A Question Returns To Three Judges

In November 1941, in the second year of World war-II, the House of Lords decided a case about a man called Robert Liversidge. He had been locked up by the Home Secretary, Sir John Anderson, under a wartime regulation that allowed detention if the Secretary had “reasonable cause to believe” the detainee was of hostile association. Liversidge asked a simple question. What were the reasons? The Home Secretary declined to say. The Lords, four to one, held that he need not. Reasonable cause, in wartime, meant whatever the Secretary in his discretion thought reasonable. The courts would not inquire further.

One judge refused to go along. Lord Atkin wrote a dissent that has outlived every other line in the case. He said that in England, amid the clash of arms, the laws were not silent. He said that he had listened to arguments that might have been addressed acceptably to the court of King's Bench in the time of Charles I. He said that judges, when the liberty of the subject was at stake, should not show themselves more executive-minded than the executive. His brother judges, he added, had treated words like Humpty Dumpty in Alice in Wonderland, who had made words mean what he chose them to mean.

Lord Atkin was ostracised. The Lord Chancellor cut him in the corridors. His colleagues stopped lunching with him. He died three years later, his name a byword for stubbornness. Today, no serious lawyer cites the majority. Everyone cites Atkin. History does these corrections, though it takes its time. I retell that story because something of its shape is now playing out at the Supreme Court of India.

Lawyers will remember the question. Section 43-D(5) of the Unlawful Activities (Prevention) Act bars bail when the court has reasonable grounds to believe the police case is true. A three-judge bench in K.A. Najeeb in 2021 said that this bar must yield to Article 21 where trial is nowhere in sight and incarceration has lasted long enough. Two-judge benches then began to read Najeeb narrowly. Gurwinder Singh in 2024 and Gulfisha Fatima in January 2026 treated it as rare exceptions. In May, Justices Nagarathna and Bhuyan, in Andrabi, said openly that this was a hollowing out of the larger bench's ruling, and granted bail.

Now comes the next chapter. On 22 May, in Tasleem Ahmed and Khalid Saifi, Justices Aravind Kumar and Prasanna B. Varale have referred the question to the Chief Justice for constitution of a larger bench. Justice Kumar, who authored Gulfisha Fatima, has effectively asked the institution to settle what he and Justice Bhuyan could not settle by trading paragraphs. He has, to his credit, granted interim bail to the appellants for six months while the question is resolved. That is the right course. A bench of two cannot answer a bench of two by writing more pages. The proper response to perceived error is reference, not rejoinder.

What will the larger bench have to decide?

It will have to decide, first, what Najeeb actually held. Did the three-judge bench lay down a binding constitutional rule, or did it carve out a narrow factual exception to Watali? Andrabi reads it the first way. Gulfisha reads it the second. The distinction matters. A rule travels. An exception does not.

It will have to decide, second, what weight prolonged incarceration carries when the trial has barely begun. The petitioner has spent five years in jail. The prosecution has three hundred witnesses still to call. The case will not end this decade. Is liberty available on those facts alone, or must the court also weigh the gravity of the allegation, the centrality of the role, and the risk of interference with the trial? Andrabi leans towards the first. Gulfisha insists on the second.

It will have to decide, third, what the conviction figures mean. The Minister of State for Home Affairs told Parliament in December 2025 that ninety-four to ninety-eight per cent of UAPA cases end in acquittal nationally, and ninety-nine per cent in Jammu and Kashmir. The numbers are not evidence against any particular accused. They are, however, evidence about a statute. They tell us that the Act is being used to imprison people who, more often than not, will eventually be found innocent. A court that ignores those figures is not being judicial. It is being polite.

It will have to decide, fourth, the question of judicial discipline. Justice Kumar's order on 22 May rightly notes that a coordinate bench cannot answer another coordinate bench through emphatic disagreement. The proper course is reference. That principle protects the institution. But it must be applied symmetrically. If the original drift began when two-judge benches narrowed a three-judge ruling, the original lapse of discipline lay there. The way to set it right is to restore the larger ruling, not to split the difference.

And it will have to decide, fifth and underneath everything else, what posture the court adopts when the executive presents a serious allegation.

This is where Lord Atkin returns to the picture. The temptation, in a case involving terror, riot, conspiracy or the security of the State, is to defer. The papers look weighty. The investigating agency speaks gravely. Sealed covers are produced. The prosecutor invokes the nation. Judges, who are human, are reluctant to be the ones who let a dangerous person out. Better, the thought runs, to err on the side of caution. Better to keep him in for another year. Better to let the trial decide.

That posture, repeated case after case, produces a court that is no longer a constitutional court. It becomes an executive court. It validates whatever the State puts before it because the State has put it. It treats reasonable grounds to believe as whatever the prosecution believes. Words begin to mean what the executive chooses them to mean.

Atkin's warning was not about wartime alone. It was about the habit of judicial deference that wartime exposes. The habit does not need a war. It needs only a sufficiently grave-sounding allegation, a sufficiently confident State, and a court that prefers the company of the powerful to the liberty of the citizen.

The Indian Constitution does not contemplate such a court. It contemplates a court that examines, that doubts, that asks for reasons, that releases the prisoner when the State cannot show why he must remain. Article 21 is not a sentiment. It is a working command. Section 43-D(5) is a statute. Statutes work within Article 21, not above it.

I hope the larger bench will, say so. It will reaffirm Najeeb. It will read Watali down to its facts. It will tell judges that scepticism of executive claims is not disloyalty to the State. It is the daily business of a constitutional court. It will remind itself that the figures from the Lok Sabha matter, that five years in jail is not a procedural detail, and that the burden of justifying continued incarceration lies on the prosecution and not on the prisoner. But diamond hard, diamond bright hope in the Supreme Court, has at times been betrayed.

If the court leans towards liberty , it will have done, what its judges in ADM Jabalpur with the honourable exception of Justice HR Khanna, could not bring themselves to do. To stand in a difficult moment, on the side of the citizen is what a constitutional court is for.

Sanjay Hedgde is a Senior Advocate of Supreme Court. Views are personal.