Critics tell experts: Show us the science
By Steve Mills and Flynn McRobertsTribune staff reporters
October 17, 2004
In exonerating scores of prisoners in recent years, new DNA testing has turned an unflattering light on a whole array of forensic evidence. Two of the oldest disciplines have responded to the challenge in dramatically different ways.
The pressure persuaded researchers to test the validity of bullet-matching methods for the first time. But fingerprint examiners, who use perhaps the most common forensic tool, have resisted.
Few doubt that fingerprints are unique, but that agreement has obscured a troubling reality: No research has been done to answer such questions as how much of the partial fingerprints found at crime scenes is needed to reliably declare a match.
"The scientific basis may be there for the whole print, but is it there for that fragment found on the handle of the attache case or on the counter in the kitchen?" said American Academy of Forensic Sciences president Ronald Singer.
"Those are the questions that need to be answered with objective scientific research, not subjective experience or anecdotal evidence."
But the fingerprint community has balked. "Most of the fingerprint people are against that," Singer said, "because they're reluctant to open up themselves to the criticism that what they've been doing is somehow unscientific."
In light of such questions, the research arm of the Justice Department in September 1999 approved a request for studies meant to scientifically validate fingerprinting. But then it stalled.
According to senior officials at the National Institute of Justice, the head of the FBI lab asked the institute to delay its release until the end of a trial in Philadelphia where defense attorneys were arguing there was no scientific basis to support fingerprint matches.
Other institute officials contend that funding and a need to clarify the language of the proposal led to its delay. In any case, shortly after the defendant was convicted in early 2000, the institute released its call for research.
The proposal caused such a stir in fingerprinting circles that the institute's acting director sent out an unusual clarification, backpedaling from any suggestion that fingerprinting was in doubt.
The research, Julie Samuels wrote, was intended only to "further confirm the already existing basis" that fingerprints can be used to identify individuals, noting what she called "the success of the current procedures."
Yet her own agency had issued an assessment of the state of forensics that described the shortcomings of fingerprinting. Such evidence "has historically been `understood' to hold individuality," read the February 1999 report. "However, the theoretical basis for this individuality has had limited study and needs a great deal more work."
The institute's request eventually drew four proposals from researchers, including one submitted by a team led by University of Illinois at Chicago professor Robert Gaensslen, who heads the university's forensic science program.
The Gaensslen team wanted to put fingerprinting on a firmer scientific foundation, exploring such basic questions as whether individuals share fingerprint characteristics.
The institute returned his proposal for standard revisions, which he said he made. Not long after, the call for research died.
But that may be about to change. In an interview last week, the institute's current director, Sarah Hart, said an international panel of experts will meet in the next couple of weeks to discuss what issues ought to be explored in a second call for fingerprint research, which could go out by the end of January.
Hart said the new call for research likely would address a central unanswered question: Can examiners attach statistical probabilities to their declarations that the partial prints found at crime scenes are a match. "It's going to give jurors information about the significance and weight they should attach to fingerprint testimony," she said.
Though fingerprint experts resisted such study for years, researchers are already examining another staple of forensics--firearm identification, the process of matching bullets or casings found at crime scenes with bullets fired from suspects' weapons.
The tricky part about firearm identification is that bullets fired from the same model of gun will share certain characteristics, while other marks will be unique to each gun. The firearm examiner must pinpoint the truly distinguishing marks.
At a private lab in Rockville, Md., Benjamin Bachrach and fellow researchers are trying to insert some science into that subjective exercise.
"The core of the work that we're doing is to show that ballistics matching is not a matter of opinion," said Bachrach, vice president of the lab, Intelligent Automation.
Funded by the same agency that withdrew the request for a fingerprint study, Bachrach's team is test-firing nine models of 9 mm handguns, one of the most commonly used in street crime.
The team already has discovered, for instance, that different brands of bullets are imprinted differently, even when they go through the same barrel. Whether this matters in the ability to declare accurate ballistic matches has yet to be determined.
But Bachrach said he hopes to provide firearm identification experts with the research needed to give their testimony more scientific precision: estimated error rates in the matching of bullets.
Critics of how forensic science is used want to see the same sort of inquiry applied to other disciplines.
"We're saying, `Show me the work,'" said Lisa Steele, a co-chair of the forensic evidence committee of the National Association of Criminal Defense Lawyers. "Otherwise, we're taking their word for it. And that's not good enough in criminal defense."