A single fingerprint found at the scene of a crime is such powerful evidence that it's almost an automatic conviction. Fingerprints never lie: Juries have been told that for more than a century.
But a criminologist at the University of California, Irvine, has documented 22 cases, most involving violent crimes, in which fingerprint evidence turned out to be dead wrong, usually discovered after defendants had served time for crimes they did not commit.
The fingerprints didn't lie. But the experts who matched them with a suspect were wrong, and subsequently had to admit it, according to court records analyzed by Simon Cole, assistant professor of criminology, law and society at Irvine.
Cole's exhaustive research argues that the "zero error rate" claimed by fingerprint experts needs serious retooling. His findings are published in the current issue of the Journal of Criminal Law & Criminology.
The cases include one man who volunteered for hazardous assignments while in prison to earn money to pay for DNA research involving his case. That research eventually cleared him, despite the fact that experts had testified that his fingerprint implicated him in the attempted murder of a police officer.
In another case, a corpse found in the Nevada desert was identified – on the basis of fingerprints – as that of a California woman who later turned up alive.
Most of the cases studied by Cole were covered by the media, partly because they involved sensational events like the Madrid train bombing last year that killed 191 people. Cole suspects there are many other cases, possibly more than 1,000 each year in the United States alone, in which fingerprints have been matched erroneously with the wrong person.
Cole isn't saying that fingerprints are not a useful tool for law enforcement. All he's saying is their "error rate" is more than zero, and some way needs to be found to put their reliability into perspective.
"To tell the jury that it's positive identification, that it's infallible, that it's 100 percent certain is overstating the value of it," Cole said in an interview. "It may well be that it's right 95 percent of the time, but when it's wrong it's very unlikely for us to know about it. So it's likely to result in a miscarriage of justice."
There are multi-layered programs designed to prevent errors, including a requirement for independent verification. Cole found only 22 cases, spanning several decades, but he was limited to cases that have come to public attention.
"An analysis of these cases shows that they are most likely only the tip of the proverbial iceberg of actual cases of fingerprint misattribution," he writes.
In most of the cases he studied, the error did not surface during the routine processing of the case. It surfaced because of "extraordinary circumstances," such as the confession of someone else, which Cole describes as "always a fortuitous and highly unlikely event." So he concludes there are probably many less "fortuitous" cases where the error is never discovered.
Part of the problem stems from the nature of the evidence itself. Fingerprints taken from the scene of a crime (called latent prints) are not always clearly defined.
"They are typically partial, smudged or otherwise distorted," according to Cole.
That would seem a clear path for errors, but experts routinely testify that the "methodological error rate" is zero. How can that be, especially now that several cases have surfaced?
When fingerprint analysts testify that the method is perfect, they are talking about the fact that no two fingerprints are exactly the same, so one and only one person can match the prints. They frequently use mathematics as an analogy.
The equation 2+2=4 is correct. If a mathematician comes up with an answer of five, the method isn't wrong – the mathematician is.
But the chance that the fingerprint analyst is wrong is not taken into account when the evidence is presented to the court, Cole says. Repeatedly, during the 22 cases he documents, the experts testified that the chance of error was essentially zero.
The most celebrated case of mismatching involved a Portland, Ore., lawyer, Brandon Mayfield, who was arrested and held for two weeks as a suspect in the Madrid train bombing last year.
Spanish National Police sent fingerprints from the crime scene to law enforcement agencies around the world, including the FBI, in an attempt to identify the bomber.
Mayfield, in retrospect, may have been a sitting duck. He was a Muslim convert with an Egyptian wife, a United States Army veteran, and he had once represented one of the "Portland Seven" who had pled guilty to conspiracy to wage war against the U.S.
Spanish authorities questioned the match, and the FBI even sent agents to Spain to try to convince them that Mayfield was their man.
Spanish police, however, found another man who was a better match for the prints, and Mayfield was released. At the time of his arrest, Mayfield told authorities he had not been out of the country in 10 years, and didn't even have a passport.
And according to Cole, one of the FBI agents who had insisted the fingerprints were those of Mayfield had been reprimanded for making false attributions in 1969 and 1974.
One of the most troubling cases involves Stephan Cowans, who was convicted of attempted murder in 1997 for allegedly shooting a police officer while fleeing a robbery in Roxbury, Mass. He was implicated in the crime by the testimony of two witnesses, including the victim, and a fingerprint found on a cup.
Cowans insisted he was innocent, but several experts testified that the fingerprint was his. He was convicted and sent to prison.
While in jail he volunteered for "biohazard" work assignments to earn money for DNA tests. Three DNA samples from the mug and from a hat and a sweatshirt discarded by the perpetrator all excluded Cowans.
As a result, the Boston police department re-examined the fingerprint and determined that the match had been an error. It was later found that one of the fingerprint experts who had testified at Cowans' trial had "discovered" the error but concealed it.
Cowans was released, but he served six years in prison for a crime he did not commit.
And he owes his freedom not only to his own perseverance, but to happenstance as well. If the real villain had not been "so obliging" to drink from the cup and discard two pieces of clothing containing his DNA, "it is virtually certain that Cowans would have served his full sentence of 35 years without anyone ever knowing that the fingerprint evidence (and the eyewitness evidence) was erroneous," Cole writes.
It doesn't exactly sound infallible.
Lee Dye's column appears weekly on ABCNEWS.com. A former science writer for the Los Angeles Times, he now lives in Juneau, Alaska.