Gray, R v

Where later disclosure materially undermines an identification expert’s reliability, an appellate court must ask whether the jury would inevitably have convicted without that evidence. It should test its provisional conclusion by asking whether the evidence might reasonably have affected the verdict. If it might have done so, the conviction is unsafe.

Facial imaging evidence may assist a jury by demonstrating relevant characteristics. However, without an objective database, agreed formula or other objective measure, an expert should not express a probabilistic assessment of the support which facial characteristics provide for identification.

The appellant was convicted by a jury of robbery and possessing a firearm or imitation firearm with intent to commit robbery. The Crown case included a fingerprint on material found in the getaway car, CCTV images, eyewitness height evidence, the appellant’s silence at trial, and facial-imaging evidence from Mr Harrow identifying similarities between the robber and the appellant.

After trial, material emerged showing that Mr Harrow had made unreliable identifications in other investigations. The Crown accepted that it would not have called him had that material been available. The issue on appeal was whether the convictions remained safe without his evidence.

1. Appeal allowed. The court quashed both convictions. It held that the appropriate approach was that stated in Pendleton [2002] 1 WLR 72: ask whether the jury would inevitably have convicted without the impugned evidence, and test that view by asking whether the evidence might reasonably have affected the verdict.

2. Mr Harrow’s evidence had been an important part of the Crown case. Without it, the remaining evidence comprised a non-distinctive estimate of height, the jury’s unaided comparison of CCTV stills with the appellant, the fingerprint on foil in a car stolen 17 days before the robbery, and the appellant’s silence. The fingerprint did not inevitably establish that he was in the getaway car when it left the robbery, and silence could support, but could not found, a conviction.

3. The trial judge’s comment could not neutralise the effect of the unreliable expert evidence. The eyewitnesses had provided no relevant evidence about facial characteristics, and the judge had correctly left assessment of the expert evidence to the jury. The court could not say that the jury would have convicted without Mr Harrow’s evidence; the convictions were therefore unsafe.

4. The court added general observations. Reliable facial imaging or mapping evidence, including opinion evidence, may assist a jury by demonstrating facial characteristics, with enhancement where necessary. But absent an objective database, agreed mathematical formula, or comparable measure, an expert’s assessment of the probability or degree of support supplied by characteristics is subjective and cannot be objectively validated. Such assessments should generally not be expressed; the evidence should instead enable the jury to reach its own conclusion.

5. The court directed a fresh indictment and re-arraignment within two months, refused bail, and made a defendant’s costs order in respect of the solicitors’ representation.

• Court of Appeal (Criminal Division) Allowed the appeal in [2003] EWCA Crim 1001, quashed the convictions and ordered a retrial.
• Trial court A jury unanimously convicted the appellant of robbery and possessing a firearm or imitation firearm with intent to commit robbery. The judgment does not identify the trial court or provide a citation for its decision.