I was interested in this when I too went looking:
In dismissing Sam’s appeal, Latham LJ faced a clear problem concerning Bilel Khelfa 's evidence at trial. The transcript of his testimony shows that he disowned his identification of Sam Hallam as set out in his second written statement of 20th October 2004. Without this statement, there existed no other evidence to link Sam Hallam with the murder. Latham LJ's bizarre explanation for the manifest discrepancies between Bilel Khelfa 's second statement and his trial testimony was
the nature of the evidence given by Khelfa was such that we suspect that the true flavour of that evidence does not emerge from the cold typescript of the transcript ... what appears to this court to be equivocal may in fact have indicated to those who were present that indeed the witness... may have given the impression, even if not expressly in his answers, that he was accepting what he had said on 20th October..
The background to this is that Khelfa's was the only identification evidence at the trial that actually named Hallam as the murderer. In his first statement he said unequivocally that he could not identify the alleged perpetrator because he never looked at him. After he was reminded of Hallam's name by another iffy witness whose evidence simply placed him at the scene (although no-one else did) he suddenly came out with a detailed statement saying he had looked at the perpetrator, had seen him standing over the victim with a baseball bat, and it was Hallam. It was that second statement that he disowned at the trial, saying he had been upset at the time over the murder of his friend, he had only named Hallam because the iffy witness had suggested the name to him, and all he had seen of the perpetrator was a boy on a bike wearing a hood which obscured his face. He gave evidence unwillingly and was apparently visibly upset. After he had sworn on the Koran he was heard to say he must now tell the truth about Hallam.
So what M'Lud seems to be saying is that the appeal court can decide that, when a witness says X was untrue, they can nevertheless uphold a conviction based on the assumption that the jury found that X was true purely because the appeal court "suspects" that the jury might
have gained the impression from his demeanour in court that his words mean the opposite of what he said. Of course in a case like this a jury is entitled to decide that it thinks the witness' disowned statement actually represents the truth, but logically that can only be if there is good evidence suggesting it - e.g. other witnesses or forensic evidence: in this case there was no other witness suggesting Hallam was responsible, and no forensic evidence against him whatsoever. The jury equally can decide that the witness' demeanour was so shifty that they don't believe the evidence he gave, but in that case they can't conceivably rely on his disowned statement either as evidence beyond reasonable doubt against the accused. There was never any suggestion that he could have been threatened into disowning the second statement.
It seems to me a classic example of the sort of intellectual dishonesty which the appeal courts regularly show when they don't want to allow an appeal. In effect they are saying that the conviction is safe - despite the fact that the principal witness categorically said he has no idea whether the perpetrator was Hallam - simply because the jury might
have decided from the way he gave evidence that his words meant the opposite of what they actually meant. It's precisely the sort of wriggling around that we saw from the appeal courts in cases like the Birmingham Six, and it really does the judiciary no credit at all.