Prosecutors have convicted suspects with fingerprints for more than a century - but the once unshakable certainty of fingerprint experts might be crumbling.
When a Baltimore County judge barred fingerprint evidence in a murder case this week, her concerns echoed those of critics who say fingerprint identification remains as much art as science.
"Basically, it's 'trust me' forensic science, and that's scary," said Sandy L. Zabell, a Northwestern University mathematician who studies how statistics are used in court cases.
Typically, when detectives recover a fingerprint at a crime scene, experts scan computer databases for possible matches among the millions of known prints. But a human, combining rational analysis and gut feeling, makes the final call.
With that comes the inevitable potential for human error. But critics say that fingerprint examiners, unlike scientists in any other discipline, refuse to acknowledge any uncertainty.
Proponents counter that fingerprint examiners use a rigorous, time-tested process. When adhered to carefully, it leaves "zero" possibility that a suspect will be wrongfully linked to a crime, according to the International Association for Identification, which certifies fingerprint examiners.
Recent cases of mistaken fingerprint identity, however, have raised concerns that these supposedly foolproof methodologies may be prone to human error. As a result, a few judges now entertain defense attorneys' attacks on the underlying principles of fingerprinting.
This emerging skepticism is generating speculation that fingerprints could join bite marks, handwriting, hair and gunshot residue on the list of forensic evidence that defense attorneys challenge successfully before judges increasingly open to such arguments.
Baltimore County Circuit Judge Susan M. Souder is thought to be the first jurist in the United States to have excluded fingerprint evidence from a murder trial based on broad concerns about reliability. She wrote that she is not convinced that the methods are based on "a reliable factual foundation," as Maryland law requires.
The methodology, known as Analysis, Comparison, Evaluation and Verification, or ACE-V, is employed by the FBI and crime labs around the country.
Examiners focus on patterns left by "friction ridges," the raised lines on the skin of the fingers and palm that form the whorls, arches and loops in fingerprints.
As a first step, an examiner studies a crime scene print to determine if enough of the finger surface was captured and whether it's clear enough to make a comparison. If so, he compares the crime scene print with high-quality prints inked from a suspect. If the overall flow of lines matches, the examiner then looks for "points of similarity," tiny pieces of the pattern at the same place on each print.
If there is enough similarity, the examiner declares a definitive match. Then another examiner goes through the same process - and either verifies or disputes the first examiner's findings. Unless they agree, the prints are not classified as a match.
Given that thoroughness, Thomas P. Mauriello, an adjunct professor of criminology and criminal justice at the University of Maryland, College Park, described challenges to the reliability of the method as "absurd."
"You can question the way a print was taken or question the chain of custody," he said. "But questioning the science is hard to do, because we have never, ever found two people with the same fingerprint."
He and other proponents say the technique has weathered the test of time. "If this didn't work almost 100 percent, there would be a hue and cry from people who were identified but it wasn't them," said Joseph Polski, chief operations officer for the IAI.
The case of Brandon Mayfield, however, spurred critics to charge that fingerprinting advocates are overly confident.
After a March 2004 terrorist bombing killed 191 people in a train station in Madrid, Spanish investigators lifted a print from a bag of detonators and provided a copy to U.S. authorities. FBI computers matched the print to Mayfield, a 37-year-old Muslim lawyer living in Portland, Ore.
A senior FBI examiner who identified more than 15 points of similarity between the Spanish print and Mayfield's declared a "100 percent identification." Two other FBI examiners and an outside expert confirmed the match.
Spanish investigators disagreed, but the FBI arrested Mayfield anyway. Mayfield was released two weeks later after Spanish authorities matched the latent print to an Algerian and matched that man's DNA to that found in a cottage where the bombs were built.
FBI officials later told members of Congress that the mistake stemmed from the poor quality of a digital copy of the latent fingerprint.
"It turns out they are all wrong, and nobody would have known," said Robert Epstein, an assistant federal public defender in Philadelphia. "But for the fortuity that you had investigators in another country looking at it, Mayfield would probably still be in jail."
FBI spokeswoman Ann Todd said agency experts had testified in the Baltimore County case and that it would be inappropriate to comment about the Mayfield case or related issues.
But Epstein said the Mayfield case shines long-overdue light on the limitations of fingerprinting - and the need for research into the reliability of its methods and practitioners. "They have never done any real studies or any scientific testing to figure this out," he said.
One major concern, he said, is the absolute certainty that fingerprint examiners claim in the courtroom. DNA experts, for example, typically present evidence to judges and juries by noting the statistical probability of a match between two DNA samples.
Fingerprint examiners, however, use a binary, all-or-nothing approach. After determining whether a recovered print is good enough to make a decision, they decide whether it is a match for the defendant: a simple yes or no.
"They dramatically oversell their opinion," said Epstein. "The claim of 100 percent certainty is absurd. There is no recognition [of] how blurry the lines are between those choices, the amount of gray area there is in making these decisions."
Critics also attack the argument that no two people have the same fingerprint. Often, crime scene investigators recover only partial prints - and with a sample that small, it is possible that two similar but different prints can be confused as a match.
Confounding the problem, critics say, is a lack of standards on how many points of similarity make a match. Thus, fingerprint experts cannot be held accountable, and judges, juries and attorneys have no choice but to trust an examiner's opinion.
Epstein said researchers should determine the potential accuracy of fingerprint matches, given the condition of the print, the number of possible matches and other variables such as the surface from which the print was lifted.
"Then we might be in a better place to evaluate the examiner's opinion," he said. "Maybe if they have 25 points they are right 95 percent of the time. If they have 15 points they are less likely to be sure."
Proponents say fingerprint examiners follow rigorous procedures and that rigid rules concerning a match would undermine their intuitive judgment.
IAI's Polski compared print experts to medical pathologists who determine whether cells taken from a patient are cancerous.
"The pathologist can look at a cell and tell you it's cancerous," he said. "But they can't begin to tell you exactly why they know. To a certain extent, that is what the trained fingerprint examiner is going to tell you."
Mauriello of the University of Maryland said the Mayfield case and other instances of mistaken fingerprint identity are the result of rare cases of human error, not a problem with procedures or training.
But Zabell remains skeptical. "From a scientific perspective, just because it's been used to convict someone doesn't mean its accurate," he said.