An
examiner at the Wisconsin State Crime Lab tries to identify fingerprints.
The
idea that people could be identified with fingerprints was proposed back
in 1880 by Henry Faulds, a Scottish missionary. His proposal went nowhere
until it was taken up in 1888 by Francis Galton. Galton, a biologist and
cousin of Charles Darwin, argued that the tiny ridges and canyons on fingers
could finger criminals based on objects they'd touched at crime scenes.
By 1905, the director of Scotland Yard was a believer in fingerprinting, and the technique was used in the 1902 murder of an elderly couple (see "Fingerprints..." in the bibliography).
The FBI championed fingerprints while opening its highly hyped crime lab in 1932, and the technique became a key element in forensics -- the science of detective work. During the 20th century, fingerprints helped convict an uncountable number of defendants, and national and international databases gathered hundreds of millions of prints.
Oddly, in all that time, the central premise of fingerprint identification -- that every print is unique -- was never systematically tested, even if identical prints are obviously at least extremely rare. Since the practical question of whether examiners can make accurate identifications has been systematically tested, the question becomes: Is fingerprint identification a science, or just a gimmick that's had a run of good luck and sterling PR?
Send
lawyers, guns and money
This
belated question reflects the demands of the 1993 Daubert ruling, which
held that scientific evidence must be science rather than belief or, say,
voodoo. Do the conclusion flow from the facts? What was the method? Are
the results statistically valid? Can they be replicated? Are the assertions
subject to disproof, as all scientific assertions must be? Was a written,
approved protocol used to collect and analyze the evidence?
Faigman, of Hastings College of Law, says Daubert wrought a "revolution.... Now courts have to ask, 'Where are your data to support what you say you can do, or that supports the validity of the technique generally?'"
Daubert poses difficulties for fingerprint identification, he adds. The technique "easily met" the old standard of general acceptance in the field, "because the only field was fingerprint analysts, and their livelihood depended on it being accepted in court as valid."
Now, law professor Faigman says, "The proponent of expert testimony has burden of proving reliability and validity, and failure to demonstrate that leaves it inadmissible."
But the ultimate proposition of fingerprint identification -- that no two people have identical prints -- can only be proven by snagging and comparing prints from the human population, past, present and future.
That's an impossible assignment. Much easier would be showing that identical prints are statistically quite rare. That was the goal of a recent effort of the National Institute of Justice, which earmarked $500,000 for grants on fingerprint procedures. To satisfy the grant, it specified that "Procedures must be tested statistically in order to demonstrate that following [them] allows analysts to produce correct results with acceptable error rates. This has not been done." The request amounted to an official admission that fingerprinting rested on a shaky scientific foundation, and it supplied ammunition to defense attorneys questioning fingerprint identifications. However, the program's future is uncertain (see "Fact Is..." in the bibliography)
It
ain't CSI, but this truck makes it much simpler to get clean evidence
from a crime scene.
And mistakes happen. Simon Cole, author of a forthcoming book on fingerprinting, notes that Richard Jackson, a Pennsylvania prisoner, was released in 2000 from a life sentence after the prosecution conceded an error in the fingerprint ID that nailed him (see "The Myth..." in the bibliography).
Here
comes the judge
Jerome Geurts, director of the Wisconsin State Crime Laboratory in Madison,
still puts his faith in fingerprints. "There have been hundreds of millions
of prints catalogued to date, now with the AFIS (Automatic Fingerprint
Identification System) they are being loaded into master files, and they
have yet to find two that match." Geurts adds that even identical twins
have non-identical prints, and that the technique survived a Daubert challenge
in federal court.
However, Faigman calls that decision "badly reasoned." Although he says judges are "extremely resistant" to exclude fingerprint evidence, he expects them to follow the "courageous leaders... All you need is one well-known judge to exclude fingerprint evidence and others will follow... Some judge will say, 'You produce the research ... if you don't produce it, I won't allow the evidence.'"
That would exclude a type of evidence that's been extremely useful for almost 100 years. It sounds radical, but some judges already have excluded analysis of handwriting and hair, two forensic sciences that could not meet the Daubert standard.
I
got sued, Babe
We've posed the fingerprint dispute in perfect terms: proponents say fingerprints
can infallibly identify criminals while skeptics say the entire basis
for fingerprinting is unproven.
Perhaps the important debate lies between the extremes. Both sides recognize, for example, that the ultimate assumption -- that every print is unique -- can never be proven as a practical matter. And skeptics recognize that prints are a valuable identification tool -- if not a conclusive one.
Is there a better way to do fingerprinting?
Bibliography | Credits | Feedback | Search