A conviction cannot be based on a retracted and uncorroborated confession if there are circumstances that cast doubt on the case.

Judgment

BOSE J. : The facts of this case can be placed in short compass. The appellant Muthuswami has been convicted of the murder of Nachimuthu Goundan and sentenced to death. The evidence consists of three eye-witnesses and a retracted confession. The learned Additional Sessions Judge disbelieved two of the eye-witnesses, namely Hanifa (P. W. 2) and Ghouse (P. W. 5), and rejected the confession on the ground that it was not voluntary. But he believed the third eye-witness Jamal (P. W. 1) who he thought was corroborated by certain other evidence and based his conviction on that. He also convicted another accused Pongiannan, with whom we are not concerned, on the same evidence and sentenced them both to death.

2. The High Court considered that P. W. 1 was as unreliable as the other two eye-witnesses and so refused to believe him. But they thought the confession had been wrongly rejected and, believing it to be voluntary, they upheld the conviction relying on the confession alone. They acquitted the other accused Pongiannan because once the eye-witnesses were discarded the only evidence implicating him was this uncorroborated confession of a co-accused. The question is raised whether a conviction can be based on a retracted and uncorroborated confession.

3. We do not intend to answer this in a general way because on the facts of this case it is enough to say that it would be unsafe to act on this particular confession. The deceased was murdered about midday on 14-8-1949. The eye-witnesses P. Ws. 1, 2 and and two more eye-witnesses who have not been called were examined by the police on the same day at the inquest. Despite that neither the appellant nor his co-accused were arrested. This may have been because they could not be found or it may be that the descriptions given were not enough for identification. That is conjecture. But what we do know is that none of the three eye-witnesses who have been called knew the accused before. They saw them for the first time in the actual act of committing the murder.

4. Next, we are told by Muthusami, P. W. 7, and Palanisami, P. W. 8, that the accused were detained for at least a fortnight in police custody two months after the murder, that is to say, they were detained in police custody for some days before their arrests on 23rd and 25th of October 1949. The police witnesses deny this but the learned Additional Sessions Judge believed that they were in police custody for at least six days before their arrests. The High Court preferred to believe the police witnesses but the learned Judges have brushed aside the testimony of these two witnesses somewhat summarily.

The learned Additional Sessions Judge who saw them and who saw the police witness Ratnaswami, P. W. 24, believed the former. The High Court give no reasons for preferring the latter despite the fact that they did not have the advantage of observing their demeanour. All they say is that they see no reason to disbelieve him. But that, in our opinion, is not sufficient to displace the conclusion of the Judge who actually saw the witnesses in the box. We would require more convincing reasons in a case of this kind and would want to be shown how and where the learned Additional Sessions Judge went wrong; also, in any case, the fact remains that the appellant and his co-accused were not arrested till nearly two and half months after the murder and the 5 High Court considers that the investigation was perfunctory.

5. Next come the identification parades held on the 1st and the 4th of November. The appellant was identified by each of the three eye-witnesses who have been called. We consider it would be unsafe to accept this identification two and a half months after the event. It is true that the murder was committed in the middle of the day but it is equally true that the three witnesses saw the assailants for a very brief interval of time even if the story about them walking by in single file ten minutes after the occurrence is true. It wo