If the Referendum Provision is Back, So are Its Restraints

Can a constitutional safeguard be restored by a court yet depend on Parliament’s permission to exist?

On October 6, Home Minister Salahuddin Ahmed invoked the doctrine of eclipse to explain why invalidating the Fifteenth Amendment’s deletion revived the earlier referendum provision.

He also said full restoration required parliamentary incorporation. That combination calls for legal clarity, not competing political assurances.

The contrary view is that parliamentary restoration of Article 142’s constitutional machinery must precede a referendum. This disagreement deserves examination as a question of law. Its starting point must be the judicial order, not the governing party’s description of it.

In its December 17, 2024 judgment, on page 137, the High Court expressly declared: “Consequently, Article 142 as it then was vide Twelfth Amendment Act, 1991 is hereby restored.” The Appellate Division subsequently upheld the judgment on July 9, 2026.

Restoration therefore cannot be dismissed as an inference extracted from judicial silence.

The case for automatic revival must nevertheless confront the General Clauses Act. Article 152(2) of the Constitution applies that Act to the Constitution and treats enactments made void by it like legislation repealed by Parliament.

Sections 6 and 6A generally prevent repeal from reviving extinct provisions or undoing textual amendments, unless a different intention appears; section 7 requires an express purpose when legislation revives a repealed enactment. Merely distinguishing judicial invalidation from legislative repeal does not fully answer Article 152.

But the opposite conclusion does not follow automatically either. Those statutory rules must be read alongside the court’s constitutional jurisdiction and express remedy. Here, restoration was actually ordered. The stronger reading, in my view, is that the restored provision does not await parliamentary permission to exist.

Otherwise, a judicially restored constitutional constraint would depend on the consent of the institution it constrains. This interprets the particular restoration order; it does not claim that every objection has been expressly adjudicated. 15th amendment(4)

An important analogy appears in the Appellate Division’s Thirteenth Amendment review judgment. On pages 62 and 63, the Court held that the caretaker provisions were restored forthwith, but could operate only when their constitutional conditions arose, at the earliest after dissolution of the 13th Parliament.

That judgment did not decide this referendum dispute. It nevertheless distinguishes restoration from activation. A provision may exist even though its use remains conditional.

Further parliamentary action need not contradict revival. But its purpose matters. Updating the published Constitution to reflect a judgment, supplying legislation for administering a referendum, and amending the restored provision are different exercises.

The first records existing law; the second supplies procedural machinery; the third changes the constitutional settlement. Calling all three “restoration” obscures who is exercizing which power.

The restored Article 142 has a defined reach. Its referendum requirement concerns bills amending the Preamble, Articles 8, 48 and 56, and Article 142 itself. It does not require a referendum for every constitutional amendment.

The High Court reproduces and explains this framework on pages 78 to 80. Restoration brings back a specified procedure, not an unrestricted political licence.

On the minister’s own account, the forthcoming amendment will address the Preamble and Articles 48, 56 and 142. The question is therefore practical: The government’s stated programme engages the very provisions whose referendum safeguards it says have returned.

Parliament must first pass the relevant bill by at least two-thirds of its total membership. Within seven days of its presentation, the president must cause the question of assent to be referred to a referendum. Seven days is the referral deadline, not the deadline for conducting the poll.

The Election Commission conducts it under procedures prescribed by law. Upon declaration of the result, assent is deemed given if a majority of votes cast supports the bill, and withheld otherwise (pages 77 to 79).

It follows that a bill changing Article 142 must comply with Article 142 as it already stands. Parliament cannot assume that the referendum requirement becomes binding only after the amendment which changes it. The government’s proposed referendum could satisfy the existing requirement.

A promise to introduce that requirement later could not. The High Court identified precisely this procedural defect in the 2011 deletion: referendum safeguards were removed without a referendum (pages 85 and 86).

Restoration also differs from implementing the reform proposals. The July Charter proposes referendum protection for the caretaker provisions alongside the previously specified subjects.

The Constitution Reform Commission recommended referendum approval for all constitutional amendments following approval by two-thirds of each parliamentary chamber (volume I, page 53). Restoring the 1991 text does not, by itself, enact either expanded design.

That difference has practical consequences. Citizens should be told whether they are receiving the old safeguard, an expanded safeguard, or a differently designed one.

Judicial restoration should not become a substitute for explaining which reform commitments a proposed amendment fulfils. Equally, agreement that Article 142 has revived does not amount to endorsing the substance of whatever amendment the government eventually presents.

The design of that amendment deserves scrutiny too. A voter might support stronger protection for neutral elections while opposing an unrelated change to executive power. Bundling distinct reforms into one bill can force that voter into an artificial choice.

This is not an argument that Article 142 prohibits every composite bill. It is a demand that lawmakers justify the coherence of their package and permit public objections to reshape it before parliamentary passage.

One restored safeguard is especially important. Article 142(1D), reproduced on page 79 of the High Court judgment, prevents the referendum outcome from being deemed a vote of confidence or no confidence in the cabinet or Parliament.

A rejected amendment does not automatically dismiss the government. Campaigners should respect that distinction instead of turning a constitutional choice into a plebiscite on those in office. 15th amendment(4)

Meaningful consent also requires more than an accurately counted ballot. The public should have accessible explanations of the competing positions, reasonable opportunities to campaign against the proposal, and confidence that state resources will not be used to make disagreement prohibitively costly.

These are standards against which the process should be judged, not benefits the governing party may selectively confer.

The immediate task is transparency. Publish the restored text and the judicial orders supporting it. Identify which clauses of the proposed bill reflect existing law and which make new policy choices.

Explain the referendum’s statutory framework, and give the public time to scrutinize the actual bill. A ballot cannot provide informed consent to language voters have not had a fair opportunity to understand.

The government has substantial support for its claim that Article 142 has revived. But restoration is not a selective privilege. It brings back the people’s power to withhold assent, as well as the rules limiting government initiative.

If the referendum provision is already back, its restraints bind now. They are not future concessions for the ruling party to announce.

Barrister Khan Khalid Adnan is advocate of the Supreme Court of Bangladesh, fellow at the Chartered Institute of Arbitrators, and head of the chamber at Khan Saifur Rahman and Associates in Dhaka.