IN THE 19th century, Mark Twain played a key role in popularising the then newfangled method of fingerprinting. 鈥淓very human being carries with him from his cradle to his grave , and by which he can always be identified 鈥� and that without shade of doubt or question,鈥� he wrote.
Until recently, most of us believed that this meant a fingerprint match from a crime scene was proof of guilt. That impression has been cemented by their widespread use, by US immigration services, for instance. But in recent years, several miscarriages of justice have highlighted how fingerprints can 鈥� and occasionally do 鈥� finger the wrong person. Studies have also shown that fingerprint analysts can come to different conclusions over whether two prints match, and even the same analyst may not make the same decision about two prints every time.
Many fingerprint analysts are also waking up to the fact that if they want to be called forensic scientists they鈥檙e going to have to make their work more scientific.
Advertisement
鈥淚f they want to be called forensic scientists they鈥檙e going to have to make their work more scientific鈥�
This week, we report on some of the first moves in this direction, with the development of software that is able to apply statistical weight to fingerprint evidence 鈥� in much the same way that statistics are used to quantify the likelihood of a match when DNA evidence is presented in court (see 鈥淔ingerprint evidence to go scientific at last鈥�).
This is a monumental shift, and one which could finally force examiners to give up the outdated notion that a crime scene print could only match a single individual to the exclusion of all others. The researchers involved 鈥� some of whom are fingerprint analysts themselves 鈥� should be applauded for their candid, sensible and scientific approach to fingerprint identification.
In order for such tools to find practical applications within the courtroom, however, they will need to be validated by the analysts who will use them. We also need to be sure that the statistics this software produces are watertight, so as to avoid a repeat of the bitter squabbles that still haunt the use of DNA evidence.
The researchers also need to convince everyone who works with fingerprint evidence 鈥� including agencies such as the FBI 鈥� that such models won鈥檛 undermine their roles. This will require the biggest shift in thinking of all.
Sometimes fingerprint analysts can feel under pressure from the court to come up with unequivocal results. So analysts need to ask themselves a question: would I treat the samples before me in a different way if they were unconnected with a court case? For the sake of forensics, the answer should always be no.