NEW DELHI: On 14 May 2015, five Dalit men were killed in collective violence at Dangawas village in Rajasthan’s Nagaur district.  Several others were injured. Homes and property had been damaged and burnt.

More than 11 years later, a Rajasthan court said it had no doubt that the violence at Dangawa had occurred.

Yet not one of the 40 men prosecuted for it could be convicted.

Special judge Ashish Bijarnia summed up the outcome in his 5 August 2026 judgment in a sentence that captures the failure at the centre of the case: “The occurrence is proved; the perpetrators are not proved.”

The court described what happened at the disputed field in Dangawas, in Nagaur district, as “heinous and heart-rending collective violence” in which five people from the complainant side died, and many were seriously injured. It said it had never doubted that the violence occurred.

The Dangawas families waited more than 11 years through an investigation first by the Rajasthan police, then by India’s premier federal investigative agency, the central bureau of Investigation, the CBI, only to reach a judgment in which the court could say with certainty that the violence occurred but could not hold one person criminally responsible for the five deaths beyond reasonable doubt through legally admissible evidence.

The Rajasthan police investigated first. Nineteen days later, the Central Bureau of Investigation, or CBI, took over.

The judgment records that neither agency carried out some of the most basic investigative steps needed in a case involving a large crowd and disputed identification.

Neither the Rajasthan Police nor the CBI conducted a basic identification parade; no DNA or ballistic examination was carried out; a tractor central to the prosecution’s strongest allegation was never objectively linked to the accused said to have driven it; and crucial evidence needed to prove recoveries and some serious injuries could not be used at trial. 

Deaths On Both Sides

The violence arose from a decades-old dispute over 23 bighas and five biswas of agricultural land.

According to the first statement of injured complainant Arjunram, recorded at the government hospital at Merta on 14 May 2015, around 200 people came to the disputed field on three tractors and about 50 motorcycles. He alleged that the group was armed, demanded that his family leave the land, used caste abuses, demolished a room and a hut, set property on fire and attacked members of the family.

Five men from the Meghwal family—members of a Scheduled Caste — eventually died.

There was, however, a sixth death arising from the same confrontation.

The police investigation found that Rampal Goswami, from the opposite side (the judgment does not identify him as Jat), had suffered a gunshot injury and was declared dead in hospital. 

Ramdev Khadav, a Jat from the opposite side who was also among the accused, suffered a gunshot injury in the same confrontation. A separate counter-FIR was registered over Rampal’s death.

The Dangawas court did not decide who fired the shot that killed him. It expressly said that the question belonged to the counter-case and that its judgment acquitting the 40 accused should have no prejudicial effect on that proceeding.

FIR 168/2015, concerning the deaths on the Dalit side, was registered at Merta city police station on 14 May.

The transfer followed demands from the victims’ families and Dalit-rights groups for an independent CBI investigation amid criticism of the local police’s handling of the dispute and allegations of negligence. 

The Nagaur SP (superintendent of police) recommended a CBI probe; the state had initially shifted the investigation to its CID-Crime Branch before referring it to the CBI. On 2 June—19 days after the violence—the CBI registered the two cases and took over the investigation.

The agency filed its first chargesheet against six men on 14 August 2015. A supplementary chargesheet the following year added 34 more, taking the number of accused to 40.

The chronology is important.

The Rajasthan Police handled the most valuable first days, when the scene was fresh, and eyewitness descriptions could most easily be recorded.

But the CBI took over less than three weeks after the violence and controlled it thereafter.

The final judgment shows that crucial gaps survived both investigations.

Not One Identification Parade

Perhaps the clearest example was the identification of the alleged assailants. The prosecution case concerned a crowd said to number in the hundreds. Many alleged participants were not known to individual witnesses beforehand.

Yet the court found that neither the Rajasthan Police nor the CBI conducted a single test identification parade.

A test identification parade, commonly called a TIP, is used when witnesses are asked to identify a suspect from among several people before trial. It is particularly useful where a witness did not already know the alleged offender.

That was only half the problem.

The first investigating officer also admitted that no physical descriptions were recorded for the unnamed people in the crowd.

The police therefore failed to create an early record of what unidentified assailants allegedly looked like.

The CBI, which inherited the case 19 days later, did not conduct an identification parade to test whether witnesses could identify them.

Years later, those omissions mattered.

Prosecution witness Shobha Devi, herself injured in the violence, admitted that she did not know any of the accused before the incident and became familiar with them only afterwards.  

Prosecution witness  Shravanram was described by the court as “an injured eyewitness witness for the prosecution”. The court also noted that he was from Khakhadki, not Dangawas, and was related to the deceased family through marriage. He said he did not know any of the accused either by name or face. 

The prosecution examined Jasoda Devi as both an eyewitness and an injured witness. She was the daughter of the deceased, Ratanaram, and the wife of Shravanram. She admitted that she knew only three or four people in the estimated crowd of 250-300 and had never visited the field before the day of the incident. 

Prosecution witness Sonki Devi was an injured prosecution witness and alleged eyewitness, and the wife of the deceased Pancharam. The court records that she could not identify the persons who assaulted her.

Another prosecution witness, Bhanwari Devi, acknowledged that she did not know one accused by face and that family members had supplied his name to her. The prosecution argued that both sides came from the same village and therefore knew each other.

The court said their own witnesses disproved that assumption.

By the time several witnesses testified, three-and-a-half to six-and-a-half years had passed.

The court did not accuse them of lying because they could no longer identify particular attackers. On the contrary, it said some of these admissions demonstrated their honesty.

But an honest admission of “I do not know who hit him” cannot prove who did.

That is precisely why identification evidence should have been secured at the beginning.

From 12 Names To 40 accused

There was another problem.

The number and identity of alleged attackers changed as the case progressed.

Around two hours after the incident, the first investigating officer had 13 names for verification.

Arjunram’s hospital statement contained 12.

The CBI’s first chargesheet accused six men.

Its supplementary chargesheet added 34 more.

The court recognised that there is nothing improper in finding additional suspects during further investigation.

What troubled it was that the expansion also appeared in witness testimony.

Kishanaram, a prosecution witness, named 19 accused when he first testified in January 2019.

When he testified again in January 2021, he named 36.

Another witness added two names during her second round of testimony that she had not named when she testified about two years earlier. Three other names disappeared.

The court observed that memory normally fades with time; it does not become richer.

One admission was more damaging.

Kishanaram accepted in court that Chutaram, one of the accused, had been wrongly included.

The judgment also records his statement that the CBI had called him back and said, in substance, that if some names were wrong, they could be corrected.

Other prosecution witnesses said Ramdev, another accused, was not present at the scene.

The judge did not conclude that the witnesses had invented the violence.

He concluded that their identification of particular accused required reliable corroboration.

That corroboration was often absent.

The Tractor That Left No Evidentiary Trail

The clearest example of how the investigation failed to secure objective evidence concerned a tractor. The prosecution accused Manaram alias Mehram Danga of driving a tractor into a room and hut occupied by members of the Dalit family, causing the structures to collapse.

Five witnesses connected him with this act.

The court itself called it the prosecution’s “most consistent and strongest allegation”.

But a tractor is very different from a fleeting glimpse of a man wielding a stick in a large crowd.

It is a registered vehicle. Driving it into a building should leave evidence that can be checked: the vehicle, registration and ownership details, physical damage, tyre marks, and traces around the demolished structure.

The judgment found none connecting a tractor to Manaram and the alleged act.

No tractor had been seized from him.

No registration, chassis or engine number established such a connection.

No mechanical or scientific examination linked a tractor to the collapsed structure.

The site plan did not record tyre movement near the building.

Nor did witnesses provide a distinguishing description, such as colour or model, by which the particular tractor could be separated from other tractors.

The court pointedly observed that this was not evidence that was impossible to obtain. Two agencies had investigated the case over nearly three years.

Another tractor was mentioned elsewhere in the evidence, but that did not solve the problem.

The prosecution relied on disclosure statements said to have led to recoveries of lathis and a tractor from some accused under section 27 of the Evidence Act.

The court eventually held that these disclosures and recoveries had not been legally proved and could not be used against the accused.

So there were two distinct failures.

The tractor in the prosecution’s strongest allegation was not objectively tied to the man alleged to have driven it.

A separate tractor recovery on which the prosecution relied could not legally connect another accused to the crime.

Evidence That Could No Longer Be Used

The recovery evidence presented another unusual problem.

The officer who had recorded important disclosure statements and conducted the subsequent recoveries, Poonaram Dudi, died before he could testify.

The court repeatedly said his death was nobody’s fault.

But the officer who eventually testified had neither heard the accused make those disclosures nor personally witnessed the recoveries.

The independent seizure witnesses also did not provide the missing proof on the crucial parts.

As a result, the disclosure statements and consequential recoveries could not legally be read against the accused.

The judgment also records that two CBI officers who had taken statements from certain witnesses were not examined at trial.

That meant some earlier witness statements could not be properly proved when contradictions became important.

These were not abstract technicalities.

In a trial where eyewitness identification had already become uncertain, physical evidence and properly proved earlier statements could have provided independent support.

Instead, evidence that might have strengthened or tested the witness accounts either did not exist in the required form or could not be used.

No DNA, No Ballistics

Science did little to fill those gaps.

Forensic examination found human blood on several objects.

But blood grouping was inconclusive, so the prosecution could not connect a particular weapon to a particular dead or injured person.

The judgment then records that no DNA examination was conducted.

There was no ballistic examination, despite firearm components being recovered from the scene.

No call-detail-record evidence was placed before the court either.

DNA testing could potentially have provided powerful, even conclusive, evidence linking biological material to a particular person, while ballistic examination could have linked firearm evidence to a particular weapon. 

Neither test was carried out. 

That left the prosecution without potentially important scientific evidence that might have independently supported, or contradicted, the eyewitness accounts.  

It meant that when eyewitness identification became uncertain, the prosecution lacked reliable scientific or other objective evidence to independently connect a particular accused to particular acts. In other words, there was little reliable independent evidence to confirm who had done what. The importance of that omission grew, not shrank, as the trial moved further from 2015.

What Was Wrong With The Medical Evidence

The medical evidence requires a more careful description.

It did not fail to prove that people had died or been injured.

The five deaths on the complainant side were established. The court also accepted that many people suffered injuries.

The problem concerned some of the prosecution’s claims that particular injuries amounted to fractures and therefore to the more serious offence of grievous hurt.

The doctor relied on X-ray material prepared elsewhere. He was not the radiologist who had conducted or interpreted those X-rays.

The radiologist was not examined in court.

Applying Rajasthan High Court precedent, the judge held that where proof of a fracture depended on X-ray evidence rather than what the treating doctor himself had directly observed, the radiologist and X-ray material had to establish that link.

They did not.

The result was not that the injuries disappeared.

Instead, the prosecution could not legally prove that some of them were grievous injuries for the more serious charges.

Once again, the court accepted the harm but found the evidentiary bridge to a particular criminal charge missing.

Caste Was Proved. Not A Caste Offense 

The court accepted that the five dead men and the injured complainants were Meghwals, a scheduled caste in Rajasthan.

It also accepted that the 40 accused were Jats and were not members of a scheduled caste.

But that did not automatically establish offences under the Scheduled Castes and Scheduled Tribes (Prevention of Atrocities) Act, 1989.

For the law applicable to the 2015 incident, the prosecution had to prove the additional ingredients of the particular offences charged.

For one major charge, this included proving the required connection between the crime and the caste of the victim.

The court said the evidence established a long-standing land dispute but did not establish the required caste nexus beyond reasonable doubt.

It also found that the prosecution could not reliably identify which particular accused allegedly uttered caste abuses and that other statutory requirements were not proved.

Whether a land struggle between Dalits and a dominant caste in a Rajasthan village can socially be separated from caste is a larger question.

But the criminal case still required investigators to collect evidence proving the caste-related ingredients prescribed by the law.

The court held that they had not.

The Sixth Death Complicated The Prosecution Story

Rampal Goswami’s death also mattered to the court’s assessment of how the violence began.

The record showed that he had been shot dead and Ramdev had suffered a firearm injury in the same confrontation.

The police also found firearm material at the scene.

One prosecution witness acknowledged mutual stone-pelting for around an hour.

The court therefore said the record did not permit it simply to accept the prosecution’s description of an entirely one-sided attack.

But it was equally careful not to decide the opposite.

It did not say who started the violence.

It did not find that the accused were acting in private defence.

And it expressly refused to decide who killed Rampal Goswami.

Those questions, it said, remained open, with Rampal’s death to be determined independently in the counter-case.

This is important because the acquittal cannot accurately be described as a judicial finding that the accused’s version was true.

The court said something narrower: the prosecution’s version had not been proved against the particular accused beyond reasonable doubt.

The Victims Bear The Consequences

Near the end of the judgment, the court made clear what its acquittal did not mean.

It did not mean that the crime had not happened.

Five people on the Dalit side had died. Many others had been seriously injured.

What had failed was the legally admissible identification of the people responsible.

The court acquitted all 40 accused with the benefit of doubt.

The court also referred the victims and dependants to the District Legal Services Authority for compensation under Rajasthan’s victim-compensation scheme. 

Separately, it directed the Nagaur district administration to determine how much statutory financial relief had already been paid to them under the SC and the ST (Prevention of Atrocities) Rules, 1995, and to disburse any balance within three months. There is no contradiction in law between those two orders.

There is, however, a devastating outcome for the families.

The state can recognise that their relatives were killed.

A court can call the violence heinous.

It can accept that the victims were Dalits.

It can order compensation.

But compensation is not criminal accountability.For the victims, the distance between those two findings is the distance between a crime recognised and justice delivered.

(V. Venkatesan is contributing editor at Supreme Court Observer. The views are his own.)

Get exclusive access to new databases, expert analyses, weekly newsletters, book excerpts and new ideas on democracy, law and society in India. Subscribe to Article 14.