Can Eyewitness Identification in a Pitch-Dark Night Be Relied Upon for Conviction?

An eyewitness may be sincere yet mistaken, particularly where identification takes place in darkness, from a considerable distance, or under inadequate illumination.

In Kartika @ Kirtan @ Kirtan Charan Jena & Anr. Etc. v. State of Odisha, 2026 INSC 1015, the Supreme Court found the eyewitness identification unsafe and held that it could not thereafter be relied upon as a circumstance against the accused. As the remaining circumstantial evidence also failed to establish guilt beyond reasonable doubt, the convictions were set aside, and the appellants were given the benefit of doubt. 

The Real Question Was Not Whether a Murder Occurred

A useful starting point in understanding the judgment is to separate two questions that are sometimes unconsciously merged in a criminal trial.

First: Was the deceased murdered?

Second: Did these particular accused commit the murder?

Proof of the first does not automatically establish the second. In Kartika, the medical evidence established that the death was homicidal. The Supreme Court nevertheless emphasised that the question before it was whether the prosecution had successfully established the appellants’ authorship of the crime.

This distinction goes to the heart of criminal jurisprudence. The brutality of an offence cannot compensate for weakness in proof of identity. A court may be completely satisfied that a grave crime has occurred and simultaneously remain reasonably doubtful about the identity of its perpetrators.

That doubt cannot be filled by the seriousness of the offence.

What Happened on the Night of the Incident?

The prosecution case began with events on 14 May 2001. At about 4:00 PM, accused Narendra @ Narahari Behera allegedly came to the informant’s house, abused members of the family and threatened to kill Dhruba Pradhan, the deceased.

At approximately 7:00 PM, a group of accused persons allegedly arrived armed with weapons, searched for the deceased and threatened to kill him before proceeding towards Abhayapur. Around the same time, the deceased, who was travelling on a motorcycle, was allegedly intercepted at Agi Chhak.

According to the prosecution, he attempted to escape but was caught, taken into a nearby paddy field and assaulted with various weapons. His motorcycle was subsequently thrown into a pond.

An FIR was lodged at 2:00 AM on 15 May 2001. The investigation ultimately resulted in a chargesheet against 18 persons. Fourteen faced trial.

The Trial Court convicted six accused under Sections 148, 506 and 302 read with Section 149 IPC, while eight others were acquitted by giving them the benefit of doubt. The convicted accused were sentenced to life imprisonment for murder.

The matter then reached the Orissa High Court. And this is where the evidentiary problem became particularly interesting.

The High Court Rejected the Eyewitness Version, But Still Upheld the Conviction

The High Court found that the persons committing the assault could not have been identified in a pitch-dark night merely from the sounds of assault. It therefore discarded the eyewitness testimony.

Yet it did not acquit the accused. Instead, the High Court concluded that although the prosecution had failed to prove the charges through direct evidence, the accused could nevertheless be convicted on the basis of circumstantial evidence.

That approach created the principal issue before the Supreme Court.

If the eyewitness account could not safely establish the identity of the assailants, what remained of the prosecution case? And were those remaining circumstances independently sufficient to form an unbroken chain pointing only towards the guilt of the appellants?

Before answering that question, the Court closely examined the alleged eyewitnesses themselves.

When an Eyewitness Account Collides With Physical Possibility

The prosecution primarily relied upon PWs 3, 4, 10, 11, 17 and 26. They were relatives of the deceased: PWs 3 and 4 were his brothers, while PWs 10, 11, 17 and 26 were his sisters-in-law.

Their relationship with the deceased was not, by itself, enough to discard their evidence. Related witnesses are not legally incompetent witnesses. The real difficulty lay elsewhere, in contradictions concerning who actually witnessed the incident and in the physical conditions under which identification was supposedly made.

The FIR lodged by PW-3 indicated that only PW-17 and PW-26 had witnessed the occurrence. PW-4, however, gave a materially different account and suggested that PW-3, PW-10, PW-11, PW-17, PW-26 and he himself had been observing the accused during the occurrence.

Thus, the prosecution evidence did not merely contain minor discrepancies about peripheral details. It raised a more fundamental question: who was actually present to see the crime?

The Supreme Court then turned to an even more basic problem, the darkness.

Can a Person Really Be Identified From 60 Cubits Away in Pitch Darkness?

The Court recorded that the occurrence had taken place on a pitch-dark night. The alleged eyewitnesses were observing the occurrence from approximately 60 cubits away.

The question was therefore not an abstract one about the credibility of eyewitness testimony. It was a practical question concerning visibility:

Could the witnesses actually have seen and identified the assailants under those conditions?

The Supreme Court referred to its earlier decision in State of U.P. v. Ashok Kumar, (1979) 3 SCC 1. In that case, identification from a substantial distance at night had been considered unsafe even where some moonlight existed. The possibility of mistaken identification could not be excluded.

The contrast with Kartika was significant. Here, the evidence indicated a pitch-dark night.

There was another complication.

The source of light was not being directed at the accused by the witnesses. According to the evidence, the accused themselves were carrying torches and using them to identify the deceased.

That changes the mechanics of visual identification considerably.

A Torch in the Accused’s Hand Does Not Necessarily Illuminate the Accused

It might initially appear that the existence of torchlight solves the problem created by darkness. But the Supreme Court explained why this assumption could be misleading.

The Court relied on Tamilselvan v. State, (2008) 7 SCC 755. There too, the prosecution claimed that witnesses could identify assailants at night when the accused were carrying torches.

The Court had found such identification doubtful. If the assailants themselves were holding the torches, the light would ordinarily be directed away from their own faces. Depending upon the positioning of the witnesses, the light might even interfere with their ability to see the persons holding the torches.

The Supreme Court applied that reasoning to the present case. The witnesses’ own evidence showed that the accused had focused torchlight to identify the deceased. This did not satisfactorily explain how witnesses standing at a distance in a pitch-dark environment could reliably identify the persons holding those torches.

This is an important evidentiary point.

The existence of light and the illumination of the accused are not the same thing. Whenever nighttime identification forms a crucial part of the prosecution case, courts must consider where the light came from, where it was directed, its intensity, the distance between witness and accused, the duration of observation and whether the witness had a genuine opportunity to recognise the person.

Simply stating that “there was a torch” does not answer these questions.

Can Identification Be Made Merely From the Sound of an Assault?

The prosecution case also involved identification connected with sounds emanating from the occurrence. The High Court itself had found that it was not possible for the witnesses to identify the assailants merely on the basis of the sounds of assault. The Supreme Court treated this finding as significant.

It ultimately concluded that the eyewitness testimony could not safely be relied upon because the witnesses could neither be said to have properly witnessed the occurrence in pitch darkness nor reliably identified the assailants merely from the sounds associated with the attack.

The circumstance of identification was therefore not proved. That conclusion had an immediate consequence. The case was no longer one supported by dependable direct eyewitness evidence. It became a case resting upon circumstantial evidence.

And the standard applicable to such a case is exacting.

Once Direct Evidence Falls, Circumstances Must Stand on Their Own

After discarding the ocular version, the High Court had relied on circumstances to maintain the conviction. The Supreme Court therefore revisited the classic principles laid down in Sharad Birdhichand Sarda v. State of Maharashtra, (1984) 4 SCC 116.

The familiar “panchsheel” or five principles governing circumstantial evidence require, in substance, that:

  1. the circumstances from which guilt is inferred must be fully established;
  2. the established facts must be consistent only with the hypothesis of guilt;
  3. the circumstances must be conclusive in nature;
  4. they must exclude every reasonable hypothesis except guilt; and
  5. the chain of circumstances must be so complete that it leaves no reasonable ground for a conclusion consistent with innocence.

The Court also referred to Kamal v. State (NCT of Delhi), 2023 SCC OnLine SC 933, reiterating the crucial distinction between circumstances that “may be” established and those that “must or should” be established.

Suspicion, even very strong suspicion, is not proof beyond reasonable doubt.

This principle became particularly important because some of the circumstances relied upon by the High Court were themselves dependent upon the very eyewitness testimony that had already been found unreliable.

Discarded Evidence Cannot Be Recycled as a Circumstance

Among the circumstances relied upon by the High Court were the earlier threat to the deceased, the accused allegedly coming armed to his house, the search for him, their presence at Agi Chhak, identification through voice, the witnesses hearing sounds of assault, recovery of weapons from the spot and recovery of the deceased’s motorcycle.

But the Supreme Court noticed an evidentiary problem. Several important circumstances substantially rested on the depositions of the same prosecution witnesses whose accounts contained material inconsistencies about who was actually present at the scene.

This matters because circumstantial evidence cannot become a substitute label for evidence that has already failed the test of reliability.

If a witness could not reliably see the accused, a court cannot first reject his visual identification and then use substantially the same identification as one of the circumstances completing the chain.

Every circumstance must possess independent evidentiary strength.

Witness Conduct Also Entered the Court’s Assessment

The Supreme Court additionally examined the conduct attributed to PW-3. Although PW-3 claimed to be an eyewitness, he was unable to state which weapon had been carried by which accused. More importantly, according to his evidence, after the assault he went home and returned to the scene after approximately two hours.

The Court considered it significant that he apparently made no attempt during that period to inform other family members or anyone else in the village about such a serious occurrence.

Witness conduct is not governed by a rigid formula. Different people react differently to traumatic events. Courts therefore have to be cautious before treating an unusual reaction as inherently false.

But where the prosecution case already suffers from contradictions and doubtful identification, conduct inconsistent with the narrative may become another factor in assessing whether the evidence inspires confidence.

The 53-Kilogram Laterite Stone Raised Another Unanswered Question

The prosecution theory also involved a laterite stone weighing approximately 53 kilograms. The Supreme Court found no material satisfactorily explaining how such a heavy stone had been brought to the place or who had brought it. The alleged stone used in the commission of the offence had also not been recovered in the manner necessary to connect it with the accused.

The medical officer had stated that a laterite stone of the relevant type could cause fracture and brain haemorrhage if thrown from a height of two feet upon the head of a person lying on the ground. At the same time, no weapon of offence, including a laterite stone, had been sent to him by the Investigating Officer for his opinion.

Again, the problem was not whether such a stone could theoretically cause the injuries.

The problem was whether the prosecution had proved the evidentiary link between the alleged object, the assault and the accused.

Every Link Must First Be Proved Individually

The Court referred to Abdul Nassar v. State of Kerala, 2025 SCC OnLine SC 111, for the principle that each incriminating circumstance must first be individually proved. Only thereafter can the court ask whether all the proved circumstances collectively form an unbroken chain consistent exclusively with guilt.

The process therefore has two stages.

A court cannot take several doubtful circumstances, combine them together and hope that their cumulative quantity cures their individual weakness.

Weak links do not become a strong chain merely because there are many of them.

The Court reiterated that each circumstance must be established beyond doubt and that, collectively, those circumstances must point towards the guilt of the accused.

Identification Evidence Must Be Tested Against the Conditions in Which Identification Occurred

The broader significance of Kartika lies beyond nighttime murders. Identification evidence is inherently fact-sensitive.

Where identification is disputed, the court should not examine only the witness’s confidence in saying, “I saw him.” It must examine whether the conditions permitted a reliable identification in the first place.

Relevant factors may include the distance between the witness and the accused, available illumination, direction of the light, duration of observation, obstruction of view, familiarity with the accused, number of persons present, movement during the occurrence and consistency of the witness’s earliest version with later testimony.

The central question is therefore not merely:

Did the witness identify the accused?

It is:

Was the witness placed in circumstances in which reliable identification was realistically possible?

That distinction protects the criminal trial from a particularly dangerous category of error, an honest but mistaken identification.

Related Witnesses: Relationship Was Not the Decisive Defect

Another point deserves clarification. The prosecution witnesses in this case were closely related to the deceased. The Supreme Court noted that fact while evaluating their evidence, but the judgment should not be read as creating a rule that testimony of relatives is inherently unreliable.

The decisive difficulties were the contradictions concerning presence at the occurrence, the physical impossibility or serious improbability of reliable identification under the prevailing conditions, the absence of dependable illumination and weaknesses in the circumstantial chain.

The correct judicial approach is therefore scrutiny, not automatic rejection. A relative may be the most natural witness to an offence. Equally, relationship cannot rescue testimony that otherwise fails tests of consistency, probability and reliability.

Benefit of Doubt Is the Consequence of an Unresolved Evidentiary Gap

The Supreme Court eventually found that the prosecution had failed to establish a complete and consistent chain of circumstances excluding every hypothesis other than the appellants’ guilt.

There remained substantial gaps in the evidence. The Court reiterated that where two views are reasonably possible, the view favourable to the accused must prevail. It consequently held that the High Court was not justified in affirming the conviction and that the appellants were entitled to the benefit of doubt.

The appeals were accordingly allowed, and the judgments of conviction were quashed and set aside.

What Does the Judgment Actually Establish?

The judgment should not be reduced to an absolute proposition that no identification made at night can ever be accepted.

That would go beyond what the decision establishes.

Nighttime identification may be reliable where the prosecution proves adequate illumination and a sufficient opportunity for observation, for example, where there are functioning streetlights, headlights, nearby artificial lighting or another credible source illuminating the accused, coupled with proximity and adequate duration of observation.

The principle emerging from Kartika is more precise:

Where an offence occurs in pitch darkness and the prosecution cannot satisfactorily establish how witnesses could have seen and identified the accused, particularly from a considerable distance, the court cannot safely base a conviction upon such identification.

And where that direct evidence fails, a conviction can survive only if the remaining circumstantial evidence independently forms a complete chain satisfying the standard of proof beyond reasonable doubt.

Click Here to Read the Official Judgment

Conclusion

Kartika @ Kirtan @ Kirtan Charan Jena & Anr. Etc. v. State of Odisha is ultimately a judgment about the discipline required in criminal fact-finding.

The existence of eyewitnesses does not automatically make a case one of reliable direct evidence. A court must examine whether those witnesses actually possessed the physical opportunity to observe what they claim to have observed. In a pitch-dark night, identification from a considerable distance cannot be accepted merely because a witness confidently names the accused.

Nor can unreliable identification be salvaged by relabelling it as circumstantial evidence. Once direct identification fails, every remaining circumstance must be independently established and collectively form an unbroken chain pointing exclusively towards guilt. If substantial gaps remain, suspicion, however grave, cannot perform the work of proof.

The enduring lesson of the decision is therefore simple but important: criminal conviction depends not upon how strongly an accusation appears to fit the narrative, but upon whether reliable evidence proves the identity and guilt of the accused beyond reasonable doubt.

Important Link

Law Library: Notes and Study Material for LLB, LLM, Judiciary, and Entrance Exams

Read More