The history of UAPA bail begins with a puzzle of proportion, because Section 43D(5) of the Unlawful Activities (Prevention) Act, 1967 occupies only a small space in the statute book, and its operative proviso is shorter still, yet within six years the Supreme Court gave that proviso several distinct readings, each reading altering what a court was permitted to look at, how closely it was permitted to look, and what it was permitted to do with what it saw. A provision drafted as a threshold condition for release became, through interpretation, a contest about the nature of judging at the bail stage, about the weight to be given to the prosecution case before any evidence has been tested, and about the place of liberty in such a prosecution. The words themselves did not change in that period, so the movement in the law has to be explained by the way different benches understood the same sentence, the same materials, and the same instruction to form an opinion about an accusation. That is the interpretive tension with which any account of the provision has to begin, because the difficulty has almost always been about how a court is to obey the section in a concrete case where a person remains in custody, the investigation papers are before the court, and the accusation has not yet been put to proof at a trial.

The statute the courts were given: Section 43D and the architecture of the bar

To understand what the courts were being asked to interpret, the provision has to be placed inside the Act of which it forms a part, because Section 43D is not a freestanding bail code and it borrows much of its meaning from the offences and procedures that surround it. Taken as a whole, the Unlawful Activities (Prevention) Act, 1967 is the statute through which Parliament provides for the prevention of unlawful activities and for dealing with terrorist activities, setting out definitions, offences, penalties, forfeiture, and special procedures in a single enactment, and a reader who wants the full shape of that enactment set out in one place can follow the complete guide to the UAPA 1967 before returning to the narrower question of bail. The Act separates its offences into chapters, and that separation matters far more for bail than a casual reading might suggest, because the bail bar in Section 43D(5) is written by reference to chapters rather than by reference to the Act as a whole. A provision that looks like a general rule for every person prosecuted under the statute turns out to be a rule with a defined perimeter, and much of the later litigation has turned on whether a particular accusation falls inside that perimeter or outside it. Section 43D also gathers several distinct procedural subjects in one section, dealing with the period of investigation, with the exclusion of anticipatory bail, with the conditions for release on bail itself, and with the relationship between its restrictions and those already found in the Code of Criminal Procedure. Reading subsection (5) in isolation makes the bar appear simpler and more absolute than the section as enacted actually makes it.

Section 43D was inserted into the Act by the Unlawful Activities (Prevention) Amendment Act, 2008, an amendment enacted after the Mumbai attacks of November 2008, and the timing of that insertion is part of the context in which the section has always been read, even by courts that have insisted that context cannot displace the words enacted. The insertion brought a special procedural regime for offences under the Act into the statute in a consolidated form, rather than leaving the questions of custody, investigation time, and bail to be answered only by the ordinary provisions of the Code of Criminal Procedure, and the steps by which that regime was built up across successive changes to the Act are traced in the amendment trail of the UAPA, which shows Section 43D as a product of a particular legislative moment rather than as part of the original 1967 design. The point is not merely historical, because the origin of the section explains why its subsections pull in different directions at once. Parliament was concerned with the time an investigation of this kind might reasonably require, with the risk that anticipatory bail might present, and with the seriousness of releasing a person against whom an accusation of a specified kind had been made and recorded in the investigation papers, and it answered those concerns in different subsections with different devices. An extension of time is directed at the calendar of the investigation. An exclusion of anticipatory bail is directed at the stage before arrest. A bar on release, conditioned on the opinion the court forms from the case diary or the police report, is directed at the merits of the accusation as they appear on paper at the bail stage. Keeping those devices distinct is essential, because confusion in argument has often come from treating the section as if it expressed a single policy of detention, when the text expresses several policies, each with its own conditions and limits.

What does Section 43D(5) of the UAPA actually say?

Section 43D(5) provides that a person accused of an offence under Chapter IV or Chapter VI of the Act, if in custody, shall not be released on bail unless the Public Prosecutor has been heard, and shall not be released if the court finds reasonable grounds for believing the accusation is prima facie true.

The first thing the full text establishes is that subsection (5) contains two conditions, not one, and they operate in sequence. The opening words provide that no person accused of an offence punishable under Chapters IV and VI of the Act shall, if in custody, be released on bail or on his own bond unless the Public Prosecutor has been given an opportunity of being heard on the application for release. That hearing requirement is procedural, and it is mandatory in form, because release is forbidden unless the opportunity has been given, so a court that released an accused person without hearing the Public Prosecutor would act against the opening limb even before it reached any question about the merits. The proviso then adds the second and far more contested condition, providing that such a person shall not be released on bail if the court, “on a perusal of the case diary or the report made under section 173 of the Code”, is of the opinion that there are “reasonable grounds for believing that the accusation against such person is prima facie true”. Every element of that sentence has been pressed into service in argument. The material is specified, the case diary or the report under Section 173 of the Code. The mental act is specified, an opinion formed on a perusal. The standard is specified, reasonable grounds for believing. The object of the belief is specified, that the accusation is prima facie true. And the consequence is specified in prohibitory language, because the person shall not be released. The brevity of the proviso is therefore deceptive, because it compresses a source, a method, a standard, and a prohibition into a single sentence, and the disputes in the cases are disputes about the relationship among those four elements.

The reference to a perusal of the case diary or the report made under Section 173 of the Code deserves attention first, because it tells the court where to look and, by implication, raises the question of where else, if anywhere, the court is entitled to look. A perusal is a reading, and the word suggests an examination of the material placed before the court rather than a hearing of evidence in the manner of a trial, with witnesses examined and cross examined and documents proved. The report under Section 173 of the Code is the report that the police submit on completion of an investigation, and the case diary is the record maintained during the investigation itself, so the proviso directs the court to the prosecution’s own paper record of the accusation at the stage at which bail is sought. That direction has mattered for two opposed reasons. For the prosecution, it has supported the argument that the court’s task is confined and documentary, because Parliament has named the material from which the opinion is to be formed. For the accused, the same direction has supported the argument that a perusal must be a real reading rather than a formal glance, because an opinion formed without engaging with the diary or the report would not be an opinion formed on a perusal at all. Both arguments start from the same words, and the tension between confinement to the named material and genuine engagement with that material runs through the case law that followed.

The standard set by the proviso, reasonable grounds for believing that the accusation is prima facie true, is equally resistant to a single, settled meaning, and its internal layering is part of the reason why. The phrase prima facie, taken alone, ordinarily describes a case that appears sufficient on its face, without deeper inquiry, and a court asked only whether an accusation was prima facie true might be understood to be asking a relatively light question about appearances. The words reasonable grounds for believing, taken alone, point in a different direction, because they ask whether grounds exist, whether those grounds are reasonable, and whether they support a belief, which is a stronger state than suspicion and a more considered state than a first impression. The proviso places the two together, so that the belief to be formed is a belief that the accusation is prima facie true, and the grounds for that belief must be reasonable, and a court has to decide how much work each part of the phrase is doing. If the prima facie element dominates, the inquiry stays near the surface of the prosecution material and the bar operates readily. If the reasonable grounds element dominates, the inquiry acquires a more evaluative character, and the court must satisfy itself about the quality of the material from which the belief is to be drawn. The case law examined later in this article is, in large measure, an argument between those two emphases conducted case by case, because Parliament used both formulations in one breath and left the courts to construct a workable test from the combination.

Scope is the third feature of the subsection that a careful reading has to settle before any test is applied, because the bar does not attach to the Act in general. By its opening words, subsection (5) speaks of a person accused of an offence punishable under Chapters IV and VI of the Act, and offences that sit elsewhere in the Act are therefore outside the words of the bar, however serious they may be in other respects. The consequence can be stated concretely. Section 13 of the Act, which deals with unlawful activities, sits in Chapter III, and a bare charge under Section 13, standing alone and unaccompanied by an offence under Chapter IV or Chapter VI, falls outside Section 43D(5), so the special bar does not govern an application for release on such a charge and the ordinary principles of bail under the Code apply instead. The Supreme Court later confirmed that reading in Thwaha Fasal, treating the location of the offence within the chapter structure of the Act as decisive for the applicability of the bar. The confirmation matters because it shows that the perimeter of the proviso is a question of statutory construction to be answered before any question about the merits arises. It also matters for practice, because charge sheets may contain several offences drawn from different chapters, and a court has first to identify which accusations attract the bar and which do not, rather than treating the presence of the Act in the charge sheet as sufficient to engage subsection (5) for every count. A bar that is severe within its field may still be limited in its field, and the severity and the limit come from the same sentence.

The remaining subsections complete the architecture. Subsection (2) modifies the ordinary rule in Section 167 of the Code for offences under the Act, by allowing the period within which the investigation must be completed and the charge sheet filed to be extended from 90 days to a possible 180 days, where the court is satisfied on the report of the Public Prosecutor indicating the progress of the investigation and the specific reasons for detention beyond the initial period. The extension is therefore neither automatic nor a matter of administrative convenience, because it depends on a report, on reasons, and on the satisfaction of the court, and it doubles the outer limit rather than removing a limit altogether. Subsection (4) works at the opposite end of the process, providing that nothing in Section 438 of the Code shall apply in relation to any case involving the arrest of a person accused of having committed an offence under the Act, which means that anticipatory bail is excluded for such cases and the question of release can only arise after arrest and custody. Subsection (6) then states the relationship between the special restrictions and the general law, providing that the restrictions on granting bail specified in subsection (5) are in addition to the restrictions under the Code or any other law for the time being in force on granting bail. The word additional does a great deal of work, because it establishes that satisfying the ordinary conditions for bail does not by itself overcome the special bar, and, equally, that the special bar is layered over a body of general law that continues to exist beneath it, so that an accused person may pass the ordinary principles and still meet the bar, or may fall outside the bar altogether and be governed by the ordinary principles alone. The section, read whole, is therefore a set of modifications to a general code, each with a defined subject, and the courts were given that layered structure, not a single command, when they came to interpret the proviso.

The road the bar does not close: default bail

Against that architecture, one route to release stands apart from the merits inquiry altogether, and it has to be understood before the case law on the proviso can be followed, because argument in bail applications has often moved between the two routes as if they were one, when in law they rest on entirely different foundations. That route is default bail, which arises under Section 167(2) of the Code as modified for offences under the Act by Section 43D(2), and it depends on the calendar rather than on any opinion about the truth of the accusation. The investigation must be completed within the period allowed, which is 90 days in the ordinary course and, where the period has been extended in accordance with Section 43D(2), a possible 180 days. If the investigation is not completed within that period, the accused person who is prepared to furnish bail must be released on bail, because the authority to continue detention on the basis of an unfinished investigation has run out. The entitlement does not ask the court to weigh the accusation, to read the case diary for reasonable grounds, or to form any view about whether the accusation is prima facie true, and that is precisely why the proviso to Section 43D(5) does not close this road. The proviso governs release where the court is deciding, on the merits as they appear in the investigation papers, whether a person in custody should be let out, while default bail governs the different situation in which the time allowed for the investigation has expired without its completion. A reader tracing how the general provision operates across offences will find its structure set out in the complete guide to the Code of Criminal Procedure 1973, and the point for the purposes of the Act is that Section 43D modifies the length of the period without abolishing the consequence that follows when the period expires.

The constitutional character of that entitlement was settled for the purposes of this line of authority in Bikramjit Singh v State of Punjab, decided on 12 October 2020 by a three judge bench of the Supreme Court comprising Justices R. F. Nariman, Navin Sinha and K. M. Joseph. The Court in Bikramjit Singh held that the right to default bail under the first proviso to Section 167(2) of the Code is not a mere statutory right, of the kind that a special statute might be taken to have displaced by imposing stricter conditions on release, but is part of the procedure established by law under Article 21 of the Constitution, and is therefore a fundamental right that forms part of the guarantee of personal liberty itself. That characterisation changes the way the relationship between the Code and the Act has to be described, because if the entitlement on the expiry of the investigation period is an incident of Article 21, then a provision such as the proviso to Section 43D(5), which speaks to the merits of the accusation, cannot be read as silently withdrawing it. The Court in Bikramjit Singh also addressed the machinery of extension, holding that only the Special Court may extend the investigation period to 180 days under Section 43D(2) of the Act. The two holdings belong together. The first fixes the rank of the right, placing it within Article 21 rather than at the level of a concession in a procedural code. The second fixes the identity of the court that can postpone the moment at which the right accrues, and confines that power to the Special Court acting under the subsection, on the material the subsection requires.

The consequences of keeping the two routes separate are considerable, and they explain why the sequence of events in a case determines the outcome of a release application before argument about the accusation is heard. Where an extension to 180 days has been properly granted by the Special Court, the calendar runs to the longer date and no question of default bail arises in the interval, so the accused person’s application falls to be decided under the ordinary principles as qualified by the bar in Section 43D(5), with the perusal of the case diary or the report under Section 173 and the opinion about prima facie truth at the centre of the inquiry. Where no such extension has been granted, or where the period as extended has expired without the charge sheet being filed, the calendar answers the application on its own terms, and the strength or weakness of the accusation as it appears on paper is not the question the court is asking. Confusing the two inquiries distorts both, because it allows a merits bar to swallow a time based right that the Supreme Court in Bikramjit Singh located in Article 21, and it allows a time based right to be debated as if it depended on the defence case, when its whole purpose is to operate without regard to the merits at all. The distinction also disciplines investigations, since the prosecution that wishes to hold an accused person beyond the initial period must seek the extension in time, from the proper court, on a report that shows the progress made and the reasons for further detention.

With the statute described and the default bail route marked off from it, the question that the remainder of this article answers can be framed with some precision, and it is a question about the reach of the bar rather than about its existence. On its face, the proviso to Section 43D(5) speaks to the court that is asked to release a person on bail in the course of the proceedings under the Act, which is the Special Court in the first instance, and to the appellate courts that reconsider such decisions within the same statutory scheme, and for those courts the command of the proviso is plainly addressed in terms. The harder question, and the one over which the case law has fought, is whether the bar also binds a High Court or the Supreme Court when either court is exercising constitutional jurisdiction under Article 21 of the Constitution, on the ground that detention under the Act has itself become a violation of the guarantee of personal liberty, through delay, through the state of the trial, or through some other feature of the custody that the constitutional court is asked to examine. If the proviso governs every judicial door through which release might be sought, then its interpretation is close to the whole law of release under the Act. If the proviso governs the statutory courts acting under the Act, while the constitutional courts retain a jurisdiction under Article 21 that the proviso does not exhaust, then two tracks of authority exist side by side, and everything depends on the conditions under which a case may move from one track to the other. The cases that follow are best understood as a sustained judicial argument about that boundary, conducted through the language of prima facie truth, reasonable grounds, and the perusal of a case diary, but directed throughout at the larger issue of which court, acting under which power, is finally entitled to order that a person accused under the Act be released.

A decade of quiet: 2008 to 2018

The strangest feature of the bail bar inserted into the Unlawful Activities (Prevention) Act in 2008 is not that it was severe, but that it remained so largely unexamined for a full decade. A provision that altered the ordinary relationship between accusation and custody, that required a court to refuse bail where the accusation appeared prima facie true, and that therefore placed the liberty of an undertrial prisoner behind the quality of the prosecution papers, might have been expected to generate a steady stream of appellate reasoning. It did not. Between its insertion and the end of the following decade, the higher courts produced remarkably little sustained guidance on what the proviso demanded, how a judge was to form the required opinion, what material could be looked at, or where the boundary lay between a permissible assessment at the bail stage and an impermissible trial conducted in advance. The provision operated, but it operated mostly in the shadows of trial court orders, brief appellate affirmations, and refusals that turned on their own facts without laying down a method that later courts could follow with confidence.

Several conditions combined to keep the question quiet. The first was the simple scarcity of contested bail appeals reaching the Supreme Court. Bail litigation under the statute usually began and ended far below the apex court. An accused person applied before the Special Court or the trial court, the application was resisted by the prosecution on the strength of the chargesheet and the case diary, and the order that followed was frequently treated as the effective end of the matter for a long period. A further challenge before the High Court required resources, preparation, and time, and a further journey to the Supreme Court required still more of each. Many undertrial prisoners, already in custody and already facing a prosecution with a large documentary record, had none of those advantages in abundance. The result was a body of law that developed, if it developed at all, at a level where orders were short, fact specific, and rarely reported in a form that could discipline later decision making across the country.

The second condition was the manner in which trial courts applied the bar. Faced with a chargesheet that alleged offences under the statute and with a proviso that spoke in restrictive terms, the safer course for a trial judge was refusal, and the reasoning that supported refusal could be kept deliberately thin. A court did not need to say a great deal. It could record that the material collected by the investigating agency disclosed a prima facie case, that the accusations were serious, that the investigation pointed to involvement in the alleged conspiracy, and that the statutory condition for release was therefore not satisfied. Such orders answered the application before the court, but they left the underlying method untouched. They rarely explained how much of the material had been read, whether contradictions within it mattered, whether a document that was plainly weak could be separated from a document that was strong, or whether the court was entitled to ask if the material, even if accepted, actually connected the particular accused to the particular act alleged. Mechanical application produced orders without producing jurisprudence.

The appellate route, where it was taken, offered little correction and moved slowly. Bail appeals do not advance in isolation from the rest of the criminal process. They compete for hearing time with the trial itself, with applications by co accused persons, with challenges to charges, and with the ordinary press of criminal work before the High Courts. An appeal against a refusal could therefore wait while custody continued, and by the time it was heard the very passage of time had changed the practical stakes without necessarily changing the legal answer. If the appeal failed, the prisoner returned to the same position before the trial court, with liberty dependent on a trial that had barely moved. If the appeal succeeded, it often did so on a close reading of the facts of that one case, in language that later benches could distinguish or confine. Either way, the central interpretive questions about the proviso, its degree of satisfaction, its treatment of evidence, and its relationship to prolonged custody, remained largely unanswered in any authoritative and general form.

That quiet would have mattered less if the underlying trials had been swift. They were not, and the background condition that eventually made the bail question urgent was the pace of UAPA trials themselves. Proceedings under the statute commonly involve wide conspiracies as alleged by the prosecution, several accused persons tried together, documentary records running across many volumes, electronic material whose contents must be proved through witnesses, and witness lists that extend to very large numbers. Each witness must be examined and cross examined, each document must be put in the legally recognised manner, and delays compound when accused persons are produced from different jails, when counsel for one accused is unavailable, when a witness does not appear, or when interlocutory applications interrupt the sequence of evidence. None of this requires bad faith on any side. It is the ordinary arithmetic of a heavy trial. Yet the consequence for bail is direct and unforgiving. A refusal at the beginning of such a proceeding is not a refusal for a few weeks until the evidence is tested in court. It can mean years of undertrial custody during which no witness has been cross examined, no document has been proved, and the accusation on paper remains the only version of the case that any court has examined.

How did the courts treat UAPA bail before Watali?

Before Watali, trial courts treated the statutory bar as close to decisive once a chargesheet alleged a terrorist act, the Supreme Court offered little corrective guidance, and appellate courts intervened sparingly, so bail was usually refused on the prosecution papers alone without any sustained examination of delay, custody or proportionality.

That summary, though compressed, captures a pattern that was more a habit of decision making than a settled doctrine. Courts were not ignoring the statute. On the contrary, the statute was frequently the whole of the reasoning. The proviso told the judge that bail must be refused if the court, after perusing the case diary or the chargesheet, formed an opinion that there were reasonable grounds for believing the accusation to be prima facie true. In the absence of authoritative elaboration, many courts read that instruction at its broadest and simplest. The chargesheet was perused, the allegations were noted, the gravity of the offences was recorded, and the opinion followed almost as a matter of course. What was missing was the intermediate discipline: a requirement to show how the material bore on the individual accused, how the court had treated a statement that conflicted with a document, or how a piece of material that was merely repetitive added anything to a piece that was direct. The test existed in words, but the working method for applying it remained underdeveloped.

There was also a structural reason why appellate courts intervened sparingly, beyond the ordinary respect owed to a trial court order. An appellate court hearing a bail challenge in this setting inherited the same undeveloped test as the court below. Without a settled statement of how deeply it could examine the material, the appellate bench had to choose between two uncomfortable courses. It could confine itself to asking whether the trial court had perused the record and expressed the statutory opinion, in which case the appeal added little. Or it could open the record for itself and examine documents and statements in detail, in which case it risked being accused of converting the bail hearing into a rehearsal of the trial, with findings that might embarrass the trial or prejudice one side. Many benches, understandably cautious, chose the narrower course. The consequence was cumulative. Trial courts knew that a short refusal grounded in the proviso was unlikely to be disturbed, appellate courts knew that a fuller intervention lacked a clear doctrinal footing, and counsel knew that the prospects of release before trial were poor unless the prosecution case contained a defect so obvious that it could be shown without any deep examination at all.

Delay and length of custody sat at the edge of this reasoning throughout the decade, acknowledged in individual cases perhaps, but not integrated into the test. A judge could of course note that an accused had been in custody for a long period, and such a fact might support a sympathetic hearing or an order expediting the trial. What had not been established was whether prolonged undertrial detention could itself alter the bail analysis under the statute, whether Article 21 of the Constitution supplied an independent route to release when the trial could not conclude within a reasonable time, or whether the statutory bar, once attracted on the papers, answered every argument including an argument founded on years already spent in custody awaiting evidence. Those questions were present in the background of many applications, because counsel could hardly ignore the calendar, but the reported jurisprudence of the decade gave them no firm place in the structure of the decision. Bail reasoning therefore tended to stop at the chargesheet. The passage of time after the chargesheet belonged, if it belonged anywhere, to a separate and uncertain conversation about speedy trial, and the two conversations rarely met.

Watali: the restrictive baseline

National Investigation Agency v Zahoor Ahmad Shah Watali became that decision. The accused, Zahoor Ahmad Shah Watali, was a businessman from Kashmir whom the prosecution accused of acting as a conduit in a terror funding conspiracy. The allegation, in the form in which it reached the bail courts, was not that he had himself carried out an act of violence, but that money and support had moved through channels with which he was said to be associated, for the benefit of activities that the prosecution placed within the statute. Cases of this kind present a particular difficulty at the bail stage. The accusation is built from financial trails, meetings, communications, and statements that acquire their alleged significance from the pattern they are said to form when read together. No single document necessarily speaks for itself. The prosecution invites the court to see a design in the aggregate, while the defence invites the court to see a set of disconnected transactions and associations, each capable of an innocent explanation. How a bail court is to handle that invitation, what it may examine and what it must leave alone, was the question that the litigation in Watali forced into the open.

The procedural path of the case exposed the two competing approaches in unusually clear form. The Special Court refused bail on 8 June 2018. The Delhi High Court took a different view and granted bail on 13 September 2018, after examining the material closely. The contrast between those two orders is central to understanding the Supreme Court judgment, because the High Court had not released the accused on a general ground of delay or on a broad appeal to liberty disconnected from the record. It had gone into the material that the prosecution relied upon, assessed the weight that particular items could bear, and concluded that the case presented did not justify continued custody at that stage. The National Investigation Agency carried the matter to the Supreme Court, and the question before the apex court was therefore not an abstract one. It was whether the close examination undertaken by the High Court was the kind of examination that the proviso permits at all. The answer given would define the method for every bail court that followed.

The Supreme Court, a two judge bench of Justices A. M. Khanwilkar and Ajay Rastogi, set the High Court order aside on 2 April 2019. The operative conclusion was that bail had been wrongly granted, but the lasting importance of the judgment lies in the reasoning by which that conclusion was reached. The bench treated the case as an occasion to state the approach required by the proviso in general terms, and it stated that approach in a sequence of connected propositions that left little room for the method the High Court had adopted. Read narrowly, in the manner the task of interpretation requires, the holding was about the approach a court must follow when it applies the proviso on the merits of the accusation. The court must reckon the totality of the material in the chargesheet and the case diary. It must take that material at face value. It must not conduct a mini trial. It must not weigh the admissibility or probative value of individual pieces of evidence. It must not discard the statements of protected witnesses. And the degree of satisfaction it forms under the proviso is lighter than in an ordinary bail case, because the court is forming an opinion on broad probabilities rather than proof.

Each element of that sequence carries weight, and the elements must be read together rather than in isolation. The insistence on totality was directed against a selective reading of the record. A bail court, on this view, is not free to seize upon the weakest item in the chargesheet, demonstrate that it would carry little weight at trial, and treat that demonstration as an answer to the whole application. The prosecution case at the bail stage is the case as assembled in the chargesheet and the case diary, and the court must form its opinion on the material as a whole, including statements and documents that the defence considers unreliable or incomplete. Conversely, totality also means that the court is not to stop at the summary of allegations with which such documents often begin. It is to peruse the material itself. The restrictive force of Watali does not lie in releasing the court from reading. It lies in prescribing what the court may do with what it has read. Having read the whole record, the court must accept its contents at face value for the purpose of the bail opinion, rather than test those contents as a trial court tests evidence after examination and cross examination.

The prohibition on a mini trial follows from that prescription, and it is the part of the judgment that bail courts have had the least difficulty in invoking and the greatest difficulty in applying with precision. A mini trial, in the sense condemned in Watali, means an inquiry in which the bail court resolves disputed questions of fact, decides whether a witness is telling the truth, prefers one version of an event over another, or records findings of the kind that properly belong to the judgment at the end of the trial. The bench held that such an inquiry is impermissible at the bail stage under the proviso. The reason is not hard to see. Bail is decided early, on material that has not been proved, by a court that has heard argument rather than evidence, and any detailed findings recorded at that stage could prejudice the trial, embarrass witnesses, or create an appearance that the outcome has been prejudged. Yet the prohibition also removed the principal tool by which the Delhi High Court had reached its conclusion in the case itself. If a court cannot weigh what a document actually shows against what the prosecution says it shows, beyond taking the document at face value within the totality, then the scope for a defence argument founded on the weakness of individual items is sharply reduced.

The treatment of admissibility and probative value carries the same logic further. In an ordinary trial, evidence law performs a filtering function. Material may be excluded because it is inadmissible, and material that is admitted may still be given little weight because it is hearsay in substance, because it is contradicted, because its source is compromised, or because it does not connect logically with the fact it is offered to prove. Watali held that this filtering function is not to be performed at the bail stage under the proviso. The court must not weigh the admissibility of individual pieces of evidence, and it must not assess their probative value in the manner of a trial court. The statements of protected witnesses were given express protection within this approach. Such statements, whose authors are shielded because of the sensitivity attributed to their evidence, could not be discarded by the bail court on the ground that their contents could not be tested or that their identity and credibility could not be examined. To discard them, on the reasoning of the bench, would be to write out of the totality a part of the very material on which the prosecution case rests, before the trial had given that material its proper setting.

The statement about the degree of satisfaction completes the structure and explains why the bench described the exercise as one of broad probabilities. In an ordinary bail case, a court weighs several considerations together, among them the nature of the accusation, the strength of the material, the risk of flight, the risk of interference with witnesses, and the period of custody already undergone, and it exercises a discretion that is not controlled by a single statutory opinion about whether the accusation appears true. Under the proviso as expounded in Watali, the inquiry is narrower and the threshold for the prosecution is lighter. The court asks whether, on the totality of the material taken at face value, there are reasonable grounds for believing that the accusation is prima facie true. That is an opinion formed on broad probabilities. It is not proof beyond reasonable doubt, it is not a preponderance established after evidence, and it is not even the kind of reasoned satisfaction that a court records when it frames charges after hearing both sides on the record of a completed investigation. Because the satisfaction is lighter, material that might eventually fail at trial can still be sufficient at the bail stage to attract the bar, provided it forms part of a totality that, accepted as presented, points towards the accusation.

The practical consequence of the judgment was therefore severe and was understood at once by trial courts and by counsel. Once a chargesheet was filed and disclosed material connecting the accused to the accusation, bail became very difficult to obtain. The route that had been open in principle during the quiet decade, a close item by item examination of the prosecution papers in the hope of showing that the connection was illusory or that a key document did not say what the prosecution claimed, was closed in its Watali form. A defence argument had to work within the totality and at face value. It could submit that even the whole record, accepted as presented, did not connect this accused to the accusation, and later jurisprudence would have to explore how far that submission could go, but it could not invite the court to disbelieve statements, to exclude material as inadmissible, to treat protected witness statements as worthless, or to resolve competing readings of a document in favour of the accused.

It is equally essential to state, with the same precision, the limits of what Watali decided, because the authority of a judgment extends to the questions it decided and not to questions that were never placed before the court. Watali was a decision about the approach under the proviso on the merits. The appeal challenged a High Court order that had granted bail after a close examination of the material, and the Supreme Court answered that challenge by prescribing the method for forming the statutory opinion on that material. The judgment did not address prolonged detention. It did not address Article 21. That question was not before the Court. No argument founded on years of undertrial custody, on a trial that could not conclude within a reasonable time, or on the constitutional limits of keeping an unconvicted person in prison while the state completed a lengthy prosecution, was decided by the bench either in favour of the prosecution or in favour of the accused. The silence of Watali on that subject is therefore not a rejection of the Article 21 argument. It is an absence, and treating the judgment as though it had settled a question it never heard would be an error of reading as well as an error of law.

That distinction, between the question decided and the question left open, is the hinge on which the later development of this area turns. On its own ground, Watali is a restrictive baseline of great force. It tells every bail court that the chargesheet and the case diary are to be taken as a whole and at face value, that evidence is not to be filtered or weighed item by item, that protected witness statements are not to be discarded, and that the satisfaction required is a light satisfaction on broad probabilities. Within an application that asks the court to release the accused because the prosecution material is weak, Watali will usually supply the prosecution with its answer. Outside that ground, the judgment has nothing to say. It does not measure custody against time. It does not ask whether a trial with a very large witness list can finish soon enough to keep continued detention within constitutional bounds. It does not decide what happens when the delay ceases to be an incidental hardship and becomes the central fact of the case. Those questions remained where the quiet decade had left them, unanswered by the baseline judgment, awaiting a case in which they were squarely raised and could not be avoided.

Najeeb: the constitutional track opens

For a long period after Watali, the bail enquiry under the Unlawful Activities (Prevention) Act appeared to move along a single track, and that track was narrow by design. A court asked whether the accusation, viewed on the material collected by the investigating agency, was prima facie true, and if that threshold was crossed, the statutory restriction in Section 43D(5) held the door closed. Delay, personal circumstance, and the length of custody already undergone struggled to find a place inside that enquiry, because the proviso directed attention to the accusation and its supporting material rather than to the condition of the accused person waiting for trial. Union of India v K. A. Najeeb, (2021) 3 SCC 713, did not remove that track, and a careful reading of the judgment shows a bench determined not to be understood as removing it. What the decision did, instead, was to identify a second track running alongside the first, grounded in a different source of power and addressed to a different question, and to explain why the existence of the statutory restriction did not foreclose a constitutional court from travelling along it.

The decision was delivered on 1 February 2021 by a three judge bench of Justices N. V. Ramana, Surya Kant and Aniruddha Bose, with Justice Surya Kant authoring the judgment. Its procedural posture matters to its authority and to its limits. The Kerala High Court had granted bail on 23 July 2019, and the Union of India appealed against that grant. The Supreme Court was therefore not considering an application in the first instance, in the abstract, or on a hypothetical record of delay. It was reviewing a considered order of a constitutional court that had already exercised its jurisdiction in favour of release, and the question before the bench was whether the statutory scheme had made that exercise impermissible. That posture colours the reasoning throughout, because the judgment defends a jurisdiction already invoked by a High Court rather than announcing a free standing entitlement to bail in every prosecution under the Act, and it defends that jurisdiction by locating it outside the proviso altogether rather than by diluting the test that the proviso prescribes.

The human facts pressed hard against any purely abstract answer. K. A. Najeeb had spent more than five years in custody as an undertrial. His case had been separated from that of his co accused, whose trials had been completed. Those co accused had received sentences not exceeding eight years. The comparison embedded in those facts is difficult to evade, and the bench did not evade it. A man whose guilt had not been established had already remained in prison for a period that formed a substantial part of the longest sentence actually imposed on persons tried and convicted in the connected proceedings, and his own trial, separated from theirs, offered no assurance of an early conclusion. The complaint was about the duration and open ended character of detention without adjudication, and about whether the State could continue to hold a person on the strength of an accusation that the legal process was not bringing to a timely determination.

Stated narrowly, and the narrowness is deliberate, the holding in Najeeb is that the presence of statutory restrictions like Section 43D(5) does not by itself oust the ability of constitutional courts to grant bail on grounds of violation of Part III of the Constitution. Every element of that formulation carries weight. The bench spoke of statutory restrictions like Section 43D(5), placing the provision within a familiar legislative pattern rather than treating it as a unique code of exclusion. It spoke of constitutional courts, not of every court seized of a bail application, and that choice of forum becomes central to the architecture examined later in this part of the article. It spoke of grounds of violation of Part III, which means the release, when it comes on this track, rests on a fundamental rights violation and not on a more generous reading of the statute. And it said that the presence of the restriction does not by itself produce the ouster, which leaves intact the ordinary expectation that the restriction will govern the statutory enquiry for which Parliament enacted it.

Justice Surya Kant, writing for the bench, presented statutory restrictions and constitutional jurisdiction as capable of being harmonised, operating in different registers. Harmonisation is a conservative judicial technique, and its use here signals the method of the whole decision. A court that harmonises two sources of power is not declaring one void in favour of the other, and it is not rewriting either to avoid a conflict that it has first exaggerated. It is reading each within its proper field so that both can operate. On that reading, Section 43D(5) continues to express a legislative judgment about bail at the stage where the accusation is assessed, and the constitutional jurisdiction continues to protect the detained person against a detention that, through lapse of time and absence of a foreseeable conclusion, has ceased to be compatible with Part III. Neither power needs to swallow the other for the judgment to work, and the bench was careful to preserve work for both to do.

The temporal dimension of that harmonisation received its most quoted expression in the passage dealing with the commencement of proceedings. At the commencement of proceedings, the bench held, courts are expected to appreciate the legislative policy against grant of bail. The sentence acknowledges, without qualification, that Parliament had a policy, that the policy was against the grant of bail in this class of cases, and that courts at the early stage are expected to give that policy real effect rather than treat it as a formality to be recited before release is ordered on general grounds. The judgment then adds the qualifying movement for which it is remembered: the rigours of such provisions melt down where there is no likelihood of the trial being completed within a reasonable time and the period of incarceration already undergone has exceeded a substantial part of the prescribed sentence. The metaphor of melting is gradual rather than explosive, and it fits the logic of the reasoning. Nothing in the provision breaks, and nothing is struck down. Its rigour lessens as the factual premises that might justify rigorous pre trial detention fall away, namely a trial in prospect within a reasonable time and a period of custody that remains proportionate to the sentence the law prescribes for the offence charged.

Both limbs of that qualifying condition deserve close attention, because later argument has often tried to detach one from the other. The first limb looks forward: there is no likelihood of the trial being completed within a reasonable time. The second limb looks backward: the period of incarceration already undergone has exceeded a substantial part of the prescribed sentence. A long period in custody, without more, is not the whole test on the language the bench used, and neither is a gloomy forecast about the trial calendar, without more. The constitutional anxiety in Najeeb arose from the conjunction, from years already served joined to a trial that could not be expected to conclude within a reasonable time, so that continued detention threatened to become punishment in substance while remaining accusation in form. Read in that conjunctive way, the passage is less a licence than a threshold, and a demanding one, requiring the applicant to establish both a past that has already consumed a substantial part of the prescribed sentence and a future in which adjudication remains out of reasonable reach.

Equally instructive is the catalogue of things Najeeb did not do, because much later confusion has come from attributing to the judgment a boldness its authors withheld. The bench did not strike down Section 43D(5). The provision remained valid law after the decision, available to govern statutory bail applications in the ordinary way, and no part of the judgment depends on its invalidity. The bench did not overrule Watali. The prima facie approach associated with Watali continued to describe how the statutory question is to be approached, and Najeeb neither disowned that approach nor instructed courts to conduct a deeper merits assessment under the cover of constitutional reasoning. The bench also did not lay down any fixed period of custody after which bail follows. There is no number of years in the judgment that operates as a trigger, no calendar line on one side of which detention is statutory and on the other side of which it is unconstitutional, and no suggestion that an accused person need only count days in custody to earn release. The absence of a fixed period is not an oversight in a judgment so carefully framed. It is a consequence of the constitutional method itself, which assesses delay, custody, and the prospect of trial on the facts of a given prosecution rather than by a universal rule announced in advance.

What was distinctive about K. A. Najeeb’s own case?

K. A. Najeeb’s trial had been separated from that of his co-accused, whose trials had been completed and who had received sentences not exceeding eight years. He had by then spent more than five years in custody as an undertrial, a period approaching the maximum terms actually imposed on those already tried and convicted.

Each strand of that distinctiveness does separate work in the reasoning, and the strands should not be blurred into a general impression of hardship. The separation of his trial comes first in analytical importance, because it altered the relationship between his detention and any adjudication of his own case. He was not simply one accused among many in a single proceeding moving, however slowly, toward a common conclusion. His case had been separated from that of his co accused, which meant that the completion of their trials did not complete, advance, or even necessarily illuminate his own. The process that might have justified his continued detention by bringing his guilt or innocence to determination had divided, and he remained in custody on the separated side of that division. A court examining delay in such a posture is not confronted with the ordinary complaint that a joint trial is large and slow. It is confronted with detention attached to a proceeding that has been structurally detached from the proceedings that did reach judgment.

The completed trials of the co accused do a second kind of work, because they supply a concrete measure against which his undertrial custody can be assessed without speculation. Courts asked to decide whether detention has lasted too long often lack a reliable comparator, since the sentence that might follow conviction remains hypothetical until a trial court pronounces it. In Najeeb, the connected proceedings had produced actual sentences, and those sentences did not exceed eight years. The bench was therefore not required to guess at the punitive horizon of the case from the maximum prescribed in the abstract. It could observe what the legal process, operating on the connected allegations against the co accused, had in fact imposed at its highest point. It did, however, give practical content to the phrase substantial part, because more than five years in custody could be seen in proportion to sentences of not more than eight years imposed after trial in the separated proceedings.

The third strand is the sheer length of custody already undergone, more than five years as an undertrial, and its significance lies in the status that accompanied those years. An undertrial is detained on accusation, not on a judicial finding of guilt, and the constitutional tolerance for such detention depends heavily on the State bringing the accusation to trial. Five years is not treated in the judgment as a magic figure, and it would misread the bench to convert the facts of this case into a rule for every other case. Its force came from combination. More than five years had passed, the trial had been separated, the co accused had been tried and sentenced to terms not exceeding eight years, and there was no likelihood that his own trial would be completed within a reasonable time. Remove any one of those elements and the case becomes materially weaker, which is precisely why the bench could decide as it did without laying down a fixed period. The facts formed a constellation in which detention had outgrown the procedural justification offered for it, and the judgment is best understood as a response to that constellation rather than to delay in the abstract.

Seen in that light, the Kerala High Court order of 23 July 2019 and its affirmance on appeal acquire a significance beyond the release of one man. The High Court had been willing to treat the accumulated facts as constitutionally relevant, and the Supreme Court, in upholding the grant, confirmed that such treatment was legally available. The sequence matters for undertrial litigation across the country, where individual cases rarely attract attention yet together reveal how the enforcement record of India’s anti-terror laws is shaped by years spent in custody before any verdict is reached. A single grant of bail changes one life. A reasoned affirmance by a three judge bench changes the arguments that can responsibly be made in the next case, because counsel can point not merely to sympathy or discretion but to an identified constitutional jurisdiction that a statutory restriction does not by itself oust.

Two registers, one accused

The enduring contribution of Najeeb lies less in its outcome than in the architecture it articulated, an architecture of two registers in which the same accused person can be the subject of two different legal questions that must not be confused with one another. The first register is statutory. Within it, the court asks whether the accusation is prima facie true on the chargesheet material, applying the approach that governs Section 43D(5) and respecting the legislative policy that the bench in Najeeb expressly acknowledged. The question is accusation centred. It looks at the material gathered by the prosecution, takes that material at the level the statute and precedent require, and decides whether the threshold for withholding bail has been crossed. The personal history of detention enters this register, if at all, only at the margins, because the enquiry is directed to what is alleged and to what the collected material shows about the allegation.

The second register is constitutional. Within it, the court asks a question that the chargesheet cannot answer, namely whether detention itself has become a violation of Article 21 through delay. That question is detention centred rather than accusation centred. It looks at the period already spent in custody, at the likelihood of the trial being completed within a reasonable time, at the relationship between custody undergone and the sentence prescribed, and, on facts such as those in Najeeb, at complications like a separated trial and completed proceedings against co accused. A strong accusation, assumed for the purpose of the statutory enquiry, does not by itself answer the constitutional question, because the complaint in the second register is not that the accusation is weak. The complaint is that the State, whatever the strength of its accusation, has held the person for so long without bringing him to trial that the detention can no longer be reconciled with Part III.

Keeping those propositions apart sounds elementary, yet much of the difficulty in this field comes from allowing the answer in one register to do duty in the other. If a court treats a prima facie true accusation as a complete answer to an Article 21 complaint, the constitutional track identified in Najeeb disappears, because every applicant who fails the statutory test will fail the constitutional test for the same reason, and the harmonisation the bench described will have produced no practical space in which fundamental rights can operate. If, on the other hand, a court treats delay as a reason to re examine the accusation with a softer eye, the statutory track is quietly displaced, because the prima facie enquiry becomes a vehicle for considerations that belong elsewhere and Watali is eroded without ever being confronted. Najeeb avoids both errors by separation rather than by compromise. The statutory question remains statutory, governed by its own materials and threshold. The constitutional question remains constitutional, governed by time, custody, and the prospect of adjudication. The accused stands at the intersection of the two, one person subject to both enquiries, and that is the sense in which the heading of this section speaks of two registers and one accused.

The identity of the court asking the question is part of the architecture and not a procedural detail to be added afterward. A Special Court applying the proviso and a High Court exercising constitutional jurisdiction are not doing the same job, even when both orders concern bail and both are made in the same prosecution. The Special Court, in applying Section 43D(5), performs the statutory task assigned to the trial court structure, assessing the accusation on the chargesheet material under the restriction Parliament enacted. A High Court that intervenes on Part III grounds performs a different task, rooted in its constitutional authority to protect fundamental rights against detention that has become unlawful through the passage of time and the failure of timely trial. To demand that the High Court confine itself to the proviso’s question would be to deny the very ouster point decided in Najeeb, since the bench held that the statutory restriction does not by itself remove the constitutional ability. To expect the Special Court, in its ordinary application of the proviso, to replicate the full constitutional enquiry in every bail application would be to misunderstand where Najeeb placed the second track. The judgment speaks throughout of constitutional courts, and that language allocates responsibility within the judicial hierarchy rather than distributing the new doctrine evenly across every forum.

This allocation also explains why the same application history can legitimately produce different outcomes at different moments without any court having contradicted itself. At the commencement of proceedings, the statutory register predominates in practical terms, because the legislative policy against bail is at its strongest, custody has been short, and an Article 21 claim founded on delay will ordinarily have no factual foundation. As years pass without a trial concluding, or without a trial even becoming likely to conclude within a reasonable time, the constitutional register acquires material on which to operate. The accusation may not have changed at all, and the chargesheet material may be exactly what it was at the beginning. What has changed is the detention, and a court that grants bail at that later point on constitutional grounds is saying that continued detention, assessed on accumulated facts, has crossed a line that had not earlier been crossed. Time is therefore not background in this architecture. Time is the medium in which the second register comes into being.

Bench strength gives this architecture a further dimension that lawyers ignore at their peril. Najeeb is a decision of a three judge bench. Watali is a decision of a two judge bench. In the discipline of precedent, that difference in strength matters, because later benches, and in particular later two judge benches, were bound to work within both decisions rather than choose between them as if they were rival manifestos competing for allegiance. A two judge bench confronted with an undertrial application after Najeeb could not responsibly discard Watali by treating Najeeb as having overruled it, for the simple reason that Najeeb itself had done no such thing. Equally, such a bench could not discard Najeeb by treating Watali as an exhaustive code, because a decision of a larger bench had held that the statutory restriction does not by itself oust constitutional jurisdiction. The task for later benches was therefore synthetic and constrained: to preserve the prima facie discipline of the statutory register while giving real, if carefully bounded, effect to the constitutional register that a larger bench had identified. Much of the case law that follows in this article is best read as a series of attempts to perform exactly that synthetic task under the discipline that bench strength imposes.

Najeeb settled that constitutional courts are not ousted by Section 43D(5) from granting bail on Part III grounds, that the two jurisdictions can be harmonised, and that the rigours of the statutory restriction melt down where trial within a reasonable time is not likely and custody undergone has exceeded a substantial part of the prescribed sentence. What it did not settle presses forward into the rest of the enquiry. How long is long enough, in custody and in the shadow of an unfinished trial, for the constitutional register to prevail in a case less stark than Najeeb’s own? By what measure is a substantial part of a prescribed sentence to be judged where no co accused has been tried to provide a comparator? And could the pressure generated by prolonged detention eventually soften the merits approach within the statutory register itself, rather than merely supplementing that register from outside through constitutional jurisdiction? Najeeb leaves each of those questions open, deliberately and necessarily, because its method was to establish a track rather than to map every distance along it. The answers, insofar as the later case law supplies them, belong to the segments that follow.

The merits track under Section 43D(5) of the Unlawful Activities (Prevention) Act had, after Watali, acquired a reputation for rigidity, resting on the posture the judgment prescribed for the bail court, a posture of acceptance toward the material collected by the investigating agency and of restraint in testing that material for worth. A court told to treat the chargesheet as prima facie true for the purpose of the statutory bar had little room to ask whether a described act answers the description in the statute, or whether a connection asserted between a person and an offence is a connection in law or only in narrative. The softening of that track did not arrive as a repudiation. No larger bench reconsidered Watali, and no judgment in this sequence declared the earlier reasoning wrong in terms. The change arrived through the ingredients of the offences themselves, through the discipline with which those ingredients were read, and through an insistence that the statutory bar has a field of operation that must be established before its rigour can be applied. A bar that is never attracted leaves nothing to override, and a test that requires a surface examination of worth can fail on material that would have survived a test of mere acceptance.

The ingredient question: Thwaha Fasal and the mental element

Thwaha Fasal v Union of India, decided on 28 October 2021 by a two judge bench of Justices Ajay Rastogi and Abhay S. Oka, approached the bail question from the side of the offence rather than from the side of procedure. The accused were alleged to be associated with the Communist Party of India (Maoist), and the prosecution case rested in substance on that association and on material said to demonstrate support, from which offences under Sections 38 and 39 of the Act were sought to be made out. Section 38 concerns membership of a terrorist organisation and allied conduct, while Section 39 concerns support given to a terrorist organisation, and both provisions sit within the chapters to which Section 43D(5) applies, so that if the accusations were accepted in the frame presented, the bar on bail would have appeared in its usual formidable form. The Court declined to accept the frame without examination, and it placed the mental element at the centre of the inquiry.

The holding, stated narrowly, was that mere association with a terrorist organisation is not sufficient to attract Section 38, and mere support is not sufficient to attract Section 39, of the Unlawful Activities (Prevention) Act. The qualification that followed is the operative part of the judgment. The association or support must be with the intention of furthering the activities of the organisation, and that intention cannot be presumed from the fact of association alone. Intention has to be capable of being inferred, and the route to inference recognised by the Court was through overt acts or acts of active participation borne out by the chargesheet material. The formulation releases the court from a purely status based reading of Sections 38 and 39, while specifying the kind of material that could sustain the charge, material showing a person doing something in a manner that points toward furtherance of the organisation and its activities, rather than material showing only proximity, sympathy, or acquaintance. On the material in that case there was no prima facie showing of active participation of that character, and the foundation for invoking the bar in its strict form had therefore not been laid.

Precision about that sequence protects the judgment from two opposite misreadings. Thwaha Fasal is not a general licence to reappraise prosecution material at bail, and it is not a decision turning on sympathy for particular accused persons. Its reasoning is directed to the elements of defined offences. The first question is textual, whether Sections 38 and 39, as construed by Justices Rastogi and Oka, require intention directed toward furthering activities, so that association simpliciter and support simpliciter fall short of the offence as defined. The second question is evidential in the limited sense appropriate to a bail hearing, whether that intention can be shown at the prima facie stage by overt acts or active participation that the chargesheet material itself bears out, a formulation that neither demands proof to a trial standard nor permits intention to float free of any act to which it can be attached. The third question is consequential, whether a court that finds the ingredients undisclosed is granting bail in defiance of the bar or holding that the condition for the bar has not been satisfied. The judgment adopts the latter course.

The judgment also confirmed two connected points about the field of Section 43D(5). It confirmed that Section 43D(5) applies only to offences under Chapters IV and VI, so that an accusation under Section 13 alone, Section 13 being situated in Chapter III, does not attract the bar, however seriously the underlying conduct described as unlawful activity may be regarded. That confirmation matters where a chargesheet mixes accusations across chapters, because it requires the court to attend to the chapter location of each accusation and to resist treating the Act as a single undifferentiated code for bail purposes, when the legislature has distributed its offences across chapters and has attached the special restriction to only some of them. The judgment further reiterated the position in Union of India v K. A. Najeeb, that the bar does not prevent a constitutional court from granting bail on grounds grounded in Part III of the Constitution. That reiteration kept the delay and rights route open alongside the ingredient route, even though the principal contribution of Thwaha Fasal itself lies elsewhere.

The analytical point is that the lever applied in Thwaha Fasal is different in kind from the lever applied in Najeeb. Najeeb works where the bar is present and applicable, and it holds that constitutional courts retain a power to grant bail notwithstanding that bar where continued incarceration offends Part III, most prominently the protection in Article 21. The inquiry there is external to the offence, turning on time spent in custody, on the pace of the trial, and on proportionality between detention awaiting trial and the process that detention awaits. Thwaha Fasal works at an earlier point and asks a prior question, whether the bar is attracted at all, because the charge as framed may not disclose the offence the bar presupposes. The inquiry there is internal to the offence, turning on the text of Sections 38 and 39, on the mental element those sections require as construed by the bench, and on whether the chargesheet material bears out overt acts or active participation from which the requisite intention might be inferred. A court following Najeeb concedes, for the purpose of argument, the statutory obstacle and then finds a constitutional path around it, while a court following Thwaha Fasal examines the obstacle and finds, in a given case, that it was never placed in the path at all. The ingredient route therefore does not depend on a long period in custody or on a showing about the likely duration of a trial. The two routes are cumulative in that both should be considered, and alternative in that either may succeed where the other fails.

The limits in the holding must be preserved. The judgment did not hold that association with a proscribed organisation is legally insignificant, or that support for such an organisation falls outside the Act, or that intention must be proved at bail to a trial standard. It held that association or support must be shown, through the chargesheet material, to be accompanied by the intention of furthering the activities of the organisation, inferable from overt acts or active participation.

The High Court experiments of 2021

The same calendar year produced a second and quite different set of experiments, conducted not in the Supreme Court but in a High Court exercising constitutional jurisdiction over bail in a case of exceptional public attention. On 15 June 2021 the Delhi High Court granted bail to Asif Iqbal Tanha, Devangana Kalita and Natasha Narwal in the case concerning the conspiracy alleged in connection with the Delhi riots. It did so by reading the terrorist act provisions narrowly and by examining the chargesheet material closely rather than accepting it wholesale. The orders attracted intense scrutiny, in part because of the prominence of the underlying events, in part because the High Court distinguished protest, even protest characterised by the prosecution in grave terms, from the specific statutory conception of a terrorist act, and in part because the method of close examination sat in evident tension with the method Watali had prescribed. Where Watali had emphasised acceptance of the prosecution material as a whole, the Delhi High Court judgments proceeded by parsing allegations, testing them against the statutory definitions said to be attracted, and asking whether the material answered those definitions at the prima facie level.

The sequel in the Supreme Court must be described with the same care as the orders themselves, because the sequel defines what the episode does and does not stand for. The Supreme Court declined to interfere with the bail orders, with the result that Asif Iqbal Tanha, Devangana Kalita and Natasha Narwal stayed on bail. At the same time, the Supreme Court directed that those judgments shall not be treated as precedent. That direction quarantined the reasoning, withholding from it the authority a judgment ordinarily carries for other courts and other cases, while leaving undisturbed the operative orders. A bail order that survives without its reasoning becoming law occupies a precarious place, effective for the persons it concerns and instructive as an instance of what a constitutional court considered justified on a particular record, yet formally unavailable as a proposition upon which later courts may rest. To describe the Delhi High Court judgments as having settled the construction of the terrorist act provisions would be wrong, and to describe them as having been reversed in the sense of the accused being returned to custody would be equally wrong. An accurate account holds two facts together that pull in opposite directions, release sustained and reasoning quarantined.

The episode is valuable not because it supplies a governing precedent, for by direction it does not, but because it throws into relief a divergence among High Courts in this field. High Courts are constitutional courts, equipped with the powers that status confers, including the power recognised in Najeeb to grant bail on Part III grounds notwithstanding the statutory bar, and they hear the great majority of contested bail applications under the Act. On the question how closely chargesheet material may be examined at the bail stage, those courts pulled in different directions. Some treated Watali as requiring an approach close to acceptance of the material in the form presented, others treated the prima facie inquiry as permitting and at times requiring a closer look at whether the material answered the ingredients of the offences charged, and others located the resolution in delay and rights analysis rather than in examination of the merits. The Delhi judgments of June 2021 represented one end of that spectrum, marked by close examination and narrow construction, and the direction that followed them represented not a choice of the opposite end but a refusal to let the position taken become the governing position, a refusal that preserved the divergence it declined to resolve into precedent.

That divergence is one source of the inconsistency complaint that recurs in commentary on this area and in submissions of counsel who practise within it. An accused faces, in principle, the same statute and the same Supreme Court authorities in whichever State the prosecution is brought, yet the practical content of the prima facie test, the willingness of the court to interrogate the fit between allegation and ingredient, and the readiness of the court to travel the constitutional route may each differ according to the High Court before which the application falls to be heard. Divergence of that kind is not unique to this statute, and a judicial structure with multiple High Courts will always produce a measure of it, particularly where the governing authority prescribes a posture in general terms and leaves its application to records that differ markedly. The complaint acquires particular force in the UAPA context from the stakes attaching to the bail decision, since refusal of bail under a regime in which trials may extend over long periods shapes years of an accused person life, and because the divergence concerns the central question what a court is permitted to do with the material on which the State asks it to keep a person in custody. The experiments of 2021 made that divergence visible and gave it a concrete illustration.

A further feature explains part of the caution with which later courts have approached those judgments. The narrow reading in the Delhi cases operated, like the ingredient analysis in Thwaha Fasal, at the level of fit between conduct alleged and offence defined, but it operated on different provisions and on a far more elaborate factual narrative, and in reasoning of a length and detail that invited the objection, heard in the proceedings that followed, that the bail court had conducted something approaching a trial of the conspiracy case upon the chargesheet. Whether that objection was well founded is not a question the quarantine direction answered, since the direction withheld precedential status without interfering with the orders, leaving later courts without either an endorsement or a repudiation of the method in terms that bind them. The result is a body of reasoning that continues to be read and cited in argument while formally standing outside the line of authority. In the architecture of this article the episode belongs between the ingredient question in Thwaha Fasal and the surface analysis Vernon would later require, as evidence that pressure against wholesale acceptance was building in the constitutional courts before the Supreme Court supplied a formulation that could be applied within Watali.

Is possessing radical literature enough to refuse UAPA bail?

No. Per Vernon, mere possession of literature, even if its content propagates violence, does not by itself constitute an offence within Chapters IV and VI of the Act, and such possession, standing alone, cannot supply the prima facie foundation that Section 43D(5) requires for refusing bail.

That distinction, between inclination and offence, is the bridge to Vernon. A text possessed is not, without more, an overt act of active participation in a terrorist organisation, and a proposition found in a book is not, without more, a terrorist act committed by its possessor. Vernon resolved the literature question on a record where letters and literature featured prominently, and its reasoning altered the merits track well beyond books and papers.

Vernon: surface analysis enters the merits track

Vernon v State of Maharashtra, reported as 2023 INSC 655 and decided on 28 July 2023 by a two judge bench of Justices Aniruddha Bose and Sudhanshu Dhulia, is the judgment in which the softening of Watali on the merits received its most considered formulation. The companion appellant was Arun Ferreira, and the appellants were accused in the Bhima Koregaon case, a prosecution involving allegations of association with a proscribed organisation and of participation in a wider design presented through documents, correspondence, and material recovered during investigation. The bail question reached the Supreme Court against the background of Watali, and the bench did not disown that background. Its holding, stated narrowly, was that the prima facie test described in Watali is not satisfied unless there is at least a surface level analysis of the probative value of the evidence, and that the quality or probative value of the material must satisfy the court of its worth. Bail was granted to the appellants on the application of that test to the material relied upon against them.

Each element of that formulation does distinct work. The first element is the location of the requirement within Watali itself, because Justices Bose and Dhulia presented surface analysis not as an addition imposed from outside but as a condition of the earlier test being satisfied at all, so that a court omitting the analysis had failed to complete the inquiry Watali, properly understood, demanded. The second element is the level of analysis, described deliberately as surface level, a phrase marking both a permission and a boundary, permission to examine probative value rather than merely to register the existence of material, and a boundary against converting the bail hearing into a full evaluation of credibility or of the ultimate weight a trial court might attach after evidence has been tested. The third element is the object of analysis, the probative value or quality of the evidence, which directs attention away from volume and away from the seriousness of labels, and toward what the material tends to prove about the ingredients charged against the particular accused. The fourth element is the standard, the court being satisfied of the worth of the material, a standard remaining within the prima facie register, since such satisfaction is not proof of guilt, but requiring the court to form and articulate a judgment rather than merely to record allegations.

Applied to the record before it, that discipline produced findings illustrating its operation. The letters relied upon against the appellants were in the nature of hearsay, they were recovered from a co accused, and they attributed no covert or overt terrorist act to the appellants. Each feature diminished probative value in a different way, hearsay character affecting the directness with which the material spoke to the appellants conduct, recovery from a co accused affecting the connection between the material and the appellants themselves, and absence of attribution of any covert or overt terrorist act affecting the fit between the material, even taken at its highest, and the offences the bar presupposes. A court confined to wholesale acceptance might have registered the letters as material collected during investigation and treated the prima facie test as satisfied by their existence in the chargesheet, while a court undertaking the surface analysis required by Vernon had to ask what the letters were, where they had come from, and what they actually attributed to the persons seeking bail, and on asking those questions the material did not satisfy the Court of its worth. The judgment also held that mere possession of literature cannot by itself constitute an offence within Chapters IV and VI.

The relationship between Vernon and Watali is the point on which the most careful statement is required. Vernon did not overrule Watali, and it purported to apply Watali, a self description the bench maintained throughout and that later courts are bound to respect in characterising the authority. No declaration was made that the earlier test was wrongly framed, no reference to a larger bench was sought, and no substitute test was promulgated displacing the language of prima facie truth upon which Watali had insisted. What changed was the discipline of the inquiry, the set of operations a court must perform before it can say the test has been satisfied, and the direction of change was from acceptance of the material as a whole to a surface check of what the material is worth. That is not a trivial change, because in a field where decisive contests are fought at the level of method, over what a judge may look at and what questions a judge may ask, a change in required discipline is a change in the law as it operates upon the person in custody. Equally, the change should not be exaggerated into a permission Vernon withheld, since the surface check remains a check at the surface, undertaken upon the material as presented for the limited purpose of the bail bar.

It is for that reason that Vernon represents the most significant softening of Watali on the merits rather than on delay. The delay route leaves the merits test exactly where it found it, and grants relief for reasons independent of whether the prosecution material is strong, weak, direct, or hearsay in character, so that an accused unable to show the requisite period of custody or prospect of protracted detention gains nothing from it, however thin the material may be. Vernon operates within the merits inquiry itself, and it therefore assists, in principle, at an earlier stage of custody and on a ground going to the justification for applying the bar in the first place, namely that the material has not been shown, even at the surface level the prima facie standard permits, to be worth what the bar requires. Read together with Thwaha Fasal, the softened merits track becomes intelligible as a sequence of questions a bail court can ask in order, first whether the accusations fall within the chapters to which Section 43D(5) applies, second whether the material bears out the ingredients of the offences charged, including the mental element in Sections 38 and 39 and the active participation from which it may be inferred, and third whether the material, subjected to surface analysis, has the probative value and quality to satisfy the court of its worth, with hearsay letters recovered from another person and attributing no terrorist act to the accused, and literature merely possessed, standing as illustrations of material that may fail at the second or third question.

2024: the two lines collide and coexist

By the opening months of 2024 the bail jurisprudence under the Unlawful Activities (Prevention) Act, 1967 had developed along two lines that pulled in different directions, and the year did not choose between them. One line, associated with the strict reading of the proviso to Section 43D(5) and with Watali, treated the statutory bar as the governing discipline: where the accusation, taken on the prosecution material, was prima facie true, bail was to be refused. The other line, associated with K. A. Najeeb, treated prolonged custody and an unending trial as a distinct constitutional problem, holding that a constitutional court retained power under Article 21 to grant bail even where the statutory test was not satisfied in the ordinary way. The four decisions of 2024 considered here confirm that both lines remain alive, that each operates within limits identified by the benches that applied it, and that the outcome may turn as much on the state of the trial, the quality of the chargesheet material, and the length of custody already undergone as on any abstract statement of the test. Gurwinder Singh v State of Punjab, (2024) 5 SCC 403, decided in February 2024 by a two judge bench of Justices M. M. Sundresh and Aravind Kumar, was the first of the four. The appellant, according to the prosecution, was linked to the Sikhs for Justice organisation. He had been in custody for over five years, the trial was under way, and witnesses had been examined. Bail was refused. The Court restated a two prong approach: whether the allegations make out an offence attracting the bar, and, if the accusation is prima facie true, bail must be rejected as a rule. Stated in that sequence, the test leaves little room for a balancing exercise once the second prong is satisfied. The bench distinguished K. A. Najeeb on the footing that there the trial had been separated and the co accused had been tried and sentenced to terms not exceeding eight years, while here the trial was progressing. It also held that mere delay in trial is no ground for bail in grave offences. That proposition, read in its factual setting, was directed at delay asserted as a standalone reason, detached from any showing that the trial had stalled or that the material itself failed at the prima facie stage. Whether later benches would read it more narrowly became the immediate question, because applicants who had spent comparable or longer periods in custody continued to press Article 21.

Shoma Kanti Sen v State of Maharashtra, decided on 5 April 2024 by a two judge bench of Justices Aniruddha Bose and Augustine George Masih, with Justice Bose authoring the judgment, supplied the first part of the answer. The applicant was a former professor at Nagpur University and an accused in the Bhima Koregaon case. She had been in custody for nearly six years without charges being framed, a chronology that placed delay in a very different posture from a trial actively progressing with witnesses under examination. Bail was granted on two independent footings. The first was on the merits under the statute itself: the material did not disclose a prima facie case against her within the proviso to Section 43D(5). That holding locates the grant inside the statutory scheme: if the accusation is not prima facie made out on the material as it relates to the particular accused, the bar does not require refusal.

The second footing in Shoma Sen was constitutional, and the Court expressed it in terms that directly addressed the argument pressed against bail. The prosecution had submitted that bail is not a fundamental right. The Court rejected that submission, holding that the Court had already accepted the right of an accused under the 1967 Act to be enlarged on bail founded on Article 21, and that any deprivation of liberty must be reasonable, follow a just and fair procedure, and be proportionate in the facts of the case. The Court then dealt expressly with Gurwinder Singh, treating it as a judgment on its own facts which did not dislocate the axis of reasoning on constitutional ground in Najeeb. The judgment also noted the delay precedents cited to it, Shaheen Welfare Association v Union of India (1996) 2 SCC 616 and Angela Harish Sontakke v State of Maharashtra (2021) 3 SCC 723, each decided on its own facts. That qualification prevents a mechanical rule in either direction, whether a fixed period in custody automatically producing bail or the seriousness of the charge automatically defeating a claim founded on delay. What remains is an evaluative inquiry by a constitutional court on a particular record, into whether continued detention remains reasonable, just, fair, and proportionate.

Sheikh Javed Iqbal @ Ashfaq Ansari @ Javed Ansari v State of Uttar Pradesh, 2024 INSC 534, decided on 18 July 2024 by a two judge bench of Justices J. B. Pardiwala and Ujjal Bhuyan, with Justice Bhuyan authoring the judgment, took the constitutional line further on a starker record. The appellant was a citizen of Nepal, arrested in February 2015 at the India and Nepal border with counterfeit currency of over 26 lakh rupees, according to the prosecution. He was charged under the Indian Penal Code and under the UAPA with damaging the monetary stability of India. The Allahabad High Court had refused bail on 3 April 2023. By July 2024 he had been in custody for nine years and only two witnesses had been examined. Bail was granted. The Court held that a constitutional court cannot be restrained from granting bail where the Article 21 right to a speedy trial is infringed, and it described Article 21 as overarching and sacrosanct. It held further that, where a trial is prolonged, the prosecution cannot oppose bail only on the ground that the charges are serious, and bail cannot be denied on seriousness alone when there is no end in sight. The reasoning identifies the vice that the Article 21 track addresses: the conversion of pre trial custody into an open ended punishment by delay, in a proceeding that has not been brought to a conclusion within a reasonable time. Seriousness remains relevant to the statutory inquiry and to the conditions that may be imposed on release, but it cannot perform unlimited work. The nine years were therefore not treated as a number that triggered a rule. Together with the examination of only two witnesses, the period was treated as evidence that the trial had been prolonged to a point at which opposition founded solely on gravity had ceased to be a sufficient constitutional answer.

Jalaluddin Khan v Union of India, decided on 13 August 2024 by a two judge bench of Justices Abhay S. Oka and Augustine George Masih, with Justice Oka authoring the judgment, approached the problem from the merits side and from a restatement of first principles. The appellant was a retired police constable from Bihar who had rented the first floor of his house to alleged members of the Popular Front of India. Bail was granted. The Court held that the maxim that bail is the rule and jail is an exception is settled law and applies even under special statutes with stringent bail conditions, with the modification that bail can be granted if the statutory conditions are satisfied. It held further that, once a case for bail is made out, a court cannot decline to grant it, and denying bail in deserving cases violates Article 21. That proposition links the statutory and constitutional analyses in a manner distinct from Najeeb, where the constitutional power was invoked to overcome the bar in circumstances of delay. In Jalaluddin Khan, refusal in a case that satisfied the statutory conditions was itself characterised as a violation of Article 21. On the facts, the chargesheet contained nothing showing that he had taken part in unlawful activities, a witness statement had been distorted in the chargesheet, and the trial had made no progress. The absence of material on participation undermined the prima facie case against this accused, as distinct from any case that might exist against others, and the distortion undermined confidence in the chargesheet as a reliable basis for the prima facie assessment. The four decisions therefore do not form a ladder in which each overrules the last. Gurwinder Singh reasserts the statutory rule where the trial is progressing. Shoma Sen preserves the constitutional reasoning of Najeeb and applies the proviso on its merits after nearly six years without charges being framed. Sheikh Javed Iqbal gives the speedy trial guarantee its strongest formulation on a record of nine years and two witnesses. Jalaluddin Khan restores the bail is the rule maxim to special statutes and scrutinises the chargesheet for distortion. A lawyer who cites only one of the four misstates the law that emerged from the year.

Is bail impossible under the UAPA?

No. That description was close to accurate after Watali, but bail is obtainable through the ingredient inquiry, through surface analysis of the material, and through the Article 21 delay track, typically where custody has run for years or the material plainly fails to disclose an offence.

The fuller answer is that impossibility was never the text of the proviso, even if it often described the practical experience of applicants after Watali. The proviso refuses bail where there are reasonable grounds for believing that the accusation is prima facie true. It does not refuse bail in every case in which an offence under the Act is alleged, and the difference between an allegation and reasonable grounds for believing it is the space in which the ingredient inquiry and the surface analysis operate. The delay track occupies different space. It does not depend on showing that the accusation fails at the prima facie stage. It depends on showing that custody has continued for so long, with so little prospect of conclusion, that detention can no longer be justified under Article 21 whatever view is taken of the material. An applicant may lose on the statutory track and succeed on the constitutional track, or succeed on the statutory track without needing the constitutional track at all, as Shoma Sen and Jalaluddin Khan show. An applicant may also fail on both, as Gurwinder Singh shows, where the material is held to be prima facie true and the trial is progressing.

The two-track bail test

The jurisprudence is not contradictory but bifurcated, and that is the namable claim this article advances. One track is statutory and merits based. It asks what offence is charged, whether its ingredients are disclosed, and whether the material, assessed by the method the courts have prescribed, gives reasonable grounds for believing the accusation to be prima facie true, including a surface check on whether the material is worth the weight the prosecution seeks to place on it. The other track is constitutional and custody based. It asks how long the accused has been detained, what progress the trial has made, and whether there is a realistic prospect of conclusion within a reasonable time, so that continued detention either remains a proportionate incident of a functioning prosecution or has become an infringement of the right to a speedy trial under Article 21. A narrower road, default bail, stands apart from both tracks because it turns on the expiry of the statutory period for completing the investigation rather than on the merits or on the overall length of custody. The main source of confusion in argument is that a proposition correct on one track is often advanced as if it answered a question arising on the other. That a charge is serious answers part of the statutory inquiry. It does not, after Sheikh Javed Iqbal, answer the constitutional inquiry by itself. That custody has been long answers part of the constitutional inquiry. It does not, after Gurwinder Singh, displace the statutory inquiry where the trial is progressing and the material is prima facie true. The steps in the table should be run in order, because an early step may dispose of the application without any need to reach a later and more contested one, and because the authority governing each step is different.

Step	The question the court asks	The authority that governs it	Where the answer points
First	Does the charge, as framed, disclose an offence under Chapters IV or VI at all, including the mental element the offence requires?	Thwaha Fasal (2021) on Sections 38 and 39; Section 43D(5) by its terms	If the ingredient is missing, the bar is not attracted and ordinary bail principles apply
Second	On the chargesheet material taken as a whole, are there reasonable grounds for believing the accusation is prima facie true?	Section 43D(5); Watali (2019)	If the material does not reach that threshold, bail follows under the statute itself
Third	Does the material survive a surface-level check of its probative worth, or is it hearsay, inference stacked on inference, or distortion?	Vernon (2023); Jalaluddin Khan (2024)	Material of low probative value does not satisfy the prima facie test
Fourth	How long has the accused been in custody, and is there a realistic prospect of the trial concluding within a reasonable time?	K. A. Najeeb (2021); Sheikh Javed Iqbal (2024); Shoma Sen (2024)	Prolonged custody with no end in sight opens the Article 21 track in a constitutional court
Fifth	Is the investigation simply unfinished because the statutory period has run out?	Section 167(2) of the Code as modified by Section 43D(2); Bikramjit Singh (2020)	Default bail follows by right, and the merits bar does not close this road

A practitioner running an actual case through the steps in order begins, therefore, not with delay, however striking the custody period may be, but with the charge and its ingredients, because a concession that an offence under Chapters IV or VI is disclosed, when the mental element or another ingredient is in truth missing on the charge as framed, surrenders the cleanest ground of release and forces the application onto harder terrain. The practitioner next assembles the chargesheet material as a whole and tests it against the prima facie standard, separating material that connects this accused to the offence from material that establishes only the existence of an organisation, an incident, or the conduct of others, and then subjects the connecting material to the surface check for hearsay, stacked inference, and distortion that Vernon and Jalaluddin Khan permit. Only if that statutory argument fails, or alongside it as an independent ground in a constitutional court, does the practitioner build the delay record, and that record must be documentary rather than rhetorical. The distinction at the heart of that exercise is the distinction drawn in the comparison between preventive detention and arrest, because bail after arrest in a pending prosecution is a judicial assessment of custody within a trial process that must move towards a verdict, whereas detention without trial rests on a different justification altogether, and a delay application succeeds by showing that custody which began as the former has, through inaction and the passage of time, acquired the practical character of the latter without any adjudication of guilt.

The same discipline governs preparation. The team acting for the applicant should be able to open and organise the custody chronology and trial record files with this office file viewer in spreadsheet form, so that the custody chronology, chargesheet dates, and trial progress record that a delay track application is built on can be opened and checked against their sources before the hearing, because such an application loses credibility the moment a date is misstated or a stage is misdescribed. Counsel should plead the tracks in the alternative, pressing the statutory grounds in the first instance, preserving the Article 21 ground for the constitutional court if those grounds fail, and keeping any default bail claim separate and founded on the computation of the statutory period rather than on the merits. Conditions of release should also be addressed, since a court that accepts a track in principle may still need assurance on attendance, on non interference with witnesses, and on residence and reporting.

The counter-readings, taken seriously

No account of this jurisprudence is honest if it treats the arguments against the delay track and against a wider reading of the prima facie inquiry as merely formal obstacles. Those arguments, at full strength, identify real institutional risks and a real question about the role of a court confronted with a considered legislative choice. The first counter reading is operational. Parliament enacted the bar in Section 43D(5), on the case made by the Union of India and the National Investigation Agency in these appeals, because organised violence cases present real risks of witness intimidation and evidence tampering, which is the operational reason for a higher threshold. That reasoning, which the Gurwinder Singh bench accepted, deserves to be stated without caricature. As the Union and the Agency have submitted, a witness frightened into silence, or evidence destroyed before trial, cannot be restored later. The bar, on this reading, is a prophylactic rule which prefers the integrity of the trial to the liberty of a person against whom a prima facie case has been made out, for the period until that trial concludes.

The second counter reading is constitutional in the opposite direction, and it concerns the separation of powers. A court granting bail on delay, where the statutory conditions for refusal are otherwise satisfied, is in substance declining to apply a valid statute, on the case advanced by those who press this objection. The statute has not been struck down, and its bar remains in force. If the bar yields whenever custody becomes prolonged, the exception may consume a considerable part of the rule without any bench having held the rule unconstitutional after full argument on its validity. That is a genuine constitutional question and not an obviously correct outcome in favour of the applicant. Defenders of the Najeeb line answer that Article 21 is also part of the Constitution, that a statute must be applied consistently with the fundamental rights of the person detained, and that a court which enforced the bar in a manner permitting indefinite detention without trial would subordinate a constitutional guarantee to a statutory provision. That answer is powerful, and Shoma Sen and Sheikh Javed Iqbal adopt versions of it, but it does not dissolve the difficulty: the point at which a valid scheme applied to a slow trial becomes unconstitutional detention in an individual case is identified by evaluation rather than by rule.

The third counter reading, the inconsistency objection, is in practical terms the most damaging criticism of the present position, and it is honestly difficult to answer. Outcomes appear to depend in significant measure on which bench hears the application. Gurwinder Singh and Shoma Sen were decided within a short interval by benches of the same strength, on custody periods both measured in years, and they emphasise propositions that point in opposite directions: that mere delay is no ground for bail in grave offences, and that deprivation of liberty must be reasonable, just, fair, and proportionate, with Gurwinder Singh confined to its own facts. Sheikh Javed Iqbal and Jalaluddin Khan then granted bail in July and August, in one case on the speedy trial guarantee in strong language and in the other on a restoration of the bail is the rule maxim. A doctrine whose application varies with the composition of the bench fails an elementary test of equality before the law, because two applicants whose chargesheets are equally thin and whose custody is equally long should not face materially different prospects for reasons unrelated to either fact. The answer usually offered, that each case was decided on its own facts, is true as far as it goes: the trials stood at different stages, the material differed in quality, and the roles attributed to the accused differed in kind. That answer explains some of the variation, but not all of it.

The general position described in this article is legal information about how courts have approached these questions, not advice on any individual case, and anyone facing proceedings under the Act needs advice on their own facts. Subject to that limitation, the verdict that the record supports is measured. The bifurcation is real and defensible, because Watali and Najeeb answer different questions: Watali prescribes the method for deciding whether the statutory bar applies on the prosecution material, and Najeeb decides whether a constitutional court may nevertheless release an accused whose detention, through the passage of time and the failure of the trial to conclude, has infringed Article 21. A method for assessing material and a limit on the duration of detention are not in logical conflict, even if their interaction in a particular case is contested. The absence of any custody threshold, and the absence of a larger bench settlement fixing the relationship between the two lines with authority binding on all benches of lesser strength, nevertheless mean that the weakest point of the doctrine is predictability. An applicant cannot be told, by reference to a period of years or a number of witnesses, that the delay track will open, and a prosecutor cannot be told that a progressing trial will always defeat it. On the record considered here, the popular claim that bail is impossible under the UAPA is no longer accurate, because bail has been granted on the merits, on surface analysis of distorted and insufficient material, and on the constitutional track after prolonged custody, in reported decisions of the Supreme Court. The claim that bail is readily available would be equally false, because the bar remains stringent, the prima facie test remains deferential to the prosecution material in the manner Watali prescribed, and Gurwinder Singh stands as a reminder that a grave charge, a prima facie case, and a progressing trial will together defeat an application founded on delay alone.

A single sub section inserted in 2008 has been given several distinct meanings by the same Court within six years, as a deferential test of prosecution material, as a threshold that fails where an ingredient is missing or a chargesheet distorts its own witness, and as a bar that a constitutional court may pass by where detention without a concluding trial infringes Article 21, and only the sequence of decisions, not the text alone, explains the law that a bail applicant actually faces.

Frequently Asked Questions

Q: Why is bail so hard to obtain under the UAPA?

Bail is hard to obtain under the UAPA because Section 43D(5) imposes a statutory bar that operates before ordinary bail principles apply. A person accused of an offence punishable under Chapters IV and VI, if in custody, shall not be released unless the Public Prosecutor has been given an opportunity of being heard. Even after that hearing, release is forbidden if the court, on a perusal of the case diary or the report under Section 173 of the Code of Criminal Procedure, forms the opinion that there are reasonable grounds for believing the accusation is prima facie true. Section 43D(6) makes these restrictions additional to the restrictions in the Code and in any other law.

Q: What did the Watali judgment decide about bail?

The Watali judgment, delivered on 2 April 2019 by a two-judge bench of Justices A. M. Khanwilkar and Ajay Rastogi, settled how a court must approach bail under Section 43D(5) of the UAPA. The court held that at the bail stage the totality of the material in the chargesheet is taken at face value. There is to be no mini-trial, and the court does not weigh the admissibility or the probative value of that material. The degree of satisfaction required at this stage is lighter than the degree required in ordinary bail proceedings. Once the chargesheet material, accepted as it stands, gives reasonable grounds for believing the accusation is prima facie true, the statutory bar applies.

Q: How did K.A. Najeeb change the position on UAPA bail?

K. A. Najeeb, reported at (2021) 3 SCC 713 and decided on 1 February 2021 by a three-judge bench of Justices N. V. Ramana, Surya Kant and Aniruddha Bose, opened a constitutional route around the rigour of Section 43D(5). The Court held that Section 43D(5) does not oust the power of constitutional courts to grant bail on the ground of violation of Part III of the Constitution. The rigours of the provision melt down where there is no likelihood of the trial completing within a reasonable time and the incarceration undergone has exceeded a substantial part of the prescribed sentence. The Court laid down no fixed period for this purpose, so each case turns on its own facts.

Q: Can a constitutional court grant bail despite Section 43D(5)?

Yes, a constitutional court can grant bail despite Section 43D(5), because that provision does not oust the jurisdiction of constitutional courts to grant bail for violation of Part III of the Constitution. That is the holding in K. A. Najeeb, (2021) 3 SCC 713, decided by a three-judge bench of Justices N. V. Ramana, Surya Kant and Aniruddha Bose. The statutory bar continues to govern an ordinary bail application, where the court asks whether the accusation is prima facie true on the case diary or the Section 173 report. The constitutional route is different in character. It rests on fundamental rights, and it succeeds where prolonged custody combines with no likelihood of the trial completing within a reasonable time.

Q: Does long incarceration alone justify bail under the UAPA?

Long incarceration alone does not justify bail under the UAPA, and it is not a free-standing entitlement to release. The authorities tie release to a combination of factors: incarceration undergone must be coupled with no realistic prospect of the trial concluding within a reasonable time. In K. A. Najeeb the Supreme Court added that the custody undergone should have exceeded a substantial part of the prescribed sentence, while laying down no fixed period. The limit of the principle appears in Gurwinder Singh (2024), where bail was refused despite custody of over five years because the trial was progressing. A court therefore examines custody together with the pace and likely course of the trial, rather than custody by itself.

Q: Does Section 43D(5) apply to every UAPA offence or only to some of them?

Section 43D(5) applies only to some UAPA offences, not to every offence under the Act. By its own terms the bar covers a person accused of an offence punishable under Chapters IV and VI of the UAPA. In Thwaha Fasal, decided on 28 October 2021, the Supreme Court confirmed that the provision applies only to Chapters IV and VI. An accusation under Section 13 alone, which falls in Chapter III and concerns unlawful activities, does not attract the bar. The consequence is practical. Where only a Chapter III accusation is made, the court decides bail without the prima facie true test in Section 43D(5), although the restrictions in the Code and in other laws continue to apply.

Q: Is anticipatory bail available to a person accused under the UAPA?

Anticipatory bail is not available to a person accused under the UAPA. Section 43D(4) provides that nothing in Section 438 of the Code of Criminal Procedure applies in relation to the arrest of a person accused of an offence punishable under the Act. Section 438 is the provision under which a court may direct that a person be released on bail in the event of arrest, before any arrest takes place. By excluding that provision altogether, the UAPA removes the anticipatory bail remedy for its offences. A person who apprehends arrest must therefore rely on other remedies open in law, because a direction under Section 438 cannot be granted in relation to such an arrest.

Q: How does the 180-day investigation window affect default bail under the UAPA?

The 180-day investigation window directly governs default bail under the UAPA. Section 43D(2), read with Section 167(2) of the Code of Criminal Procedure, sets an investigation period of 90 days, which the court may extend to 180 days on a report of the Public Prosecutor. If no chargesheet is filed within the period as extended, an accused who is prepared to furnish bail is entitled to default bail. In Bikramjit Singh v State of Punjab, decided on 12 October 2020 by Justices R. F. Nariman, Navin Sinha and K. M. Joseph, the Supreme Court held that this right is part of the procedure established by law under Article 21, a fundamental right and not a mere statutory right, and that only the Special Court may extend the period to 180 days.

Q: Which court hears a bail application in a UAPA case?

A bail application in a UAPA case is heard by the Special Court constituted under the National Investigation Agency Act. Where no NIA Special Court is constituted, the application is heard by the designated Court of Session. Appeals from the Special Court lie to the High Court. That structure does not exhaust the forums available to an accused. The High Court and the Supreme Court can also be approached in their constitutional jurisdiction, and that is the route governed by K. A. Najeeb, under which a constitutional court may grant bail for violation of Part III despite Section 43D(5). The forum therefore depends on the nature of the application: a regular bail application before the Special Court or the designated Court of Session, an appeal to the High Court, or a constitutional application.

Q: Must the Public Prosecutor be heard before bail is granted or refused under the UAPA?

Yes, for offences covered by Section 43D(5), the Public Prosecutor must be given an opportunity of being heard before bail is granted. The provision states that no person accused of an offence punishable under Chapters IV and VI of the UAPA shall, if in custody, be released on bail unless the Public Prosecutor has been given an opportunity of being heard on the application for release. The hearing is a condition of release, not a formality added afterwards. Section 43D(5) then adds a second condition: the court must also be satisfied, on a perusal of the case diary or the report under Section 173 of the Code of Criminal Procedure, that there are no reasonable grounds for believing the accusation is prima facie true. Section 43D(6) makes both restrictions additional to those in the Code.

Q: What did Vernon Gonsalves add to the bail test?

Vernon v State of Maharashtra, decided on 28 July 2023 by Justices Aniruddha Bose and Sudhanshu Dhulia, added a surface-level check on the quality of the material relied on by the prosecution. The bench held that the prima facie test is not satisfied unless the court undertakes at least a surface-level analysis of the probative value of the evidence presented against the accused. That requirement stops a court from treating the chargesheet as sufficient merely because allegations appear in it. The bench also held that mere possession of literature, even if that literature propagates violence, is not by itself an offence within Chapters IV and VI of the UAPA. Possession therefore had to be connected to the statutory ingredients, rather than assumed to prove them.

Q: Is mere membership of a banned organisation enough to deny UAPA bail?

No, on the reasoning in Thwaha Fasal v Union of India, decided on 28 October 2021 by Justices Ajay Rastogi and Abhay S. Oka. The bench held that mere association with, or support to, a terrorist organisation does not attract Sections 38 and 39 of the UAPA unless the association or support is done with the intention of furthering the activities of the organisation. That intention cannot simply be presumed from a label of membership or sympathy. It must be inferable from overt acts or from active participation shown in the chargesheet. A bail court therefore has to ask what the material actually attributes to the accused, and whether that conduct, if accepted, shows the required intention, rather than denying bail on association alone.

Q: Do the UAPA bail rulings contradict each other?

The rulings create tension in method, but the later benches present themselves as working within a single framework rather than openly departing from one another. Watali treated the chargesheet at face value on broad probabilities, while Thwaha Fasal and Vernon required attention to intention, overt acts and the probative value of the material at a surface level. K. A. Najeeb, a three-judge bench decision, also remains binding on the two-judge benches that decided the other cases. A two-judge bench cannot overrule a three-judge bench; it can distinguish a case on its facts or refer a question to a larger bench. Shoma Sen captured that position by saying Gurwinder Singh did not dislocate the axis of reasoning in Najeeb.

Q: What does it mean for an accusation to be prima facie true at the bail stage?

An accusation is prima facie true at the bail stage when, on the approach stated in Watali, the court forms its opinion from the totality of the chargesheet material taken at face value and assessed on broad probabilities. The phrase does not mean that the prosecution has proved the offence, or that the accused has lost any defence available at trial. It means the material, accepted for this limited purpose, discloses the ingredients of an offence attracting the statutory bar. Gurwinder Singh v State of Punjab used that finding as a threshold: once the accusation is found prima facie true at that stage, bail must be rejected as a rule. The inquiry is therefore preliminary, record-based and directed to the chargesheet as a whole.

Q: Can a court look at the quality of the evidence when deciding UAPA bail?

The answer depends on which strand of authority the court follows, because Watali and Vernon describe the exercise differently. Watali held that the bail court does not weigh admissibility or probative value, and instead forms its opinion on the totality of the chargesheet material taken at face value on broad probabilities. Vernon held that the prima facie test is not satisfied unless there is at least a surface-level analysis of the probative value of the evidence. That surface-level analysis is not a full assessment of credibility or proof. It allows the court to ask whether the material has any apparent capacity to support the statutory ingredients, while stopping short of deciding contested questions that belong to the trial.

Q: Why do courts refuse to hold a mini-trial at the bail stage?

Courts refuse a mini-trial because the bail stage is meant to test the accusation on the chargesheet material, not to decide the case before evidence is formally tested. In Watali, the court formed its opinion on the totality of that material taken at face value and on broad probabilities, without weighing admissibility or probative value. Deciding contested evidentiary questions at that point would require the court to resolve disputes that the trial itself is designed to examine through witnesses, cross-examination and a complete record. A premature finding could prejudice the trial itself, either by appearing to prejudge guilt or by appearing to reject material before it is properly presented. The refusal therefore preserves separate functions for bail and trial proceedings.

Q: What is the two-prong test for UAPA bail stated in Gurwinder Singh?

Gurwinder Singh v State of Punjab, (2024) 5 SCC 403, decided in February 2024 by Justices M. M. Sundresh and Aravind Kumar, stated a two-prong approach to the UAPA bail bar. The first prong asks whether the allegations make out an offence attracting the bar, which is the threshold at which the court decides whether the accusation is prima facie true. The second prong follows from a positive finding on the first: once the accusation is found prima facie true, bail must be rejected as a rule. The bench applied that approach strictly on its facts, refusing bail despite over five years of custody because the trial was progressing, and distinguishing K. A. Najeeb on its facts.

Q: Is there a fixed number of years after which a UAPA accused becomes entitled to bail?

No decision in the material lays down a fixed number of years after which a UAPA accused becomes entitled to bail. K. A. Najeeb coupled incarceration with other circumstances, including no likelihood of the trial being completed within a reasonable time and incarceration exceeding a substantial part of the prescribed sentence. Sheikh Javed Iqbal involved nine years of custody with only two witnesses examined, a combination that exposed severe delay. Gurwinder Singh points in the opposite direction on different facts: over five years of custody did not secure bail because the trial was progressing, and the bench distinguished Najeeb. The period spent in custody therefore matters as part of a wider assessment of trial progress and likely completion, not as a standalone threshold.

Q: Why does the size of the bench matter in the UAPA bail cases?

Bench size matters because K. A. Najeeb, (2021) 3 SCC 713, was decided by a three-judge bench of Justices N. V. Ramana, Surya Kant and Aniruddha Bose, while Watali, Thwaha Fasal, Vernon, Gurwinder Singh, Shoma Sen, Sheikh Javed Iqbal and Jalaluddin Khan were two-judge bench decisions. A two-judge bench cannot overrule a three-judge bench. It can distinguish the earlier decision on its facts or refer a question to a larger bench. That discipline explains why several later decisions present themselves as applying Najeeb rather than displacing it, even when their outcomes differ on custody or trial progress. It also explains why Shoma Sen said Gurwinder Singh did not dislocate the axis of reasoning in Najeeb.

Q: How should a UAPA bail application be framed to give it the best chance of success?

As general legal information, and not advice on any individual case, the practical lesson of the case law is to choose the track that the facts fit and to plead it with a precise record. One track is an ingredient or merits attack, arguing that the charge does not disclose a Chapter IV or VI offence, or that the material fails the prima facie test on a surface check. The other is the delay track in a constitutional court, built on the custody period, chargesheet dates, witnesses examined against witnesses cited, and the realistic prospect of conclusion. A clean custody chronology and a trial progress record support either presentation and help the court match the application to the correct line of authority.