Fingerprint Evidence Needs New Tools

The New Scientist carried an article recently on how courts’ use of fingerprints “is on the cusp of a much-needed revolution.” First used to obtain a conviction in Argentina in 1892, fingerprint evidence has long been the paragon of forensic science. Typically, experts will testify that prints taken from a suspect are “a match” with those found at the scene of a crime, offering, however, no “error rate” to accompany their statement, as would be the case if, say, DNA evidence were given. Yet, as the article shows, significant errors are indeed possible.

The problem is not so much one of matching two sets of perfect prints but of judging whether one smudged or partial print, for example, is a match to a more perfect example. (Try your hand at discerning possible matches with the New Scientist a quiz.)

The law school at UCLA has been commissioned by the US National Institute of Justice to look into error rates in fingerprint identification, aiming to have the study assist counsel and the courts. And in Britain forensic scientists are working on a computer program that will give “statistical weight” to fingerprint matches, where, as is often the case, the crime scene prints are less than perfect instances.

My impression from a very brief look at Canadian judgments is that little if any evidence as to the degree of likelihood of a match accompanies the presentation of fingerprint evidence here. However, I’m sure that the criminal law experts among our readers can provide a better overview.