A conviction for various sexual offences was unsafe where the two victims gave mutually corroborative evidence, as the judge's direction on collusion and possible contamination was somewhat garbled and no guidance was given as to the approach the jury should take.Appeal, with leave of the single judge, against convictions for gross indecency with a child, rape, buggery and attempted buggery. The defendant ('D') was sentenced to a total of eight years' imprisonment. There were two victims, D's niece ('B') and D's daughter ('S'). Between 1972 and 1975, B, who was aged between 7 and 8 years old, used to stay at her grandparents' home where she sometimes shared a bed with D. It was during those times that the offences occurred. S was sexually abused between 1987 and 1992 whilst she was aged between 9 and 14 years old. D was arrested and gave a no comment interview. At trial applications were made to sever the indictment, the judge refused as there were striking similarities between the offences. D gave no evidence. The prosecution's opening speech stated that the evidence of each victim had to be treated separately. However, at the close of evidence, on an application by the prosecution, the judge indicated he intended to give a direction on mutual corroboration of B and S's evidence and it was open to the prosecution to refer to it when making closing speeches. The judge summed up and directed the jury accordingly, setting out the similarities between the complaints. D appealed the conviction as unsafe on the ground that the judge was wrong to direct the jury that the evidence of B and S was mutually admissible. D relied on R v Davies & Poulton (1999) (Unreported 16/12/99) and contended that if he had been aware the judge was going to treat the case as he did he would have cross-examined specifically on the possibility of collusion and contamination. Further, the judge's summing up in relation to cross corroboration and similar fact evidence was flawed.HELD: (1) There was cross-examination on collusion and discussion on the possibility of contamination, the point was canvassed sufficiently so as to avoid unfairness. It was unreal to conclude that the present case was like R v Davies and Poulton (supra) as that trial would have taken a different course had the defendants been aware the judge would direct as he did. (2) It was clear the judge had in mind the JSB specimen direction dealing with the possibility of actual collusion. The judge was obliged to direct the jury to consider the real possibility that B and S had put their heads together to make false accusations. Whilst the judge sought to put the message across it could not be certain that the jury had got the message loud and clear. (3) The judge should also have directed on the possibility of innocent contamination and what approach the jury should take if they considered that that was a real possibility. The directions given were somewhat garbled and no guidance was given on the approach the jury should take. It was important as the case involved family members and the jury had to be alert to innocent contamination. (4) It was incumbent on the judge to direct the jury on the approach to take and in the present case the conviction could not be regarded as safe.Appeal allowed.