Breakthrough research by an English geneticist 25 years ago revolutionised the investigation of crimes and the prosecution of the perpetrators, but DNA profiling is not foolproof. No one knows if Archimedes really did run down the street of his hometown shrieking "eureka" - Greek for "I have found it" - after discovering his celebrated law of buoyancy. What is certain is that in science, such moments are rare indeed. Most of the time, the frontiers of knowledge advance at a glacial pace, with researchers finding that - in the words of the American Nobel-Prize-winning physicist Murray Gell-Mann - their most valuable tool is the wastebasket.
Small wonder, then, that scientists are this month celebrating a genuine eureka moment, which took place exactly 25 years ago. On that September morning in 1984, a 30-something geneticist at the University of Leicester in England by the name of Alec Jeffreys was examining the DNA of three members of the same family. As expected, he found many similarities; what stunned him were the many differences. It seemed as if certain parts of DNA might act as a kind of genetic bar code, unique to everyone except identical twins.
Dr Jeffreys quickly realised such a "DNA profile" would have a host of applications, allowing scientists to identify people from minute traces of the genetic molecule. Within six months, DNA profiling was put to use on its first case, resolving a dispute about the paternity of a boy involved in a UK immigration application. But the power of DNA profiling first entered public consciousness in 1987, when it was used to investigate the rape and murder of two teenage girls in a small village just down the road from Dr Jeffreys' laboratory.
A 17-year-old had confessed to the double murder, and police decided to use the newly patented DNA technique to confirm his guilt. To their astonishment, the samples of DNA found at the crime scene showed no similarities to those from the teenager. Realising that he was a vulnerable attention-seeker, police released him - the first of hundreds of people to have their innocence confirmed by DNA evidence.
With the culprit still at large, detectives then announced that DNA samples would be taken from thousands of local men in an attempt to track down the killer. Among them was a 27-year old bakery worker named Colin Pitchfork, who persuaded a friend to take the test for him. When the deception was uncovered, police obtained a DNA sample from Pitchfork - which revealed his guilt. He is now serving a 28-year prison sentence.
A more spectacular showcase for the powers of DNA profiling is hard to imagine. Sure enough, the technique began to acquire an almost mesmerising reputation for its ability to implicate or exonerate. It is an image underpinned by the fact that DNA profiles seem able to put precise odds on the innocence of the accused. Forensic scientists began presenting juries with DNA evidence that they claimed showed that the chances of the accused being innocent were just one in a million, or even less.
In fact, and contrary to what even many scientists believe, DNA profiling can do no such thing. What it actually gives are the odds of getting so good a match as that found, on the assumption that the accused is innocent. Yet a moment's thought shows this is quite different from what juries are supposed to assess, namely, the odds of the person being innocent, given the DNA evidence. Confusing the two is akin to a doctor believing that the near-certainty that patients will have spots given they've got measles implies the near-certainty of having measles, given the presence of spots, which is patently not the case.
The only way to resolve this mix-up - known to statisticians as the prosecutor's fallacy - is to take account of any other evidence relevant to the case. And if there is none, even apparently compelling DNA evidence can be utterly undermined. Despite repeated warnings by experts, it's far from clear that most juries - or lawyers, for that matter - yet understand the precise meaning of DNA evidence. Worse, surveys suggest the sheer size of the odds presented in DNA cases may lead juries to ignore other forms of evidence, unaware of the dangers of doing so.
There is an even more pernicious problem with DNA evidence: human error. Over the years, there have been many cases of compelling DNA evidence implicating people who simply could not have been involved in the crime. In 2003, an Australian woman found herself being questioned for the murder of a toddler when her DNA profile was found to match crime scene samples, with the odds against a fluke apparently less than one in 100 billion. The absence of any other evidence linking the woman to the crime prompted police to investigate further. They found that she had herself been the victim of a crime investigated years earlier by the same forensic laboratory - and that her DNA had contaminated samples of clothing belonging to the toddler held in the same lab.
Perhaps the most shocking DNA blunder emerged this year, when police in Germany finally solved the mystery of the female serial killer dubbed the Phantom of Heilbronn. Suspected of up to six murders and dozens of other crimes dating back to the early 1990s, the Phantom had been identified via the DNA she left on a host of crime scene samples, from cars to a kitchen drawer. Despite the wealth of evidence, police could not identify the culprit. In March, the reason became clear: the female serial killer did not exist. The "Phantom" is now believed to be nothing but the result of wholesale contamination by DNA on the cotton swabs used by investigating officers.
There is no doubt that DNA profiling is the single biggest breakthrough in forensic science since the development of fingerprinting more than a century ago. Since Professor Jeffreys's eureka moment 25 years ago, it has brought thousands of criminals to justice around the world. It has also brought justice to thousands wrongly accused of crimes, including some facing capital punishment. Arguably, one of the biggest injustices surrounding DNA profiling is that it has yet to bring Professor Jeffreys a Nobel Prize, an omission that may yet be rectified when the latest prizes are announced next month.
Even if he does win this ultimate scientific accolade, we can expect the famously self-effacing geneticist to continue to stress the importance of wielding his discovery with care. "When DNA profiling was first commercialised, I was shocked to hear that one of the companies involved was describing the technology as infallible," he recently told an interviewer. "There is always room for human error."
Robert Matthews is Visiting Reader in Science at Aston University, Birmingham, England