However, the interview in which Stud made the first confession was videoed. There is a gap in the recording, both sound and vision. Before the gap he can be seen repeatedly being threatened and bullied by the interviewing officer. Stud appears confused and sounds incoherent, frightened. He is complaining of a pain in his side, asks to see a doctor and says that he has been repeatedly kicked. There are traces of blood on his face.

When the recording resumes, the scene has changed. The interviewing officer is no longer visible on camera, but he, or someone like him, is in the room and questioning Stud. There is still blood on Stud’s face, and some of it has seeped into his shirt. Stud is lying back on the bench, his arms over his head. He is largely inarticulate. He is crying.

Then he says: ‘Yes, I killed that man. It was me.’ He goes on to say more.

It has all the appearance of a coerced confession, forced out of a vulnerable suspect after a long period of rough interrogation. If presented to a seigniory panel on Dearth there is little doubt that it would be deemed inadmissible.

However, for all that it was a full confession. Stud’s version of events aligned almost exactly with what the other evidence showed happened. Most compelling of all, no other suspect had been found or charged. Waller Alman did not kill himself – someone else was involved. Stud was the only possible killer in the frame.

My training had already made me familiar with certain attitudes held by police officers, and I had sometimes been aware of them myself – usually off-guard remarks while investigating a transgression. It amounted to this: when there is unchallenged partial evidence, and if it is supported by circumstantial evidence, and there is only one suspect in the case, there is a strong temptation to ignore everything else. The known or assumed facts are then fitted to the person they are questioning. It is known as confirmation bias.

If I had not already been suspecting it in the case of Waller Alman, everything was made clear when the young constable who was in the car with me said informally, as we drove away from the custody remand building: ‘As soon as Stud was brought in we all knew for sure that we had the right man. After that it was just a question of waiting for him to confess.’

There was a third evidential problem, one the local police were totally unaware of. This was not their fault as they had little or no experience of mutability, but it was one of the procedures I was tasked with carrying out. I was to try to establish if there was a mutability element to the killing, because if there was it could affect the quality of the forensic work. In the past Salay Hames had known few mutability events, but was considered high-risk on the scale of Mutability Incipience.