It has taken the case of Detective Constable Shirley McKie, wrongly identified at the scene of a murder by her own force, to convince the prosecuting authorities to take a fresh look at the reliability of fingerprint evidence.

Michael Mansfield, QC, first had doubts when he represented Danny McNamee, who was convicted of the Hyde Park bombing. "We were constantly being told that to question fingerprints was like questioning two and two and so we didn't," he recalls. "They thought that it was a science. We now discover it is no more exacting than comparing two shades of blue."

The most recent blow to the orthodoxy came when it was alleged that the print of McKie was discovered at a house in Kilmarnock, Scotland, where Marion Ross had been murdered three years ago. A man was later convicted of the crime but so convinced were the police of their evidence that McKie was tried for perjury. She was acquitted last year when a fingerprint expert gave evidence that the mark was not hers.

The shockwaves are still being felt north of the border where four fingerprint experts from the Scottish Criminal Records Office have been suspended and an investigation is under way. It has also given the Association of Chief Police Officers (ACPO), who are planning to scrap the "16-point system" in favour of a non-numerical system, something to think about.

Under the existing standard experts must find at least 16 matching marks between a suspect's prints and a print found at the crime scene. Once that criteria is reached it is almost always regarded as irrefutable evidence of the suspect's presence.

According to Mansfield, that logic is all very well provided there are controls. "If I put my forefinger down twice on a special clean piece of paper then obviously the ability to compare the two will be pristine," he explains. However, life is not always so straightforward - as was the case with Danny McNamee, who served 11 years for his involvement in the 1982 bombing that claimed the lives of four members of the Household Cavalry and seven horses. The Metropolitan Police claimed that his print was found on a Duracell battery. His conviction was quashed after a referral back to the Court of Appeal by the Criminal Cases Review Commission only two years ago.

The trial revealed a worrying lack of unanimity among the experts. "Three different squads of officers came up with three different answers about the same print," Mansfield recalls. As the trial rolled on, more experts were called upon to give their view on the mark and eventually 20 fingerprint officers were involved. "At the end of the day Lord Justice Swinton-Thomas, who was the main judge, threw up his hands and said, ‘This is impossible for us to resolve'." Peter Swann, a leading fingerprint expert and a former adviser to the Home Office, was a defence expert witness in the case. "I was the one who stuck for eight years saying that the mark was of no value," he says. "It couldn't possibly be identified to Danny McNamee." While Mansfield claims that cases like McNamee and McKie "rock the foundations" of the forensic world, Swann supports the retention of the old standard. "There is nothing wrong with the system otherwise it would not have been used for 50 years," he says.

If the criteria are met and the administrative procedures are adhered to there is little room for errors, he contends. "I know that one or two have got through in recent times but they have been down to administrative errors," he says. "There is no doubt that identification by fingerprint is unique." There is nothing "inherently wrong" with the current system, agrees Ben Gunn, Chief Constable of Cambridgeshire and ACPO spokesman on forensic science. But nor is there any basis in law, logic, science or fact to support it either, he adds. Gunn cites the case of Shirley McKie, in which at least three experts claimed that the standard was met. He says: "Now that is a misidentification and it has proved to be so. So where is the validity of 16 points where human error can actually come to the same conclusion, albeit wrongly?" A move to a non-numerical system has nothing to do with relaxation of standards, nor is it a drive to increase detection or conviction rates, Gunn insists. Instead, the new system will backed by new "standardised and accredited quality procedures".

Nevertheless, what unites Swann and Mansfield is their fear that abandoning a point system will lead to undermining of standards. "Deregulation of the field is quite the reverse of what scientific examination is all about," says Swann. "It should be about protocols and ensuring all the experts meet the same standards.'