After Tony Gauci, the forgetful Maltese shopkeeper, had been located by police he was shuffled through a string of identification spreads. First, with some justification, he picked out Mohamed Salam; then, for no good reason, he identified Abu Talb. Finally, near eighteen months after he’d been interviewed, he claimed this picture was “similar to the man who bought the clothing…“

The first thing to note is what a crummy picture this is. It looks like a mugshot of a pre-war gangster wearing a comedy wig. In a report(pdf) for Megrahi’s abandoned appeal Professor Tim Valentine, an expert in witness identification and facial recognition, noted that this compromised the spread…

[Looking at the spread he noted] that the picture of Mr al Megrahi has a grainy quality in comparison to the other pictures. An attempt has been made to degrade the quality of the other pictures. The pictures of the foils are lacking in sharp focus and lacking in contrast, but in my view do not match the grainy quality of Mr al Megrahi’s picture. Therefore, Mr al Megrahi’s picture does stand out…

Valentine later concluded that the process was “highly biased against the suspect” and “provided very strong influence, generating a serious risk of mistaken identification“.

Posters at the JREF forum (whose research is summarised in this new post from Caustic Logic) have noted another problem with the witness spread. The photo tha police used was cadged from Czechoslovakia and its date and origin is, I think, unknown. Later, though, this picture of Megrahi, from a year before the purchase, was discovered…

Do you see the likeness? Frankly, I think this would shame Private Eye on a dull fortnight. The hairdos are similar and the chins share bulk but the Czech photo displays a face that seems broader; pudgy, even; boyish in comparison with the long, drooping features of the latter pic. Even with allowances for angles and poise it’s a bad resemblance.

Rolfe points out what may have been a noble course of action…

Imagine, you’re investigating a crime, and it’s very important to find out exactly who it is the witness saw on a particular occasion.

You only have a very very bad likeness of your suspect, and your witness is pretty tentative in his comments. He doesn’t say “oh yes I recognise him that’s the man,” he just says, well this man looks similar but is too young.

You then manage to secure a much better picture of your suspect, in colour, with far better resolution, and known to have been taken only a year before the witness may have seen him.

What do you do?

If you’re really after the truth, you set up another ID session, this time using the better picture. (And you make sure to follow proper best practice this time.) Surely, if the witness does have a good memory of the person he saw, he’ll jump on the good likeness and say, oh yes that’s him!

Sadly, these investigators were “reluctant to show any more photos” for fear of “tainting what [Gauci] had already provided“. Oh, yeah, naturally, ‘cos that was crystal-bloody-clear!

(Also see.)