Science: Seduced by the gene genie: The Government wants criminals to be genetically fingerprinted. But, warns David Pringle, a DNA database may put the innocent at risk

Tens of millions of pounds of taxpayers' money could be wasted because the Home Secretary, Michael Howard, is pushing ahead too quickly with plans for a national criminal DNA database. Government and MPs may be placing too much faith in the power of DNA profiling to fight crime because they have not fully understood the limitations of the technology.

According to Professor Sir Alec Jeffreys, who invented DNA fingerprinting, the Government's plans are premature: 'As soon as you go for a major national database, that locks the technology. The current technology is still evolving - it will be completely superseded over the next five years.' Professor Jeffreys also warned that the DNA analysis might reveal a person's medical condition - information the police should have no right to know.

The Home Office's pilot study for the national database will be completed by the end of April - only two months after it was announced. The database will contain DNA profiles from unsolved crimes and convicted criminals. The breakneck speed reflects the importance the Government places on DNA as a forensic weapon. Different techniques of DNA profiling produce mutually incompatible results, so any future change in technique will mean that hundreds of thousands of offenders' profiles will either have to be scrapped or retaken.

Even before the Government started designing its database, DNA evidence was being challenged at the highest judicial level. Last December the Appeal Court ordered the retrial of Andrew Deen, convicted of three rapes in 1990. Quashing his 16-year sentence, the judges identified major flaws in the use of DNA evidence at the original trial. A forensic expert had implied that the strength of the DNA evidence proved that Deen was the source of the 'crime sample' semen. The appeal court decided that this type of assertion could not be justified by DNA evidence alone. The court also heard evidence from a prominent German scientist, Professor Bernd Brinkmann, who found the DNA evidence to be 10,000 times less incriminating than the prosecution had suggested.

Yet if the current Criminal Justice Bill becomes law, anyone charged with a 'recordable' offence - one that is entered on a criminal record - could be forced to provide a DNA sample. According to the Home Office, one third of males have such a record by the age of 30.

The Government estimates that each DNA sample will cost between pounds 65 and pounds 80 to process. If all 500,000 recordable crimes committed each year lead to DNA analysis, then the annual cost will exceed pounds 30m. By comparison, the budget for the whole of Britain's forensic science service last year was about pounds 43m. The MPs who examined the Bill in committee appear to have been unaware of this potential financial burden. And most MPs seem unaware that today's database may be overtaken by tomorrow's technology.

DNA profiling, or typing, highlights 'markers' which occur at a number of locations (loci) on a chromosome. These markers are simply a characteristic sequence of chemicals, repeated many times. The length of the repeated sequence varies greatly between individuals. The most common technique is the 'single-locus' profile, which displays two sequences at each locus, one inherited from each parent. Single-

locus markers have no known biological purpose: they can identify a person, but indicate nothing about their genetic make-up.

New DNA technology will ultimately allow DNA typing of a single cell instead of the many thousands required at present. It relies on a chemical multiplication process known as the polymerase chain reaction, or PCR. The markers that are ideal for the PCR process are still being researched. Some lie within genes themselves, others in the regions between them.

In Germany, typing DNA markers within genes is banned on the grounds that such markers could reveal medical information. In Canada, by contrast, one of the regions used for PCR typing is inside the 'androgen receptor' gene, which regulates masculine secondary sexual characteristics. Defects can lead to conditions such as feminisation and muscular atrophy.

Professor Jeffreys rejects the Canadian approach: 'That would be prying into the potential disease status of a person and the police have no right to that sort of information.'

There are other difficulties. An offenders' database with hundreds of thousands of entries highlights a key problem of DNA evidence. This relates to 'match probability'. This is the chance that someone randomly picked off the street will have the same DNA profile as the crime stain left, say, at the scene of a rape.

With single-locus profiling, the chance that two unrelated people share sequences of the same length at a particular locus might be, for example, one in 30. Generally, forensic laboratories probe sequences at four loci for identification purposes. Provided that each sequence of DNA is independent - not linked like ginger hair and freckles - then the chance that two people will have four sets of identical sequences will be 1:30 raised to the power of four. This is 1:810,000.

When presented in court, such numbers can, at first sight, be very persuasive. If a suspect's DNA profile matches that of a crime stain, the court might be told that the chance of a random match is 1:810,000. The point so often misunderstood is that if the suspect comes from a large pool of potential or hypothetical offenders - all young men in a locality, for example - then the probability of innocence can be much greater than the DNA match probability. The probability of one completely coincidental but innocent match occurring in a database containing 400,000 profiles could be as high as 50 per cent (ie 1:810,000 multiplied by 400,000). There would need to be corroborative evidence in order to convict safely a suspect found in this manner.

Peter Donnelly, professor of statistics at Queen Mary and Westfield College, London, warns: 'There is a grave danger with cases from a database. The analysis you should do is clear, but if you don't do it properly, then you'll convict people who shouldn't be convicted.'

Many MPs have been seduced by the apparent forensic power of DNA typing into calling for a national database of the entire population. The call is echoed by Professor Colin Tapper of Oxford University, a leading authority on criminal evidence. 'I wouldn't be averse to everyone being entered into a DNA database at birth, with a view to detecting crime, and preventing innocent people from being charged.'

This is causing disquiet among civil rights groups. 'There is a sense that anything is excusable in the name of law and order,' says Simon Davies, of Privacy International. 'If we're not careful, we'll be required to disclose our own DNA code at every turn - to get a job, to get a loan, to get insurance.'