Before DNA evidence took center stage on TV shows like CSI and Bones, fingerprint evidence was prized above all other forms — as far as the public was concerned, prints were the cornerstone of forensic science. In many respects, they still are: most lawyers and judges still consider fingerprints the easiest form of forensic evidence for juries to understand, and they still appear in more criminal court cases than most other types of forensic evidence. But the strength of this fingerprint evidence, and its standing in the forensic community, is far less cut and dry. And despite its use in criminal cases since the late 19th century, scientists have only just begun to examine how unique our fingerprints truly are — and how good forensic examiners are at telling them apart.
"Up until 2011, there was no scientific data on the accuracy of fingerprint matches," says Simon Cole, a criminology professor at the University of California, Irvine, and the author of a book on the history of fingerprint identification. "That year, we had the first two major studies and they showed that it's highly accurate in the test conditions."
Two people were falsely implicated in criminal cases in Los Angeles in 2008
By highly accurate, Cole means that about one in 1,000 fingerprint examinations inaccurately matches a print with a previous offender. Conversely, about 7.5 percent of examinations falsely clear someone as a potential match. This is near perfect, Cole says, and it indicates that forensic examiners tend to be conservative when matching the fingerprints found in crime scenes with those in databases. But those results were obtained under laboratory test conditions, so figuring out "to what extent those general rates would match those we see in real life," Cole says, is "a tricky question." And unfortunately, real-life errors do occur: in Los Angeles, for example, forensic examiners falsely implicated two people in criminal cases in 2008.
That's why members of the forensic community are seeking to alter the future of fingerprinting by changing the way the evidence is obtained and evaluated — not to mention how it's reported in the courtroom.
Improving the technology
Currently, fingerprints are obtained from surfaces like cardboard, leather and glass, using tools such as powders, gels, and lifting tape. They're mainly used to place a person at the scene of a crime. But Simona Francese, a forensic scientist at Sheffield Hallam University, thinks that forensic examiners can get a lot more information out of a fingerprint. "I am working on the development of a technology that enables researchers to detect molecules in fingerprints," Francese says.
Getting more information out of a fingerprint
In so doing, she hopes investigators will be able to tell if the fingerprint's owner smokes cigarettes or takes drugs. The technique, she says, can even determine if a person is a genetic male or a genetic female, because each sex produces a unique combinations of peptides — short chains of amino acids — in their sweat. By using a statistical model, Francese says that "researchers can predict the sex of the person with 85 percent accuracy." She even thinks they might be able to predict a person's ethnicity using the same method in the near future.
"Researchers can predict the sex of the person with 85 percent accuracy."
The technique that Francese and her colleagues have developed — which is already used by police officers in West Yorkshire, UK — doesn't just add more information to a suspect's profile. It also improves the reading of the print itself, because it can be used to reconstruct multiple images of a fingermark. "You can map the location of various kinds of molecules located on those ridges," she says, and tease them apart to create a number of fingerprint reconstructions that increase the accuracy of an examiner's reading.
But the method won't soon be ready for court, Francese says. Instead, it will largely be used to provide investigators with supplementary information they can use to create a suspect profile. And researchers still want to do more studies to find out if there are limitations to fingerprint sex determination. "We will need to study people with diseases or people that take medications," Francese notes, to "see if the profile survives beyond these disturbing elements."
Other projects are focusing on how investigators develop fingermarks in the field, because obtaining a fingerprint from a given surface is still considered impossible in many situations. One team of researchers in Israel, for instance, is currently trying to develop fingerprints on rocks — surfaces that investigators usually don't sample because of a low success rate. Another project is even more ambitious, as researchers in Spain are trying to develop prints that were formed underwater.
"Some of the elements that make fingermarks — those that are soluble — are lost when they come in contact with water," says Ana Castello, a forensic science researcher at The University of Valencia who worked on the project. "There are several publications on the treatment of fingerprints on surfaces that have been wet after depositing the fingermarks, but not when these fingermarks were generated underwater." Her technique, she says, makes use of various powders, including what's known as small particle reagent — a powder that investigators already use to lift prints off slightly wet surfaces. To develop a fingerprint that was formed in water, the object in question needs to be dried out properly, and the surface can only be made out of glass or plastic.
Still, Castello thinks there's a future for underwater fingerprints in forensics — and a future for technological improvements in forensics as a whole. "We need to know more and better methods that can help us get as close as possible to the truth," Castello says. This quest, she says "represents nothing more and nothing less than helping bring criminals to justice."
Changing attitudes, assessing odds
But bringing the right people to justice can only be achieved if examiners properly report the reliability of their findings. "In the past, fingerprint testimony has been badly overstated in court, but now we are working in the right direction," says Edward Imwinkelried, a law professor at the University of California, Davis, who has written about the need for fingerprint matching technique reform.
Imwinkelried credits two reports released by the US National Institute of Standards and Technology in 2009 and 2012 — both of which outline steps for examiners to improve fingerprint matching — for the change in attitude he's witnessed in the forensic community. "Some now say things like ‘you should regard my testimony as an opinion and not scientific fact.'" This is a big improvement, Imwinkelried explains, because examiners used to testify that they could match a print to a specific person, "to the exclusion of everyone else."
But David Kaye, a law professor at Penn State, says the change isn't universal. "How many people are following those recommendations, I don't know," he says. "And does it make any difference to remove a sentence that says ‘to the exclusion of everyone else?' — it seems more like a cosmetic change, really."
Quantifying the reliability of fingerprints so jurors can understand
Kaye represents a part of the legal community that thinks juries still have trouble evaluating forensic expert testimony. Views like these are why Sargur Srihari, a computer scientist at the University at Buffalo who has served on a national committee that sought to improve forensics in the US, has developed a way of quantifying the reliability of fingerprint evidence that's easier for juries to understand. "We all live our lives with odds," Srihari says. "People understand them." If a juror knows what the odds are that a print matches that of a given suspect, he says, they'll be able to gage the weight that each testimony deserves in their final decision.
"When an examiner looks at a print, they have to take into account not only the similarity of the two fingerprints, but also the rarity of those similarities," Srihari says. So he used population frequency data to determine how often one might encounter a set of fingerprint features, and how independent these features are from each other. Armed with this math, Srihari says, jurors will have the tools they need to make sense of the science.
Unfortunately, this approach assumes that jurors can understand what odds like 5 in 1000 or 50 to 1 actually mean — and some studies have already shown that jurors can have trouble understanding probabilities in courtroom settings. Moreover, some fingerprint examiners are reticent to include probabilities in their work. "The fingerprint community has been quite conservative about embracing statistical models," Kaye says. "And there are examiners who worry that they will be replaced [by computers]."
"That experience can be flawed."
But forensic examiners will undoubtedly be forced to embrace new technologies as they become more accessible — and more accurate than examiners themselves. "Fingerprint, handwriting, and tire tread evidence all involve experts who come in and testify based on their experience," Srihari says. On that basis alone, an examiner can testify that a fingerprint on the crime scene belongs to a certain suspect. "But that experience can be flawed," he says, "and you can have the same flawed experience for over 20 years."
"Well-trained analysts have been known to make mistakes," Kaye adds, "but fingerprint evidence is generally good." The hard part, he says, "is figuring out when it's not, and presenting it for what its worth — and no more than that."