BNSS and the quiet expansion of police custody

Most of the Bharatiya Nagarik Suraksha Sanhita is what its critics have said it is: a re-numbering of the Code of Criminal Procedure with a Hindi name. Sections shift; the architecture stays. The chapter on arrest is recognisable. The chapter on investigation is recognisable. The chapter on bail is recognisable. Even the deletions are conservative. If one read only the section-by-section concordance tables, the Sanhita would look like a forty-year-old code given a new typeface and a parliamentary photo-op. I want to argue that this reading is wrong in one specific place, and that the specific place matters more than the rest of the Sanhita put together. Section 187 of the Sanhita, which replaces Section 167 of the CrPC, undoes the most hard-won procedural protection in Indian criminal law. The fifteen-day cap on police custody — a cap that took the Supreme Court two decades to consolidate, that runs through Madhu Limaye and Anupam Kulkarni and Kosanapu Ramreddy, that has shaped what police investigation in India is allowed to look like — has been replaced with a window in which the police may return to custody of the same accused, in fragments, across forty or sixty days. The cap on the face of the statute is the same. The cap in operation is gone.

What the fifteen days meant under the CrPC

The architecture of Section 167 was built around three numbers and one principle. The numbers are twenty-four hours, fifteen days, and sixty-or-ninety days. The principle is that the executive's custody over the body of an arrested person is on a tight clock, and that the clock can be wound forward only by a Magistrate exercising independent judicial mind.

Article 22(2) of the Constitution does the first piece of work.⁷ No person arrested and detained in custody can be held beyond twenty-four hours without the authority of a Magistrate. Section 167(1) is the procedural instrument through which that constitutional mandate operates: the police officer must forward the arrested person, along with the case diary, to the nearest Judicial Magistrate. From that point on, custody is no longer something the police hold by their own authority; it is something the Magistrate authorises, every day, against a body of materials the police must place before the court.

Section 167(2) does the second piece of work. The Magistrate may authorise the detention of the accused in such custody as he thinks fit, but the total period of remand he can grant — Magistrate with jurisdiction or without — cannot exceed fifteen days in the whole.² The fifteen days is not a default to be shaved off if the investigation needs more time. It is a ceiling. Within that ceiling the Magistrate has a single, structural choice: police custody or judicial custody. Outside that ceiling, the choice disappears. Beyond fifteen days, custody must be judicial. The police can produce the accused, present materials, request further detention — but the body is in the custody of the court, not the station-house.

The proviso to Section 167(2) does the third piece of work. If the investigation has not been completed within sixty days for ordinary offences, or within ninety days for offences punishable with death, life imprisonment, or imprisonment of not less than ten years, the accused acquires an indefeasible statutory right to be released on bail.¹ The right is not the gift of the court. It runs against the State because the State has failed to investigate within the time the legislature allotted. The Supreme Court has called it "the most novel feature" of the new Code — the only provision in Indian criminal procedure that makes the State pay a procedural price for sloth in investigation.

The fifteen-day cap is the keystone of this arrangement. It is what makes the sixty/ninety-day clock honest. The police do not have sixty or ninety days of operational access to the body of the accused; they have fifteen, after which the body sits in jail, and the police must work with what they collected during those fifteen days plus the ordinary investigative tools — witness examination, documentary evidence, scientific tests — that do not require custodial interrogation. The Code's premise, throughout, is that custodial interrogation is a high-cost investigative tool whose use must be rationed and judicially supervised.

How the cap was won

The fifteen-day cap looks tidy on the page. Its consolidation in case law was anything but. Three lines of authority, decided across more than two decades, did the work.

The first is Madhu Limaye, In re, decided in 1969 under the old Code.⁵ The petitioners were arrested under Section 151 of the CrPC and produced before a Magistrate who passed orders of remand without applying his mind to the materials placed before him. The Supreme Court set the orders aside and held that the authorisation of custody under what is now Section 167 is a judicial act, not a clerical one. The Magistrate must be satisfied, on materials in the case diary, that further detention is justified. The reasoning is simple, but its consequence is that the fifteen-day window is not a window of police entitlement — it is a window of judicial decision-making, taken in steps, each of which requires fresh satisfaction.

The second is CBI v Anupam J Kulkarni, decided in 1992 by a three-judge bench.³ The question before the Court was whether police custody could be reordered after the fifteen-day window had closed, on the request of a fresh investigating agency that wanted to interrogate the same accused. The Court's answer was no. Police custody, the bench held, is not renewable once fifteen days have elapsed from the date of first remand. The custody beyond that point must be judicial. The principle underlying this condition, the Court said, is that "once the inquiry or trial begins it is not proper to let the accused remain under Police influence". The only exception was if the accused's complicity in a different transaction surfaced — a separate offence that had not been the subject of the original remand. In that case, for that case alone, police custody could begin afresh.

The third is Kosanapu Ramreddy v State of Andhra Pradesh, decided in 1994.⁴ The bench held that within the initial fifteen-day window, the Magistrate had jurisdiction to convert judicial custody into police custody and vice versa — the choice was open until the window closed. Once it closed, no conversion was possible. Custody beyond fifteen days had to be judicial, and judicial custody could not be reconverted into police custody for the same transaction. The two cases together produced the operative rule. Inside fifteen days, the Magistrate chooses. Outside fifteen days, the choice is unavailable.

Around this core sat the procedural superstructure. Hussainara Khatoon required the Magistrate to inform the undertrial of his right to default bail under the proviso to Section 167(2) — and to ensure that a lawyer was provided at State cost so the right could be exercised.⁸ Joginder Kumar required the police to record reasons for arrest itself, not merely for the custody that followed it.⁹ Arnesh Kumar sharpened the point in 2014 for offences punishable with seven years' imprisonment or less: no automatic arrest, and the Magistrate must scrutinise the Section 41A notice before authorising any detention.¹⁰ Satyanarayana settled how the 60/90-day clock was counted — from the first order of remand, not from arrest — and made clear that custody before the first remand was outside Section 167 altogether.¹¹

None of these decisions stood alone. They were a single architecture, built by judges who understood that the police's incentive in any investigation is to maximise time of access to the suspect, and that the only thing standing between that incentive and what the constitution calls due process was a hard ceiling, judicially enforced, on the operational meaning of the word custody. The fifteen-day cap was that ceiling. Anupam Kulkarni gave it the doctrinal name.

What Section 187 of the Sanhita does

Section 187 of the BNSS replaces Section 167 of the CrPC. Read in isolation, it appears to retain the fifteen-day police custody cap. Sub-section (2) of the new section authorises Magistrates to order remand in police or judicial custody for periods not exceeding fifteen days in the whole, in language closely modelled on the corresponding sub-section of the CrPC. The 60/90-day proviso is also there, in similar terms.⁶ A reader who came to the Sanhita expecting the framework of the CrPC would find — and many readers in the early commentary did find — that the framework had been preserved.

The change is in sub-section (3), and in the way the official commentary expects sub-section (3) to interact with sub-section (2). The Sanhita's formulation does not require police custody to be exhausted in a single block. It permits the fifteen days of police custody to be split and spread across the wider 40-day window applicable to most offences, or the 60-day window applicable to the gravest. The text contemplates that the Magistrate may, on a fresh application by the investigating officer, order the accused returned from judicial custody to police custody for further interrogation, so long as the total of police custody days across the entire 40/60-day window does not exceed fifteen.

What this rewrites, without saying that it rewrites anything, is Anupam Kulkarni. The judgment had held that once fifteen days had elapsed from the date of first remand, no further police custody could be ordered in respect of the same transaction. Section 187 abandons that holding. The fifteen days are no longer a calendar block. They are a budget the investigating officer can spend in instalments, returning to the well as the investigation progresses, with the body of the accused continuously available to be drawn back into police interrogation across the longest two months of his pre-trial life.

The defenders of the change present it in operational language. Investigations into complex offences — economic offences, organised crime, terror cases — frequently require fresh custodial interrogation as new material surfaces. The CrPC's hard ceiling, on this account, was a luxury the police could not afford in an era of long financial trails, encrypted devices, and conspiracies that unfold across borders. The Sanhita, the argument runs, retains the headline number (fifteen days) but releases its operational rigidity. This is presented as a calibration. It is, in substance, the dismantling of Anupam Kulkarni's holding under the banner of preserving its text.

Why this matters more than the rest of the Sanhita package

The first reason this matters more than the rest of the Sanhita is empirical. Police custody in India is not the procedural footnote it is in the textbooks. It is the operative location of most coerced confessions, most of the physical abuse that the Court has called the "worst kind of crime in civilised society", and the principal reason that conviction rates in this country track the quality of the chargesheet rather than the truth of the underlying allegation. Increasing the temporal reach of police custody, even by re-arranging the calendar inside an unchanged headline number, increases each of these costs in direct proportion. A custody that ends on day fifteen and cannot be reopened is a different instrument from a custody that ends on day five, resumes on day twenty-two, and reaches forward again on day thirty-five. The first is an investigative episode. The second is a continuous shadow that the accused carries through the entire pre-chargesheet phase of his life.

The second reason is structural. The fifteen-day cap was not a legislative gift. It was a doctrinal achievement, built by the Court out of the raw materials of Article 22, the proviso to Section 167(2), and the moral commitment expressed in Madhu Limaye and Joginder Kumar that the police's operational convenience is not a sufficient reason to keep a citizen in police custody. The Sanhita's reformulation of Section 167 into Section 187 returns the legislature to the position it occupied in 1973 — author of a text whose meaning was unsettled until the Court fixed it. The cost of that return is the entirety of the doctrinal work done in the intervening five decades. The Court can, of course, revisit Section 187 and read into it the limits that Anupam Kulkarni read into Section 167. There is no guarantee it will, and good reasons to think it will not. The Court that decided Anupam Kulkarni in 1992 was answering a question the legislature had not addressed. The Court that will be asked to read down Section 187 will be asked to override a legislative choice deliberately made and explicitly drafted. The institutional posture is different. The doctrinal hill is steeper.

The third reason is comparative. The BNSS package was marketed as a decolonising reform — the replacement of a British-era code with an indigenous Sanhita. The drafters did not, however, replace the parts of the British-era code that the British had built to constrain executive power. They replaced the parts that the post-independence Court had built. The fifteen-day cap is the clearest example. It is not a colonial relic. It is the post-1969 jurisprudence of an independent Supreme Court speaking against the operational habits of an independent Indian police. To dismantle it under the banner of decolonisation is to use a political vocabulary to do work the vocabulary was not chosen for. The Sanhita's other major changes — the new offences in the Bharatiya Nyaya Sanhita, the procedural reforms in trial timelines, the digitisation provisions — can be debated on the merits of what they propose to do. Section 187 cannot. What it proposes to do is to undo something specific that was achieved over half a century, and the proposal needs to be evaluated on those terms.

The counter-argument, taken seriously

I want to address the strongest version of the case for the change, because the case is not silly and a counter-argument that ignores it is intellectual cowardice.

The case runs as follows. The architecture of Anupam Kulkarni was built in 1992 for a different ecosystem of crime. A typical investigation in 1992 was a homicide, a dowry death, a property offence, a riot. The interrogative facts were available within fifteen days if they were available at all. Forensic technology was rudimentary; financial trails were on paper; the accused's electronic footprint did not exist. Today, the major investigations are different. A money-laundering case under the PMLA may produce new lines of inquiry on day twenty, on day thirty, on day forty-five, as foreign accounts are traced, as cryptocurrency wallets are de-anonymised, as the cooperation of co-accused develops. A terror investigation under the UAPA may require the suspect to be re-interrogated as new conspirators are arrested or as encrypted communications are decrypted. A cyber-crime investigation may require the same.

For each of these, the fifteen-day block is a procedural mismatch. Investigators cannot front-load all the questions they need to ask, because they do not yet know all the questions. The 154th Law Commission Report saw this difficulty in 1996, and the Standing Committee on Home Affairs saw it again in 2017.¹² Special legislation has been the substitute solution — the PMLA, UAPA, and several state organised-crime statutes contain provisions extending the investigation window beyond ninety days under judicial supervision, and the special bail standards in Section 43D(5) of the UAPA and Section 45 of the PMLA make extended judicial custody the operational norm in their respective offences. The case for Section 187 is that the BNSS internalises this special-legislation logic — without the special bail standards — and makes it available for ordinary offences too. The fifteen days are preserved. The mechanism of their delivery is modernised.

This is the steel-manned version of the case. I will say what is right about it and then say what is wrong with it.

What is right is that the 1992 ecosystem is not the 2026 ecosystem. Complex investigations do generate questions over a longer time horizon than a single fifteen-day block can accommodate. The CrPC's hard ceiling can produce, in particular kinds of cases, an investigation that closes prematurely or relies excessively on documents produced through Section 91 notices because the suspect is no longer reachable in person.

What is wrong is that the response is structurally over-broad. Section 187 does not differentiate between the complex investigations that motivated the change and the routine investigations that are the bulk of the Indian criminal docket. A theft case, a domestic assault, a road-rage offence — none of these requires a fortnight of fragmented police custody spread across forty days. None of them ever did. The drafters could have responded to the complex-investigation problem by extending the framework only for scheduled offences, on judicial application, with reasons recorded — the model the PMLA and UAPA already use. The drafters did not. They extended the framework for every investigation under the Sanhita, regardless of complexity. The over-breadth is not an oversight. It is the proposal.

And what is also wrong is the second-order consequence the proposal pretends not to notice. Section 35 of the BNSS reproduces the Arnesh Kumar safeguards on arrest itself.¹³ The Sanhita's defenders point to that as evidence of liberty-protective drafting. The point misreads the incentive structure. When arrest is procedurally cheap and custody is short, the police hesitate to arrest; when arrest is procedurally expensive and custody is long, the police arrest selectively and squeeze the suspect harder once he is in. Section 35 raises the price of arrest at the front door. Section 187 lowers the price of custody at the back. The net effect on liberty is not the sum of the two reforms. It is the product. The Sanhita's drafters have produced a regime in which the police arrest less often but extract more from each arrest. This is not what the safeguards literature contemplated.

What the right response would look like

If the problem is genuinely the inadequacy of fifteen days for complex investigations — and there is a real problem there — the response should be proportionate to the problem and shaped to it. Four features of a proportionate response are visible from the doctrinal history.

First, any extension of the police-custody window should be available only for a schedule of named offences, not for all cognizable offences. The PMLA and UAPA already operate on this model. The schedule should be tied to the empirical claim that motivates the extension — offences with cross-border evidence, electronic evidence, multi-accused conspiracies — and should not be expanded without legislative action.

Second, any extension should require a fresh judicial application supported by an affidavit from the investigating officer setting out, with specificity, what investigative steps require the renewed custody. The order extending the window should record those steps and should be subject to revision in the High Court. The model is the present extension of the 90-day window under Section 43D(5) of the UAPA, even if the existing UAPA framework has its own well-known defects.

Third, the accused should be entitled to legal representation at every Section 187 hearing, including the renewed-custody applications. The constitutional weight of Hussainara sits behind this requirement; the Sanhita should make it textual.

Fourth, the default rule for the 60/90-day proviso should be sharpened, not loosened. If the State wants more operational access to the suspect during the investigation window, the procedural price should be a shorter chargesheet deadline, not a longer one. The bargain the Sanhita has struck — more police access during the window without any compensating shortening of the window — is one-sided in a direction the constitutional baseline does not permit.

None of these four features would have been hard to draft. None of them would have closed off the investigations the BNSS's defenders worry about. All of them would have preserved the architecture of Anupam Kulkarni in substance even while modifying its arithmetic. The drafters chose not to do any of them. That choice is the proposal.

What the police actually gain — and what the accused loses

A specific scenario illustrates how the new framework operates on a body. Suppose A is arrested on 1 May for an offence punishable with six years' imprisonment. Under the CrPC, the police can apply for police custody up to fifteen days from the first remand. Suppose the Magistrate grants twelve days of police custody from 2 May to 13 May. From 14 May onwards, custody is judicial. The police can apply for chargesheet extensions, can examine witnesses, can collect documents — but they cannot return the body of A to the station-house for further interrogation. Their operational access to A's person ends on 13 May. If the chargesheet is not filed by 30 June (sixty days from first remand), A acquires the indefeasible right to default bail.

Under the BNSS, the same arrest produces a different operational reality. Suppose the Magistrate grants five days of police custody from 2 May to 6 May. From 7 May, A is in judicial custody. On 22 May, the investigating officer files a fresh application: a witness has named A in a fresh transaction; further interrogation is necessary; the officer requests police custody for five more days. The Sanhita's text permits the Magistrate to grant it, so long as the cumulative police custody across the 40-day window does not exceed fifteen days. The officer returns on day forty with a further five days of custody. By the time chargesheet falls due on day sixty, A has been moved between police and judicial custody three times. The fifteen-day budget is exhausted. The cap on the face of the statute is satisfied. The protection that the cap was designed to give A — that he would be in police hands for one defined block and then in jail — has been disassembled.

The cost of this disassembly is real and is borne unevenly. It is borne by the accused who is undefended, because the renewed-custody applications occur in the absence of effective legal representation in most district courts. It is borne by the accused whose family does not know that a renewed-custody application is even possible until it has been granted. It is borne by the accused who is poor, female, lower-caste, Muslim, or from a tribal background — every demographic on which Indian custodial abuse statistics consistently centre. It is not borne, in practice, by the accused who can afford a brief at the remand stage, because that accused will know which arguments to take against the renewed-custody application before it is made. The BNSS has introduced an instrument whose harm is concentrated where the doctrinal safeguards of Joginder Kumar and Arnesh Kumar are weakest in practice — in the lower courts, in the small towns, against the categories of accused who have always been the least able to fight back.

The doctrinal road back

There is a road back, but it is not a comfortable one. The Court that will hear the first sustained challenge to Section 187 — and the challenge is coming — has three options. It can read the section narrowly, treating the fifteen-day cap as a single calendar block rather than a divisible budget, and effectively restoring Anupam Kulkarni. It can read the section to permit splitting only on a recorded and reasoned judicial finding that the case is one of unusual complexity. Or it can read the section as the drafters intended and let the protection lapse into the form Parliament has given it.

The first option is the cleanest and the most consistent with the existing case law, but it runs into the difficulty that the drafters plainly wanted what they wrote. The second option is the compromise and is the most likely outcome — judicial readings that import Anupam Kulkarni's principle into Section 187 by way of constitutional avoidance, requiring exceptional circumstances and recorded reasons for any splitting of the fifteen days. The third option would mark the end of the fifteen-day cap as the Court has known it. It would also be the most candid acknowledgement that the legislature has rewritten settled procedural protection and that the Court cannot redo work the legislature has undone.

I think the second option is what we will get. I do not think it is sufficient. A constitutional-avoidance reading places the weight of the doctrine on the discretion of the Magistrate handling each renewed-custody application — which is, in lived practice, exactly the discretion that Madhu Limaye warned against in 1969. The hard ceiling did the work that judicial discretion could not be trusted to do reliably. Putting the weight back on the discretion concedes the central institutional point: in the absence of a hard ceiling, custody decisions track police preferences. Madhu Limaye was right about that fifty-seven years ago and is right about it now.

Where this leaves the Sanhita debate

Most of the commentary on the Sanhita package has settled into a binary. Either the new codes are a long-overdue indigenisation of colonial criminal law, or they are a re-numbering exercise that solves no real problem. Neither characterisation captures Section 187. Section 187 is not a re-numbering and it is not an indigenisation. It is a deliberate undoing of the most carefully built procedural protection in Indian criminal jurisprudence, written in language designed to make the undoing look like preservation.

The fifteen-day cap will continue to appear in the Sanhita's text. The cap that the cap means — the cap that Anupam Kulkarni built and that Kosanapu Ramreddy sealed — is gone. Whether the Court reads it back in, and what posture it adopts when asked to, is the next constitutional question the criminal procedure literature has to face. The cost of getting it wrong is paid in the body of every accused person now reachable for forty or sixty days by a police force that no longer needs to ask for fifteen days at a time, because the Sanhita has given it fifteen days at any time.

The CrPC's Section 167 was the part of Indian criminal procedure that, on a careful reading, the post-1969 Court could be proud of. The BNSS's Section 187 is the part of Indian criminal procedure that asks the Court to choose between honouring its own work and deferring to a legislature that has overwritten it. That choice is the constitutional question Section 187 puts on the docket, and the rest of the Sanhita debate is parochial by comparison.

Get the week's biggest decisions, distilled.

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