He was convicted on a single witness's dock ID—2 years after the mob attack

A mob of 1,500. One witness. Two years later, he pointed at a man in court and said, 'That's him.' The problem? He'd never seen the accused before that day. By the time the Supreme Court heard Javed Shaukat Ali Qureshi's appeal in September 2023, three of his co-accused had already walked free on the same evidence — and the court was about to ask a question that should have been asked years earlier: if the testimony against them was identical, why were some convicted and others acquitted?

The day the mob came to Shah Alam

On 7th November 2003, a crowd of 1,000 to 1,500 people gathered in the Shah Alam area of Ahmedabad. What started as a gathering turned violent. The mob stopped passersby, assaulted them, burned a two-wheeler, snatched a woman's necklace, and killed a man named Mukesh, whose body was later found in a nearby lake.

Thirteen people were charged. The trial court convicted seven of them, including Javed (accused no. 6), and sentenced them to life imprisonment. The Gujarat High Court confirmed the convictions but reduced the sentence to ten years. Four of the convicted men — accused nos. 1, 5, and 13 — appealed to the Supreme Court and were acquitted in 2018. The reason: the two police constables who claimed to have seen the attack gave testimony the court found unreliable.

Javed did not appeal at the same time. Neither did accused nos. 2, 3, and 4. By 2023, when Javed's case finally reached the Supreme Court, the question was no longer just about him.

One witness, two years, zero identification parade

The entire case against Javed rested on a single witness — PW-2, a man who had been present at the scene. He identified Javed in court, two years after the incident. There had been no test identification parade (a formal procedure where a witness picks a suspect from a lineup of strangers, conducted soon after the crime to test memory). The witness had never known Javed before that day. And he had picked him out from a mob of 50 to 100 people.

The prosecution argued that Section 134 of the Indian Evidence Act (the rule that says no particular number of witnesses is required to prove a fact) meant a single witness could be enough. The defence countered that this was not a case of a reliable sole witness. The identification was too weak, too late, and too unsupported.

When a single witness is not enough

The Supreme Court turned to its own precedent in Vadivelu Thevar v. State of Madras (1957), which classifies sole witnesses into three categories: wholly reliable, wholly unreliable, and neither wholly reliable nor wholly unreliable. A conviction can rest on a single witness only if that witness falls in the first category. If the witness falls in the third — where the testimony has some value but is not automatically trustworthy — the court needs corroboration (other evidence supporting the same fact).

PW-2, the court found, was firmly in the third category. He had no prior acquaintance with the accused. He had seen him in a large, violent crowd where identifying individual faces would have been difficult. He had not picked him out in a lineup soon after the crime. And he identified him in court two years later — a procedure known as dock identification, which courts have long treated as inherently weak because the witness sees the accused in handcuffs, in the dock, often the only person who looks like a defendant.

Citing Musa Khan v. State of Maharashtra (1976) and Pawan Kumar v. State of Haryana (2003), the court held that such dock identification, without a prior test identification parade, is unsafe to base a conviction on. Javed was acquitted.

The principle that freed three more men

But the court did not stop there. It noticed something troubling. The evidence against Javed was the same as the evidence against accused nos. 2, 3, and 4 — all had been identified by the same witness, in the same way, with the same weakness. Yet accused no. 2 had filed a special leave petition (SLP — a request for the Supreme Court's permission to appeal) that was summarily dismissed in 2018. Accused nos. 3 and 4 had not appealed at all.

The court applied the principle of parity — the rule that criminal courts must decide like cases alike. If the evidence against all four men was identical, convicting one and acquitting the others would amount to discrimination. The court cited Harbans Singh v. State of U.P. (1982), which held that making a distinction between similarly placed accused violates the fundamental right to equality before the law.

Under Article 136 of the Constitution (the Supreme Court's power to grant special leave to appeal), the court exercised its suo motu (on its own motion) powers. It recalled the 2018 dismissal order against accused no. 2, granted him leave to appeal, and acquitted him. It acquitted accused nos. 3 and 4 directly. All four men were ordered to be set at liberty.

THE PLAY: A dock identification without a prior test identification parade, by a witness with no prior acquaintance, from a large crowd, after a significant delay, cannot sustain a conviction — and if the same weak evidence convicts one accused while acquitting another, the court must step in on its own.

The walk-off

The mob had scattered years ago. The witness had done his best. But the court ended where the Constitution began: with the right to liberty, and the rule that like cases must be treated alike.