Garrison, a forty-two-year-old fishmonger, claimed to have seen Jane Morris on the evening of Friday, 17 October 1982. He waved to her as she passed the shop, then noticed her talking to the young man with dreadlocks. His statement included the line, ‘I kept an eye on her as I’m not overly fond of them sort, you know.’

His was the only witness statement to mention the dreadlocked man, but Mottram and his team took it at face value. That assumption informed the rest of the investigation, all the interviews with young West Indian men, and the community uproar. Perry had to cross-reference the original file notes for photographs, because few of these had been recorded on the microfiche. The case had attracted an enormous amount of interest, but the thing that particularly piqued his curiosity was a picture taken by a photographer at the South London Press on the first anniversary of Jane Morris’ disappearance. It showed Garrison and his wife, Linda, with a pram. The accompanying article was all about how much they missed their niece, but now had some fresh joy in their life. Perry stared at the picture. Linda Garrison was clearly in her forties, like her husband. He’d have to check her exact age. The article said they had been trying for a baby for years, and had finally been lucky to be blessed with little Graeme. If it had been a girl they would have called her Jane. None of the photographs of the bereaved family at the time of Jane’s disappearance showed anyone who was obviously pregnant.

If there was a child in the pram, could it really be theirs?

He had a growing suspicion, and decided to check it from the present-day evidence. He logged on to the police national computer upstairs, and keyed in his personal ID. He then downloaded the full PDF of Dr Delahaye’s post-mortem of Jane Morris, after she had been found in the flood. There was a lot of medical terminology he didn’t understand, and he laboriously looked each term up. One sentence intrigued him. ‘Beta HCG test positive, but uncertainty of effective freezing on human chorionic gonadotropin may cast doubt on nulliparous state.’

The word nulliparous turned out to mean never having had a pregnancy or birth. Perry picked up the phone and after numerous attempts managed to get hold of the forensic scientist. Delahaye briskly explained the terminology.

‘Forensically speaking the most reliable way to determine whether a woman has ever been pregnant is to detect in her bloodstream a particular chemical which is released by the placenta after implantation. It does not in itself indicate whether the foetus was brought to term. However there can be other reasons for the presence of an elevated level of chorionic gonadotropin, and as my report made clear the uncertainties of what happens to this chemical after the body has been frozen make it quite unreliable to deduce pregnancy from this measure alone.’

Perry was scribbling down notes as fast as he could go. ‘So on balance, doctor, would you say she had been pregnant?’