, of the kind Conrad and Wynes had argued for, of the kind Palkhivala had pressed in oral argument by reference to natural rights and the Preamble and the structure of the document? The answer is no. To the second — is the word “amendment” in Article 368 unbounded? The answer is also no, but the limit comes from somewhere else: from what “amendment” is, as a matter of legal grammar. Khanna spends seven pages of his opinion drawing the line, and the line is textual, not structural.

His most-quoted passage is the textual move stated bluntly: “I am further of the opinion that amendment of the constitution necessarily contemplates that the constitution has not to be abrogated but only changes have to be made in it. The word ‘amendment’ postulates that the old constitution survives without loss of its identity despite the change and continues even though it has been subjected to alterations.”³ The argument that follows is short. To amend something is by definition not to replace it; if at the end of the process the document has lost its identity, then what has happened is not an amendment but a substitution; Article 368 confers a power of amendment, not a power of substitution; therefore Article 368 cannot do what is not amendment. The argument runs entirely on the verb. It does not need a theory of inherent limits, or a hierarchy of constitutional values, or natural rights. Khanna says so explicitly, and at length.¹¹

The constitutional-history reader will already be a step ahead. Khosla, writing in the Oxford Handbook of the Indian Constitution, registers the distinction with care: six of the seven majority judges found inherent or implied limits in the nature of Article 368, while Khanna “focused instead upon the internal character of an amendment.”⁵ The distinction is in the scholarship. It is not in the case law. The doctrinal life of Kesavananda has, on the question of methodology, run almost as if Khanna had agreed with the other six. He had not. The order signed by the nine majority judges does not register the disagreement either; it states the operative holding — “Article 368 does not enable Parliament to alter the basic structure or framework of the Constitution” — in language broad enough to be signed by both camps, because the nine signatories could not have agreed on which camp the holding belonged to.⁸

Why the distinction is not pedantry

It is tempting to treat the difference between the two theories as a verbal one. If both theories produce the same operative holding — Parliament cannot abrogate the basic structure under Article 368 — why does it matter which route you take? It matters because the two theories generate different daughter doctrines.

An inherent-limits theory tells you that there is a substantive ceiling on amendment, the ceiling being defined by values external to the text — democracy, federalism, secularism, judicial review, the rule of law — each of which is part of the basic structure in its own right and any of which can be invoked to strike down an amendment that damages it. The ceiling has no closing list because the structure has no closing list. Each later Court is at liberty to add to the basic structure when a new challenge arises, because what the doctrine protects is not a finite set of textual provisions but a set of organising principles which the Constitution embodies and which can be added to as the Court recognises them. This is the doctrine that produced the addition of secularism to basic structure in S.R. Bommai, of judicial independence in the Second Judges’ case, of free and fair elections in Kihoto Hollohan, of the rule of law in Sambamurthy. The doctrine does not have an endpoint because nothing in its structure tells it where to stop.

A textual-meaning theory of the Khanna kind tells you something narrower. It tells you that there is a limit on Article 368 because of what the word “amendment” in Article 368 can be taken to mean. The limit is interpretive, not substantive. What constitutes a change of identity is, on Khanna’s reasoning, a question of comparing the document before and after. The role of “basic structure” in that comparison is to identify what would have to remain for the document to be the same document. It is the residual a competent reader would identify after stripping away every feature of the Constitution that is not constitutive of its being this Constitution. Khanna gives examples — turning democracy into dictatorship, abolishing the Lok Sabha, removing secularism — and each is an example of a change so radical that the new document would not be a version of the old document at all. The theory does not invite the Court to keep adding to a list of basic features. It invites the Court to ask, of a specific amendment, whether the document at the end of the amendment is recognisably the same document.

These two theories give different answers in cases that arrived after 1973. They give different answers, in particular, to three questions: whether the basic-structure limit applies to ordinary legislation as well as constitutional amendments; whether judicial review of constitutional amendments is itself part of basic structure; and whether a constitutional amendment that does not damage the document’s identity can nonetheless be struck down because it damages something — the rule of law in a specific tribunal, say — that the Court has come to regard as basic. On all three questions the inherent-limits theory leans one way and the Khanna theory leans the other.

I want to put the methodological difference in a sentence the reader can carry. The other six judges of the majority would have asked, of a challenged amendment: does this damage one of the values that the Constitution rests on? Khanna would have asked: would the document at the end of this amendment still be this document? Notice what the first question lets the Court do that the second does not. The first question requires the Court to maintain, and add to, a list of values. The second question requires the Court to decide one identity question per amendment. The first generates a doctrine that grows; the second generates a doctrine that resolves. Over fifty years the doctrine has grown, and it has grown because the methodology being used was the first one, not Khanna’s second one.

What Minerva Mills did with the wrong half

The cleanest demonstration of the substitution is Minerva Mills Ltd. v. Union of India, decided in 1980 by a five-judge bench led by Chandrachud C.J. and including Bhagwati J.⁶ Two provisions of the 42nd Amendment were under challenge. Section 4 had enlarged Article 31C to immunise all laws giving effect to any directive principle from challenge under Articles 14 and 19. Section 55 had inserted clauses (4) and (5) into Article 368: clause (4) ousted judicial review of constitutional amendments, and clause (5) declared that there were no limits whatever on Parliament’s power to amend the Constitution under Article 368. The Court struck both down. The reasoning by which it struck down Section 55 is what I want to read with care.

Chandrachud C.J. wrote that “the donee of a limited power cannot by the exercise of that power convert the limited power into an unlimited one.” That is the core sentence. It is an elegant sentence and it does important work, but it works only if you have already accepted that the power under Article 368 is a limited power before clause (5) tried to make it unlimited. That premise is precisely what was contested in Kesavananda. Six of the seven majority judges had a theory of why the power was limited that did not depend on the word “amendment” at all — it depended on implied or inherent limits. Khanna had a theory of why the power was limited that did depend on the word “amendment.” Minerva Mills takes the limit for granted and does not say which theory it is invoking.

That looks like a gap, and the gap matters because clause (5) of the 42nd Amendment was an attempt to legislate the basic-structure doctrine out of existence by changing the text. If the limit on Article 368 is textual, of the kind Khanna proposed, then a textual amendment that declares “there shall be no limit” is the most natural thing to test the doctrine against. The textualist would have to ask: does adding the words “there shall be no limit” cause the Constitution to lose its identity? The answer is not obvious, and a defender of Khanna’s reasoning would have to work hard to say yes. If, however, the limit is inherent — if it pre-exists the text and constrains it from outside — then clause (5) is the easy case. An attempt by Parliament to remove an inherent limit fails for the same reason that any attempt by an organ to enlarge its own constituent authority fails: the limit is not at its disposal.

Minerva Mills takes the easier road. It writes the limit as inherent. It then adds that judicial review of amendments is itself basic structure, on the reasoning that without judicial review the basic-structure limit would have no enforcer. This second move is the snake eating its tail — a basic-structure doctrine concludes that judicial review is basic structure because without it the basic-structure doctrine would have no force — and it is a move that Khanna would not have made. The judicial-review point was simply not the issue Khanna had been deciding. His issue was what Parliament can call an amendment. The two are not the same.

One can see Khanna’s embarrassment with the trajectory in his own later opinions. In Indira Nehru Gandhi v. Raj Narain in 1975, asked to apply the basic-structure doctrine to strike down Article 329A — the amendment that retrospectively immunised Indira Gandhi’s election — he wrote that Article 329A was “subversive of the principle of free and fair election in a democracy.”⁹ That is the language of basic-feature reasoning, not of textual-identity reasoning. He had not retracted his Kesavananda methodology. But he was applying a doctrine that had already begun to outrun his methodology, because the bench in Indira Nehru Gandhi was not a thirteen-judge bench rehearsing the methodological choices; it was a five-judge bench treating Kesavananda as settled. The settled version was the inherent-limits version. Khanna voted with the result and did not write a separate concurrence to retain his methodology. After 1975 the textual reading went underground; it has not surfaced since.

Coelho, and the migration of the doctrine to ordinary statutes

The migration matters at the second move too. By 2007, in I.R. Coelho v. State of Tamil Nadu, the Court was applying basic-structure review to ordinary statutes that had been inserted into the Ninth Schedule by constitutional amendments after 24 April 1973.⁷ The reasoning is that since the Ninth Schedule itself is amended by a constitutional amendment, the insertion is an amendment for Article 368 purposes; since amendments are subject to basic structure; therefore the insertion of any particular statute can be reviewed for whether it damages basic structure. Coelho is sometimes read as a clarification of Waman Rao; I read it as a quiet expansion. The clarification, on the textual-meaning theory, would have been: the insertion of a statute into the Ninth Schedule, considered as an amendment of the Schedule, cannot cause the Constitution to lose its identity. That is a much narrower test than “the statute itself, as it operates on individuals, cannot violate basic features of the Constitution.” Coelho writes the second sentence and treats it as if it followed from the first.

The two are not the same. On Khanna’s textual theory, the question for a Ninth-Schedule insertion is whether the Schedule’s being enlarged by this entry destroys the identity of the document. That is almost always no. On the inherent-limits theory — which is what Coelho is doing — the question is whether the substantive law thus immunised damages a basic feature. That can be yes whenever the law has any rights-restricting content, because basic features now include rights as such after Minerva Mills. The doctrinal centre of gravity moves from the amendment to the statute. Khanna’s theory could not get to Coelho. The inherent-limits theory got there in two moves.

The counter-argument, stated honestly

The strongest defence of how the doctrine has been built is that the distinction I am pressing is too clever by half. On this view, the textual route and the inherent-limits route are not competing theories; they are different vocabularies for the same insight. The insight is that there is a structural floor below which constitutional amendment becomes constitutional replacement, and any decent constitutional order must locate the line somewhere. Whether one locates it by reading the word “amendment” closely, or by positing inherent limits, or by appealing to the unamendable presuppositions of a constitution — the work being done is the same. To insist that Khanna’s route be honoured separately is to make a fetish of textualism in a field where the text was not going to do the work anyway. And to suggest that Minerva Mills and Coelho would have come out differently on a textual-meaning theory is to write an alternative history that no court would have followed, because the institutional pressures that produced Minerva Mills — a 42nd Amendment that was, in substance, an attempt to make Parliament unaccountable for the rest of constitutional history — would have forced the Court to the same destination by whatever route was available.

I take the force of that. The 42nd Amendment was, in fact, an attempt to make Parliament unaccountable, and a Court that failed to push back on it in 1980 would have been a Court that had failed at its function. The textual theory does not give the Court a comfortable seat from which to push back. It gives the Court a hard sentence to write — the kind of sentence that says, with a straight face, that “there shall be no limit on the power to amend” is itself outside the meaning of “amendment.” Khanna himself, asked the question, may well have written that sentence.

But the counter-argument proves too much. If the textual route and the inherent-limits route are interchangeable, then Khanna’s separate insistence on writing his opinion the way he did was decorative — a stylistic preference, not a methodological commitment. That reading does not fit the opinion. Khanna spends pages distinguishing his position from the implied-limits position. He explicitly rejects natural-rights reasoning. He rejects the idea that the Preamble functions as a source of implied limits. He treats the Conrad thesis with respect but reads Conrad narrowly, for the proposition that an amending body cannot, by structure, alter the basis on which it is constituted — which is a textual point about what amendment is, not a substantive point about what the Constitution requires.¹⁰ He is not being a stylist. He is being a textualist about constituent power, and he is doing so because he believes the implied-limits theory is unworkable. “The concept has no definite contours,” he writes; its acceptance would “introduce elements of uncertainty and vagueness in a matter of so vital an importance as that pertaining to the amendment of the Constitution.”¹¹ He had read what the implied-limits theory would do and he did not want to sign it.

The 31C tell

If anyone doubts that Khanna was holding two theories apart and not merging them, look at how he handled the second part of Article 31C — the ouster clause that immunised laws giving effect to Articles 39(b) and 39(c) from any judicial challenge premised on Articles 14, 19 or 31. He struck it down. He gave two grounds. The first was that the second part gave a State Legislature — by means of a self-validating declaration — the power to amend the Constitution in important respects, which is a function reserved to Parliament under Article 368. The second was that the exclusion of judicial review struck at the basic structure of the Constitution.¹²

Read these grounds in order. The first is the textual move — the State, through a declaration, is doing what only an Article 368 amendment can do, and that is impermissible because the second part of 31C makes the declaration itself the basis of immunity. The work is still being done by what “amendment” can mean, applied here to ordinary law that is functioning as a constitutional amendment in effect. The second ground — about judicial review and basic structure — sits within the textual frame because what is being challenged is, in Khanna’s reading, a back-door amendment. The exclusion of judicial review damages basic structure not because basic structure is a freestanding constraint on legislation but because, where the legislation is operating as an amendment, the amendment cannot be one that destroys identity — and identity, for Khanna, includes the availability of judicial scrutiny.

What this shows is that Khanna had a coherent position. He was not, on Issue 2, smuggling inherent limits in through the side door. He was working out, case by case, what the textual limit on “amendment” demands. The position is workable. It would produce a narrower doctrine than the one we have. It would also be a more defensible one, because it would tether each basic-structure finding to a specific argument about identity, rather than to a freestanding catalogue of values which the Court keeps the keys to.

What I want to leave the reader with

The point of this essay is not to suggest that Minerva Mills or Coelho were wrongly decided. They may have been right on the facts before them. The point is narrower. The doctrine that has built up around basic structure since 1980 is the doctrine that the six implied-limits judges in Kesavananda would have written. It is not the doctrine the seventh — the seventh whose vote made the majority — actually wrote. Khanna wrote something quieter and harder: a doctrine that lives in the verb “amendment” and asks, of every challenged amendment, whether the document at the end is the same document. That doctrine has been folded into the textbook account of Kesavananda and folded out of the Court’s self-understanding of what it inherited. When the Court talks today about basic structure as a constraint that pre-exists the text and constrains it from outside, it is not channelling the majority. It is channelling six judges out of thirteen. The seventh, who held the doctrine together, would have had reservations.

None of this is a constitutional emergency. The Court is not going to dig up Khanna’s opinion and re-found basic-structure doctrine on its textual base. The point is the more modest one that we have been talking about Kesavananda as if it produced one theory, when it produced two, and the one we built on was not the one with which the majority would have agreed if asked to vote. What we have is a doctrine with no agreed methodology because the methodology was never decided. That ought to make us a little less casual about saying, of any new constitutional question, that it is settled by basic structure. The doctrine is what the Court of the day says it is. Khanna would have asked it to say less.

Get the week's biggest decisions, distilled.

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