Sealed Covers: A Procedure Without a Source

There is a brown manila envelope, sealed with red tape and a wax stamp, that has appeared in some of the most important Indian constitutional litigations of the last decade. It contains documents the bench has asked to see and the petitioner has not been allowed to read. The bench reads what is inside, decides the case in part on the strength of what it has read, and returns the envelope to the registry. The lawyer for the side against whom the documents tell never learns what was in them. This artefact is called, in the practice that has grown up around it, a 'sealed cover'. It is now used routinely. It has been used in the Rafale review, in the Bhima Koregaon proceedings, in the Media One licence case, in INX Media, in cases involving electoral bonds, in cases involving the Adani-Hindenburg investigation, in cases involving extra-judicial killings in Manipur, and in routine bail matters of which no record exists outside the cause list. I want to argue, in this essay, that the sealed cover has no source in Indian procedure. Not in the Constitution. Not in the Indian Evidence Act 1872 or the Bharatiya Sakshya Adhiniyam 2023 that replaced it. Not in the Supreme Court Rules. Not in the writ-jurisdiction doctrine the Court itself has built. The practice should be retired - not because Parliament will or should pass a statute against it, but because no source ever authorised it in the first place.

Where the writ jurisdiction came from, and what it permits

To see the absence cleanly, you have to start from the textual foundation. Articles 32 and 226 of the Constitution confer on the Supreme Court and the High Courts the power to issue writs for the enforcement of fundamental rights and, in the case of Article 226, for 'any other purpose'.⁴ Neither clause confers a power to receive evidence ex parte. Neither contains the phrase 'sealed cover' or any cognate construction. Gopal Subramanium's chapter on writs and remedies in the Oxford handbook traces the doctrinal architecture of these provisions in detail: the discretionary nature of the writ; the requirement that it lie only where there is a pre-existing legal right; the standing rules; the procedural aspects, including res judicata, treatment of disputed questions of fact, and the limits imposed by the political-question doctrine.⁷ The chapter is forty pages long and covers, with the deliberation of a constitutional treatise, every meaningful procedural feature of Articles 32 and 226. The sealed cover is not mentioned. It is not mentioned because it does not exist as a doctrine within the framework of the writ jurisdiction. It is a practice that has accreted alongside that framework, without ever being placed inside it.

The framers of the Constitution treated the writ remedies as the operating heart of fundamental-rights protection. Ambedkar called Article 32 'the very soul of the Constitution and the very heart of it'. Gajendragadkar J, in an early formulation, called the right to move the Court under Article 32 'the corner-stone of the democratic edifice raised by the Constitution'. These descriptions were not poetic. They were technical: a guarantee without a remedy is not a guarantee, and the writ jurisdiction was the structural device by which Part III rights were to be made real. The procedural discipline of that jurisdiction - the requirement of pleadings, the requirement of evidence on the record, the requirement that parties have the opportunity to meet the case against them - was integral to the design. Writing into that jurisdiction a category of evidence the affected party may not see is not a procedural refinement. It is a procedural rupture.

What the sealed cover is, in operational terms

Let me describe the practice with as little editorial colour as possible, because the practice is doing all of the persuasive work in its own defence.

The Union of India, or sometimes a state authority, files an affidavit in a writ proceeding. The affidavit refers to a set of underlying documents - intelligence inputs, file notings, investigation reports, security assessments, internal advisories - which the executive contends ought not to be disclosed to the petitioner or to the public. The executive does not formally claim privilege under Section 123 of the Indian Evidence Act 1872, now Section 133 of the Bharatiya Sakshya Adhiniyam 2023, which is the statutory route for withholding evidence from production on grounds relating to affairs of State.¹³ A claim of privilege would have to be made formally, would have to be supported by an affidavit from the head of department, and would be subject to judicial determination on the test laid down in cases beginning with State of Punjab v Sodhi Sukhdev Singh and refined through SP Gupta and others. The privilege route has procedural costs: it produces a record, it is reviewable, and a finding against the State on a privilege claim is appealable. The sealed cover bypasses all of that. The executive simply hands the documents to the bench in an envelope. The bench, exercising what no one has yet identified as a particular statutory or constitutional power, accepts them. The opposing party is told that the bench has the materials and that the bench will form its view in the light of them. The materials themselves remain undisclosed.

The structural asymmetry is what the practice has come to depend on. The State controls the underlying record. The State frames the affidavit. The State decides what goes into the envelope. The State is in the room, through its counsel, when the bench reads the contents. The petitioner is informed only that something exists, and that the bench has read it. The hearing, which constitutional and administrative-law doctrine has treated for more than fifty years as the structural protection against arbitrary state power, becomes a hearing on one side and a silence on the other.

There is a smaller, less remarked feature of the practice that deserves to be named. The materials, once received, are returned to the registry. They do not become part of the public record of the proceeding. A reader of the judgment three years later, an academic five years later, a lawyer in an unrelated matter ten years later who needs to understand what the Court actually decided - none of them can see what the bench saw. The reasoning becomes opaque not only to the affected party but also to the institutional memory of the Court itself. Precedent, in the systematic sense in which the writ jurisdiction depends on precedent, becomes impossible to construct out of judgments that have read materials no one else can read. That is a second-order cost the practice incurs alongside the first-order denial of hearing.

Where the procedural source would have to come from, and why it isn't there

If the sealed cover were a legitimate device of Indian procedural law, you would expect to find a textual or doctrinal source. Three possible sources are usually canvassed, and each fails for a different reason.

The first possible source is the Evidence Act, now the Bharatiya Sakshya Adhiniyam. The Act and its successor describe documentary evidence in considerable detail. Documentary evidence means and includes 'all documents, including electronic records, produced for the inspection of the court'. The provisions on proof require primary evidence, that is the original document produced before the court, except in defined circumstances where secondary evidence is admitted.⁵ The classification of public documents, the rules on certified copies, the rules on the production of registered instruments, and the rules on attestation of wills and other attested documents all operate within a single architecture: a document is produced, the parties have access to it, and questions of execution, authenticity, and content are litigated on the basis of disclosed material. The Act recognises a single mechanism for withholding evidence from disclosure - the State-privilege regime under Section 123. Outside that regime, the Act has no doctrine that contemplates the bench reading a document the opposing party does not see. The procedural device that the sealed-cover practice depends on is, in plain words, not contained in the law that governs documentary evidence in India.

The second possible source is the Supreme Court Rules 2013, which govern the procedure of the Court itself. Order XIII of the Rules deals with filing and service of documents.¹² The Rules contemplate that documents will be filed, served, inspected, copied, and brought on record by the parties. There is provision for confidential treatment in some narrow contexts - for example, where the identity of a minor is involved - but no general rule authorising the reception of substantive evidentiary material from one party to the exclusion of the other. The Rules, in other words, do not authorise the sealed cover any more than the Evidence Act does. The Court is filing material outside its own rule-book.

The third possible source is Article 142, the 'complete justice' provision. This is the source the practice has tacitly leaned on, although no judgment has explicitly grounded the sealed cover in Article 142. Raeesa Vakil's chapter on jurisdiction in the Oxford handbook traces, with sceptical care, the Court's habit of treating Article 142 as a residual source for any procedural device it finds expedient: curative petitions, continuing mandamus, monitored investigations, court-evolved guidelines that operate as quasi-legislation.⁸ The sealed cover, on the view that the practice is constitutionally licit, would be one more instance of Article 142's productive capacity. The view is wrong for two reasons. The first is that Article 142 cannot, on its own terms, override express constitutional commitments - including the fair-procedure commitment of Article 21 as constructed in Maneka Gandhi.⁶ The second is that Article 142 produces no procedural innovation here. It cannot, because the device the practice introduces - reception of evidence from one party that the other party may not see - is not 'complete justice' in any sense the Constitution recognises. It is the structural opposite of complete justice. Calling it complete justice is calling a thing by its opposite name in the hope that the opposite name will stick.

Three uses, three failures

It is one thing to argue from sources. It is another to look at the cases. I want to take three uses of the sealed cover in the post-2018 period - Rafale, Bhima Koregaon, Media One - and show what the practice actually does to the writ jurisdiction it operates inside.

In Manohar Lal Sharma v Union of India (Rafale, 2018-19), the Court was asked to consider whether to direct an investigation into procurement irregularities in the Rafale fighter-jet contract. The Union filed an affidavit and a 'note on procurement, pricing and offsets' in sealed cover.¹ The Court read the note. The petitioners did not. The judgment dismissing the writ petition - and later the review - referred to the materials in paraphrased form, treating their existence and contents as established without ever placing them on the record. When errors of fact in the original judgment came to light, the review proceeded on the same procedural footing as the original: the materials remained in the envelope. The petitioners had to argue against a record they were not allowed to see, which is the operational definition of a Kafkaesque hearing.

In Romila Thapar v Union of India (Bhima Koregaon, 2018), the Court was asked to consider whether the arrests of five activists under the Unlawful Activities Prevention Act warranted independent investigation.² The Maharashtra Police produced case-diary materials in sealed cover. The majority, having read the materials, declined to interfere. Chandrachud J, dissenting, observed that the materials had been deployed in a manner that did not satisfy the requirements of disclosure that the writ jurisdiction demands. The dissent was overlooked at the time. It has aged well. The petitioners in Bhima Koregaon, several of whom remain in pre-trial detention years later, have never been told what the sealed cover contained that the majority found sufficient.

In Madhyamam Broadcasting Ltd v Union of India (Media One, 2023), a Kerala-based television channel challenged the Centre's decision not to renew its licence on national-security grounds.³ The Union produced its security assessment in sealed cover. The Kerala High Court accepted the assessment, ruled for the Union, and the channel went to the Supreme Court. A three-judge bench led by Chandrachud CJ set the orders aside. The judgment is the most careful articulation we have, to date, of what is wrong with the sealed-cover practice. It held that the practice violates the principle of natural justice and the right to a fair hearing; that resort to it must be tested by a structured proportionality enquiry; and that, ordinarily, the Court should appoint an amicus to review the materials and produce a public summary, rather than admit the materials directly into the bench's reasoning. The Media One judgment is, on its terms, the right answer. The trouble is that nothing in it commits the Court to abandoning the practice in the cases the bench finds inconvenient to abandon it. The judgment is a doctrinal limit. It is not an abolition.

The counter-argument, in its strongest form

The defence of the sealed-cover practice is not foolish, and it deserves to be stated at its strongest before being answered.

The defence runs roughly as follows. The Court is, in some matters, faced with information that genuinely cannot be disclosed without endangering national security, the lives of intelligence sources, the integrity of ongoing investigations, or the safety of witnesses in serious organised-crime cases. The privilege regime under Section 123 of the Evidence Act is cumbersome, requires multiple affidavits, produces a public record of the existence of the materials, and may itself reveal information that the State has a legitimate interest in not revealing. The sealed cover is, on this view, a pragmatic compromise: the bench is enabled to satisfy itself that the executive's claim has a factual basis without the State being forced into a full privilege procedure that could itself compromise the underlying interest. The opposing party is not, the argument continues, denied a hearing in any meaningful sense. They are denied access to specific sub-categories of material whose disclosure would defeat the protective purpose. The hearing on the legal questions - the constitutional and statutory issues - proceeds normally.

This is a serious argument. It is the argument the Solicitor-General has made, in different forms, in most of the cases that have used the device. It is the argument that animates the executive's reluctance to abandon the practice and the bench's reluctance to refuse it. I find the argument wrong, but I do not find it ridiculous.

The argument is wrong in three ways. The first is that it assumes the privilege regime is unworkable, when in fact the privilege regime exists precisely to handle the problem the argument describes. The Section 123 procedure - now Section 133 in the BSA - has been litigated and refined for over half a century. It allows for in-camera inspection. It allows for in-camera examination of the head of department. It allows for limited disclosure, summarised disclosure, redacted disclosure. It produces a reasoned order on whether the privilege is upheld and, if so, to what extent. The order is appealable. It is, in short, the system the law actually built for this problem. The sealed cover is not a more elegant version of that system. It is a way of avoiding it.

The second is that the argument proves more than its defenders mean it to prove. If the bench can receive evidence in sealed cover in a writ matter on national-security grounds, it can receive evidence in sealed cover in a criminal trial on national-security grounds, and then on organised-crime grounds, and then on protecting-the-witness grounds, and the principle that the accused has a right to know the case against him erodes one judgment at a time. The Court has already begun to drift in that direction. In INX Media, the bench accepted prosecution material in sealed cover at the bail stage and relied on it in declining bail.¹¹ The principle in P Gopalakrishnan v State of Kerala - that an accused must be supplied with materials accompanying the chargesheet, including electronic records - cuts squarely against this drift.⁹ If the sealed cover survives as a writ device, it will not stay confined to the writ side.

The third is that the argument assumes a benign equilibrium that the historical record will not bear. The Court has, on more than one occasion, accepted in sealed cover materials that turned out, when later disclosed, to be inaccurate or misleading - mis-stated values, mis-stated procurement processes, mis-stated facts about the alleged conduct of accused persons. The opportunity to test the materials at the time would have caught those errors. The sealed cover removed that opportunity. The State's reliability is not the test. The procedural discipline of the writ jurisdiction is the test, and the sealed cover bypasses it.

A fourth point is worth registering separately, because the defenders rarely answer it. The privilege regime applies to the State. The sealed-cover practice, as the Court has used it, has applied to non-State parties too - to corporate respondents in regulatory matters, to investigative bodies in PIL proceedings, even, occasionally, to petitioners producing material they prefer not to have on the public record. The justification offered for State sealed covers - national security, the protection of intelligence sources - cannot, on any honest reading, extend to a private bank's loan-portfolio data or a public-sector undertaking's tender process. Yet the practice has stretched to those contexts because, once the device exists in the procedural toolkit, it is asked to do work the procedural toolkit was never meant to do. This is the second slide. The first slide is from privilege to sealed cover. The second slide is from State use to general use. Each slide is small. The cumulative effect is a procedural regime in which any party that does not want a document on the record has a route to keep it off the record.

The most basic principle of natural justice is the right to be heard - audi alteram partem. In the post-Maneka jurisprudence, the principle is not a polite addition to the writ jurisdiction. It is a constitutional command. The Court's own decisions in Anuradha Bhasin v Union of India on the disclosure of internet-restriction orders, and in P Gopalakrishnan v State of Kerala on disclosure of chargesheet materials, mean that the same Court that uses sealed covers in some cases insists, in others, on disclosure as a constitutional minimum.⁹¹⁰ The internal contradiction is not subtle. It is the kind of contradiction that erodes the doctrinal authority of both branches over time, because nobody can predict which version of the principle the bench will apply in the next case.

Audi alteram partem does not mean that every piece of paper in every government department must be served on every petitioner. It means that the case the bench acts on must be one the affected party has had a meaningful opportunity to meet. Where the State has a genuine interest in non-disclosure, the privilege regime, the in-camera inspection regime, the redaction regime, the amicus-review regime contemplated in Media One - all of these are tools the Court can use without rupturing the basic structure of the hearing. The sealed cover is the one tool that ruptures it. That is what makes it different from the rest.

What retirement would look like

I am not arguing for a constitutional amendment. I am not arguing for legislation. I am arguing for judicial self-discipline, because the practice is judicial in origin and only judicial discipline can end it. Retirement of the sealed cover would require the Court to do three things, in roughly this order.

First, it would have to adopt a rule of presumption against acceptance of materials in sealed cover. The presumption would not be absolute - the privilege regime under Section 133 of the BSA continues to exist - but it would mean that, in the absence of a formal privilege claim that has cleared the statutory procedure, the bench refuses to read what the opposing party may not read. The presumption would not require a constitutional amendment. It would require the bench to enforce against the executive a discipline that the executive has, for a decade now, been allowed to relax.

Second, the bench would have to internalise the structural alternative articulated in Media One - amicus review, public summaries, redacted disclosure. Each of these is a tested tool. None of them produces the structural asymmetry of the sealed cover. None of them requires the opposing party to argue against an invisible record.

Third, the Court would have to be willing to do the harder work in the cases where the executive insists on the sealed cover and the bench refuses it. That harder work is the work of saying to the State: produce your evidence properly, claim your privilege properly, accept the procedural costs of doing so, or do not rely on the materials in the proceeding at all. The cases in which the sealed cover has been most consequential - Rafale, Bhima Koregaon, Media One at the High Court stage, INX Media - are cases in which the executive would have been put to that harder choice. Some of them would have resolved differently. That is the point.

What the absence of a source tells us

I want to close on a structural observation rather than a programme. The sealed cover is one of several procedural devices the Court has constructed without a source in the constitutional text or the procedural codes. Curative petitions are another. The continuing mandamus is another. Court-monitored investigations are another. The phrase that does the work of authorising these devices, when any authorisation is offered, is 'complete justice' under Article 142. The phrase has, over the last three decades, become a kind of constitutional solvent. It dissolves the procedural commitments the rest of the Constitution and the procedural codes have made.

Some of these inventions have been beneficial. The continuing mandamus, used carefully, has produced real progress in cases the executive would otherwise have ignored. Court-monitored investigations have, at their best, broken impasses that the criminal-justice system was structurally unable to break. But each invention has a cost, and the cost is paid by the procedural discipline that the writ jurisdiction was designed to enforce. The sealed cover is the inventive device whose cost is most clearly imbalanced against its benefit. The benefit is opacity for the executive. The cost is the right of the affected party to know what is being said about them and to answer it.

The remedy is not new legislation. The remedy is a Court willing to say, in the next case, that it will not open the envelope. The phrase to remember is the one from Ashby v White that Gopal Subramanium opens his Oxford chapter with: ubi jus ibi remedium. Where there is a right, there is a remedy. The corollary is just as important. Where there is a remedy that does not answer to any right - which is what the sealed cover is, a procedural facility that no constitutional or statutory right authorises - there is no source. A practice without a source is not a procedure. It is a habit. The Court has acquired the habit. It is past time the Court chose to break it.

There is one further thing worth saying about how the practice came to be normalised, because the normalisation is the part that the legal profession has been complicit in. Senior counsel who would, in a contract dispute, object strenuously to the production of an opposing party's documents outside their inspection have, in writ matters touching the Union of India, accepted the sealed cover without recorded objection. The High Court bar has, on the whole, mirrored that acceptance. Where objections have been raised - and a small group of public-interest lawyers has raised them consistently since 2018 - the objections have been treated as procedurally awkward rather than as constitutionally serious. This is the social fact behind the doctrinal fact. The sealed cover is a practice that the bar permitted as much as the bench introduced. Retiring it will require the bar to refuse it on the same instinct it refuses other procedural irregularities. The instinct exists in the profession; it has simply been suspended for this particular device. A return to the ordinary procedural reflexes of the writ jurisdiction, applied to the sealed cover with the same firmness the bar applies to any other document a party tries to keep off the record, is the proximate route by which the practice ends.

Get the week's biggest decisions, distilled.

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