Continuing Impunity: Deaths in Police Custody in Delhi, 2016-2018

March 27, 2019

Full PUDR report available here

This report came about as PUDR was investigating custodial deaths that occurred in Delhi in 2018, and sought to follow up the cases of such deaths that had occurred in previous years – to examine police investigations, magisterial inquiries, and court prosecutions and see if they had resulted in justice. The National Human Rights Commission (NHRC), in response to an RTI, stated that it had recorded seven ‘custodial deaths’ (in police custody) between 1 January 2016 and 31 December 2017, and for this report, we also investigated the aftermath of these earlier cases.

The following is a report of our findings, with the three incidents of custodial deaths in Delhi in 2018 presented first, and subsequently, the follow-up of the seven incidents of 2016-2017. The three incidents are:

  • Death of Deepak, Karawal Nagar Police Station, 16 January 2018
  • Death of Dalbir Singh, Naraina Police Station, 21 February 2018
  • Death of Komal Kaur, Tilak Vihar Police Post, 15 July 2018

State Intervention after Custodial Deaths

As we have found in the past, in most cases of deaths in police custody there are no independent (non-police) witnesses. This was true for the deaths of Deepak, Dalbir Singh and Komal (all incidents of 2018) as well as the deaths of Rajni Kant, Som Pal (both in 2016), Ramesh, and Rajkumar (both in 2017).

In all these cases, the police were the sole witnesses. In the case of Anil and Kul Bhushan Chaturvedi, there were apparently other witnesses but the police were evidently culpable in other ways, and the deaths definitely occurred in custody. The need to collect evidence properly and ensure an independent investigation into these incidents is crucial to establish guilt in any offence.

In incidents of deaths in police custody, this seems nearly impossible under the present system. We have found that it is very difficult to get information about a custodial death and the MM inquiry if one were to go to a PS and ask about it a week or two after its occurrence.

In the cases of custodial deaths that took place between 2016 and 2018 (like in earlier years), our team found that we were arbitrarily doled out some bare facts in a few instances and in others not even that, and instead told to approach the DCPs office or the MM’s Court. Information or even press statements that would have been shared by the police with the press at the time of the incidents were also difficult to get.

We were told to apply through formal channels such as RTI applications, which we have done. At one level, there is now more overt acknowledgment in police administration about the practice of custodial violence and the routine bureaucratic procedures to be followed by them, far more systematically than in earlier years. Specifically, the NHRC has to be informed; the Metropolitan Magistrate, notified to take over the inquest and inquiry; and the press, informed as well – and all this seems to be done nowadays with practised ease.

Yet bureaucratization and superficial acknowledgments of custodial violence and torture have also meant greater efficiency in denying information to citizens and civil rights groups than earlier. Any information about the case has to now be applied for and sought through formal channels, such as seeking appointments with higher-ranking police officials or MMs and putting in RTI applications.

The process is slow and equally, if not more, non-transparent, making it difficult to piece together the truth about any incident. Families too remain substantially in the dark, with little information about what happens in inquiries and investigations. A few procedural norms have thus changed in the two decades since our last comprehensive report on the aftermath of custodial deaths in Delhi (1998), though in other aspects there remain significant continuities.

One example of the latter is the secrecy around custodial death investigations and inquiries – and another is the tendency among the police to generate different kinds of fictional accounts to explain the deaths that occurred, as ‘suicide’ bids, failed ‘escape’ attempts, ‘natural’ deaths, or as not having occurred in police custody –anything but the consequence of routine custodial violence.

Fiction after fact?

Custodial deaths as ‘Suicides’ or failed ‘Escape’ attempts Going by the fact that three of the ten custodial deaths that have taken place in the last three years are regarded popularly and by the police as the result of ‘suicide’ (including two out of the three deaths in 2018), Delhi Police Stations seem to be popular sites for those seeking to kill themselves.

It appears that ‘suicide’ was understood to be the cause of death in at least 30% of the custodial deaths inside Police Stations in Delhi between 2016 and 2018, based on the incidents discussed here. In the custodial death cases in the Jahangirpuri and Karawal Nagar Police Stations, and the Tilak Vihar Police Post, people in the prime of their youth allegedly committed suicide when they were in the lock-up.

This is a puzzling and disturbing trend, and certainly something the Delhi police should worry about – why do healthy young people with no suicidal tendencies, choose to commit suicide inside Delhi’s police stations? The SHO of Karawal Nagar went to considerable lengths to explain the reason for the death of 19-year-old Deepak in January 2018 as a ‘suicide.’ He said that the rate of suicide in Karawal Nagar was very high and the area could in fact be called the ‘suicide capital’ of the city, implying that this was because of the grimness of living conditions in this marginal locality in north-east Delhi.

He also took pains to explain that Deepak had been suicidal and had (according to him) tried to commit suicide in February 2017 – a description and fact denied by his family. Since there is no evidence for either, and the police official’s qualifications to assess the psychological state of the general population or the victim are doubtful, it is somewhat difficult to believe this view.

What this reveals however is the complicity of the police, whether or not they were directly involved in the custodial death, in explaining away the victim’s death through wider sociological and psychological factors, rather than seriously examining the allegation of custodial torture. These views were held and expressed long before the MM’s report (which is still pending).

The fact that these police officials are the main sources of information for the MM, and also wield power and authority in the local area where the family members of the victim still stay, means that this explanation, however unfounded, might unfairly prevail in the official conclusion about this death. In the case of the death of Raj Kumar in Jahangirpuri PS (2017) also, the official explanation of ‘suicide’ seems unlikely, given that he had come to the PS voluntarily and had shown no earlier signs of being suicidal.

The fact that he was a migrant, and no one was following up his case, makes it likely that the official story would have prevailed, though (as mentioned) we are still waiting to gain access to the MM’s findings. In three other cases discussed here, the deaths in police custody have been attributed to ‘escape’ attempts by the accused. One of these ‘escape’ stories, that of Dalbir Singh in Naraina PS (February 2018), has been exposed by the MM’s inquiry as doubtful and an FIR has been lodged against the guilty policemen.

It remains to be seen if this results in conviction, since in this case too, the police alone were witnesses and had several hours that they could have used to tamper with or remove evidence. Nonetheless there appears to have been enough reasonable doubt for the MM to conclude that this was a custodial death. The family of the victim was able to withstand any pressure from the police, and the accused policemen could be indicted. The same cannot be said for the death of Som Pal in Adarsh Nagar PS (December 2016).

As discussed above, the social and economic marginality of the victim’s life will probably determine the way his death will be described and judged. With some members of the family accepting monetary inducement from the evidently guilty police, and others being too vulnerable within the family to raise their voice, it is unlikely that the police will be finally held guilty or prosecuted, despite weighty evidence and an FIR against the police, and an implausible police cover story.

Thus, six out of the ten cases of custody deaths that we have investigated have been explained away as being cases of ‘suicide’ or a ‘failed escape attempt’ from custody. But from the cases of custody deaths here and in the past that PUDR has investigated, the use of torture by the police seems to be a routine way of interrogation.

The ‘suicide’ and ‘escape stories most often are attempts by the police to hide their crime. But even if we assume that these ‘suicides’ and ‘failed escapes’ are not a cover-up and that these persons did actually commit suicide or so desperately try to ‘escape,’ then too this points to a very disturbing dimension of police functioning. It exposes the relentless pressure and trauma, apart from physical pain, that police inflict on those it detains, the experience or fear of which is so unbearable that it is enough to drive the accused to think of ending their lives or take enormous risks to escape instead of going through repeated ordeals during investigations.

‘Suicide’ or ‘escape’ stories that the police have repeatedly given not only evoke the possibility of police complicity in these custodial deaths, but their easy circulation and acceptance in official circles points to the lack of any institutional accountability despite claims and appearances to the contrary. This is explained in our next section.

Institutional Mechanisms of Accountability

Magisterial Inquiries

In each of the cases mentioned here, it appears that at least compared to earlier, there are more mechanisms of ensuring institutional accountability. One of the standard demands in the 1990s and early 2000s by PUDR and other civil society groups was that (a) the inquiry under Section 176 CrPC in cases of custodial death should be conducted into every case of custodial death regardless of allegations of police involvement; and (b) the executive magistrate’s inquiry which was conducted under Section 176 CrPC should be changed to inquiry by a judicial magistrate.

In 2005, changes were brought about in Section 176 CrPC as a result of which SDM inquiries after police custodial deaths were replaced by Metropolitan Magistrates (MM). The change was possibly intended to reduce the possibility of bias, as the police and executive magistrates have a close working relationship. It was presumed therefore that the judicial magistrate would be more independent.

Moreover, the findings of a judicial inquiry, and the evidence collected therein, would also be admissible in court, unlike the inquiry by an SDM or executive magistrate. While we welcomed the change initially, hoping that it would lead to greater accountability of the police and greater transparency in official responses, our findings in these recent investigations show that this is not the case.

Only in the case of the death of Dalbir Singh in Naraina was the MM’s report completed within a year, which indicted the police and led to the registration of an FIR against the guilty policemen. In all the other cases, including cases of the death of Deepak in Karawal Nagar (2018) or of Som Pal in Adarsh Nagar (2016), the reports have taken an extremely long time to be completed, even in incidents where the needle of suspicion clearly points to the police.

It appears that the MM’s inquiry and reports may not be quite as ‘independent’ as imagined by the authorities who sought to change the CrPC, or by us when we questioned executive magistrates’ inquiries into custodial deaths. While the Magistrate’s inquiries are now mandatory and the post-mortem examination etc. must be carried out under their supervision, a few key problems remain:

• The Magistrate’s report is not a public document, and there is no way for citizens to enquire if the report is completed. Moreover, given that the report is now prepared by a judicial magistrate, any inquiry about it receives the standard response that it is ‘sub-judice’. MM’s reports are supposed to be sent to the NHRC, as is information of police action. These too are not accessible or public, as PUDR found.

• Further, the fundamental problem that remains is that the MM is substantially dependent on the police for gathering evidence. There are rarely any independent eyewitnesses to custodial deaths. While the Magistrate can call the families of victims and hear their testimonies independently, the police can influence the investigations in many ways.

For instance, victims’ families continue to live in the same area and are subject to the pressure of police from the same PS (if not the same policemen) who might have tortured their family members or caused their death, or been involved in covering it up, making it likely that the investigation will be skewed.

• In most of the cases discussed here (eight out of ten), families of victims were poor and otherwise underprivileged and deprived as well. Relative to the power that the police can and do exercise on the families of victims on the ground, their ability to resist this kind of pressure, whether it be threats or bribery, is extremely limited.

MMs seem largely and curiously unaware of these realities, and are not able to give protection to families from the local police. It is perhaps not a coincidence that the only magisterial indictment amongst these cases discussed has taken place in a case (Naraina PS, 2018) in which the family is economically better-off.

• Given the shroud of secrecy around the MM’s inquiry, it is difficult for citizens and civil rights groups to even communicate to the magistrates any anxieties about and information of such pressure, and their impact on particular investigations. The kind of delays, and cover-ups that seem to have taken place in the Naraina case (2018), or the pressure exerted on the family in the Adarsh Nagar case (2016), are likely to have occurred elsewhere.

It appears that judicial magistrates cannot be held accountable by citizens for delays in their reports or for other problems with their investigations. For all these reasons, the understanding that judicial inquiries would be fairer and more likely to result in justice is not borne out by our investigation into these recent incidents of custodial deaths.

Role of the NHRC

The NHRC issued a notification in 1993, soon after its inception, that it must be informed of any incident of custodial death or rape within 24 hours of the incident.

If not, it would give rise to the presumption that an attempt was made to suppress the incident. It appears that the NHRC is routinely informed of such deaths. Yet this does not seem to result in greater chances of justice. The role of the NHRC in cases of custodial violence thus needs to be examined. In 2001, NHRC asked states to send reports of custodial deaths within two months of the incident. Several questions concerning this direction remain unanswered.

To what extent are these guidelines followed? If not followed, what measures is the Commission taking to deal with it? Has the NHRC been able to use its resources to further justice? The Commission was envisioned as an effective legal remedy, substantive in nature, having a flexible way to reach out to every case. The NHRC has been unable to effectively discharge many of its functions, however, which may be because it has the power only to make recommendations. In the cases that we have followed, it has intervened in the form of asking for reports but has never received any replies.

The number of pending cases pending brings out the limitations of the Commission as a mere elephant in the room. While the NHRC responded to an RTI application concerning custodial deaths in 2016-18, it did not give FIR numbers or any other details. The response contained only file numbers, which were not enough to find out any details from any PS or its website. We could only find out more details about the cases using their dates, media reports and fact-finding investigations.

The Commission has largely failed to pursue the cases of custodial death in Delhi, where it is located. One wonders about the condition of human rights in other parts of the country. In cases where the NHRC has taken cognizance, like in Anil’s case (Ambedkar Nagar, 2017), the report was supposed to be given by the DIG within 8 weeks. This report is either unavailable or inaccessible. The Commission has recorded this case as a custodial death but has not followed it up.

According to the guidelines of the Commission, the state has to give information/ reports about a custodial death case in two months from the incident. But in this case, the police officials present in the PS at the time of our visit claimed that the inquiry is over and that the accused policemen have been given a clean chit; but it is unclear how the NHRC has not been informed, and why the official ‘status of the case’ is unchanged?

Is the police not following the guidelines or is the NHRC not vigilant enough? We have no way of knowing the answer. In all the cases we investigated except one, there is no information of when the NHRC was informed and what information was sent to them.

In Ramesh’s death in Mangolpuri in 2017, where we do have this information, the NHRC was provided with reports of the post-mortem and its videography in 2018 almost after a year after the incident, in clear violation of the 2001 directive that these be intimated to NHRC within 2 months of the incident. (, last accessed 16.3.2019).

In the case of the death of Rajni Kant in Ghazipur PS, as in every other case, the NHRC has apparently not been informed about case updates. In the Ghazipur PS case, the NHRC had summoned the DCP North, Crime Branch, but it was yet another no-show. NHRC took cognizance of Som Pal’s case (Adarsh Nagar) and in January 2017, issued notice to the DCP to bring out a detailed report with relevant documents.

This has apparently not yet been submitted. The NHRC has not been able to or interested in actually holding the police accountable for this case. Evidently, this failure of the police and the MMs to respond to the NHRC shows that they too take such matters lightly, merely sending them information about the deaths as part of ‘Standard Operating Procedure’ but nothing further.

It appears to be a meaningless gesture, since neither the police nor the NHRC seem to be committed to stopping custodial violence and deaths, given their persistent inaction. The fact that the NHRC has no independent investigative team, nor independent powers to ensure accountability from the police and magistrate, only partially explains its inaction. It could if it so wished have at least visibly put pressure on state authorities to give reports in time, ordered compensation (as it used to till a few years ago), ordered criminal prosecution of the guilty policemen and so on.

It could have acted as an institutional safeguard, however limited, instead of a dead-end, a place where information of custodial deaths is sent, recorded and then seemingly forgotten.


A striking feature of all of the cases discussed in this report is the lack of any official compensation paid to the families of victims by any state institution. This is a significant failure and the state must be asked why no effort was made to pay compensation. In a majority of the cases between 2016 and 2018, as discussed, the victims were poor and supported their families economically.

The only instance in which there was some evidence of the family receiving a sum of money was in the case of Som Pal, where the police paid the family (the amount was not specified for fear of police intimidation, or legal action). This was apparently done entirely privately and secretly, and as ‘hush-money’ to buy their silence rather than as ‘compensation.’

In cases like the death of Deepak (Karawal Nagar PS, 2018) or Ramesh (Mangol Puri PS, 2017) too, similar problems and pressures could crop up, and indeed might have emerged already. In the case of most of these victims, the ability of their families to persevere in the search for justice would be difficult given their great social and economic vulnerability, even if there had been independent witnesses and a more thorough and fair investigation.

In the present scenario of deeply skewed and unfair investigations also eventually controlled by the police, and the enormous power that the police have over the lives of marginal populations in the city, grants of compensation are the precondition for the victims’ families’ ability to secure justice. Its denial, conversely, amounts to a denial of justice.

Present Continuous: Custodial Deaths and Impunity

There is a strange sense of déjà vu while writing about custodial deaths in Delhi, because of the stark continuities in the cases, the police and family versions, and the outcomes, between the 1980s when PUDR started investigating them and now, more than three decades after.

Perhaps the starkest continuity is that very few policemen responsible for these deaths are criminally prosecuted, even though there is more talk of human rights, and significantly more press coverage of custodial deaths today. This results from a systemic failure to indict and punish the custodians of law who violate it. This continued impunity granted in practice to the police ensures the same-ness of the ‘script’ of deaths in police custody and their aftermath, which continues in Delhi even though so much else has changed.

The main reason for this continuity is the continued prevalence of custodial torture as a primary method used by the police to ‘solve’ crimes, instead of gathering of evidence and investigation. Custodial torture is rarely reported. In one of the few cases of custodial torture in Delhi that was reported (Vijay Vihar PS in 2013) and that we have investigated (PUDR report ‘Crimes of Habit’, April 2014), we found that there were many legal and extra-legal ways in which the police were sought to be exonerated when such torture was incidentally exposed.

Those who suffer it in most cases do not report this to the police because of their own vulnerable status, and fear of further torture and harassment to their families etc. The main problem is the huge structural imbalance of power between the police and those who are tortured in police custody in the course of investigations. As we have previously argued, deaths in police custody are not aberrations, but are usually the unintentional consequences of routinised torture.

In all the cases discussed in this report, a few policemen have been transferred or at most suspended, even when accused of destroying evidence (Som Pal, Adarsh Nagar PS, 2016), and even with ample evidence of injuries inflicted in custody (Ramesh, Mangolpuri PS, 2017). Police officials do not perceive these as ‘custodial’ deaths, despite the definition of custodial death in law as ‘Death occurring during the period when some limitation is placed upon the liberty of the deceased and that limitation must be imposed, either directly or indirectly, by the police’ [Criminal Law Journal (Cr.L.J. 635 (637)1970].

Delays in MM reports seems to be the norm, with rare welcome aberrations,as in the case of Dalbir Singh’s death (Naraina P.S., 2018). As mentioned earlier, the norm also seems to be that MM’s reports uphold police versions of custodial deaths, without going into the context of the case.

The delay in magisterial reports and their frequent echoing of police versions; the poverty and vulnerability of the majority of victims in custodial deaths and their families; the normalisation of police torture in public perception and the fickleness of public memory have contributed to a reality where custodial deaths continue to occur and remain unpunished, despite apparently greater visibility and official acknowledgment of ‘human rights’.

The rate of convictions in cases of custodial deaths is a blatant reminder that even when cases are registered and police are found culpable, finally very few are convicted. In most cases even where there is evidence of custodial violence, FIRs are not registered. In the cases discussed here, FIRs to probe the role of the police in causing the death in any way have been registered in only two out of the ten cases. The social and economically vulnerable status of most of the victims also affects the chances of justice, as our findings in these cases show.

In eight out of the ten cases, the victims belonged to marginal sections of society. The majority of the victims of custodial violence tend to be from these sections. After the death, their families are particularly vulnerable to police intimidation and pressure – the Adarsh Nagar case (2016, death of Som Pal) is a case in point. The FIR in the case came to be lodged only because of the circumstances (discussed above), with the body being found and the police role coming to light at the time, as well as the public protests.

From then onwards though, the systemic attempt to scuttle the investigation started. As mentioned above, the police offered money to silence the victim’s family. The payoff by the police to the victim’s family may itself amount to an admission of guilt but no action has been taken against them, and will likely not be. The nature of the incident – Som Pal’s death explained as an attempted ‘escape’ that failed – is similar to the story offered by the police for the death of Dalbir Singh (Naraina PS, 2018), which the MM has ruled as a ‘custodial death’ and in which case an FIR has been lodged.

Here too the police could put pressure on the family in the future but the family’s social status may enable them to withstand this better, and the possibility of the case going forward and the guilty being brought to book is thus higher. Cases of custodial death also continue literally for decades, like other court cases, and even the initial indictment of the guilty, gathering of evidence etc. goes on for a few years.

In the case of families of victims from the disprivileged social contexts, this effectively means a period during which they are constantly under pressure from the police, which is nearly impossible to withstand. It is perhaps not surprising that some of the rare cases in the past in which the police were indicted and convicted are ones in which victims’ families have been able to bear such pressure.

One such case that was in the news again in 2017 was that of the custodial death of a businessman Dilip Chakravarty in July 1995 [PUDR report ‘Special Staff on Duty: Death in the custody of North East District Police’ (Sep.1995)] in which the Delhi High Court upheld the conviction of six policemen who had been charged with beating him so severely that he died of head injuries. The victim had not been formally arrested. The conviction had taken place in 2001, and the wife of the victim and other witnesses had testified in court about the case prior to that, and even there, some witnesses had turned hostile.

These instances illustrate the point mentioned earlier, about the role and need for a policy on compensation. Such a policy is not a substitute for prosecution but a measure of accountability towards the family members of the deceased, one that would allow them to withstand pressure by accused police personnel and their supporters, often others in the police force, who could exercise power over their families.

The fundamental question in all these recent cases of custodial death is: why has the state not paid any compensation in any of these cases? The bizarre truth is that even if Som Pal’s death (Adarsh Nagar PS) comes to be treated as an ‘accident’ in the course of a supposed ‘failed escape’ attempt’, there are state policies for compensation for accidents but no policy on compensation for custodial violence and death. The fact that there is not even an effort to develop one is another absence that reflects the callous response of the state machinery and institutions towards custodial violence and deaths.

The award of compensation by state agencies (the NHRC, state governments, courts etc.) in the past has been arbitrary and depended largely upon the capacity of the family to pursue the case in courts, or upon the intervention of rights groups like PUDR to fight for the grant of compensation. Since this is itself random and arbitrary, the principle of equality in the law is violated at multiple levels.

The NHRC has occasionally awarded compensation in cases of custodial deaths but again, has apparently no consistent guidelines for its award. Is a trend apparent from the fact that there has been no compensation by the NHRC or any other state institution in any case of custodial deaths in Delhi in the last three years?

If so, this is utterly dangerous for democracy, given the need for compensation in cases where where families have lost their primary earning members and sources of survival, as well as any possibility of sustaining the wait for justice. PUDR has previously asked for guidelines to be formed on compensation for custodial violence and death. Supreme Court judgments like the one in DK Basu vs. State of West Bengal (1997) have held that compensation has to be seen as relief for infringement of the fundamental right to life by the state, and that Sections 330 and 331 of the IPC effectively make torture during interrogation punishable.

DK Basu states that, “The prosecution of the offender by the state is an obligation but crime needs to be compensated monetarily too as the court cannot stop at just giving declarations.” In this context the guidelines laid down in a recent judgment of the Calcutta High Court in September 2017 are valuable and should form the basis of state policy. The judgment decrees that the state should pay basic compensation to family members for every case of custodial death (Reba Bibi vs. State of W. Bengal and Others, 6.9.17, Calcutta. High Court, last accessed 16.3.2019) and the victim’s family is free to approach other authorities for further compensation.

It is only when such a policy is applied to all states (including Delhi) that arbitrariness with compensation for custodial deaths – dependant on the whims of the NHRC and the economic clout of the victim’s family – will be addressed. This along with other measures suggested here will create the basis for combating the impunity that the police in Delhi seem to enjoy in practice to torture, and kill, those in their custody.