Skip to main content

Fingerprint evidence -- under judicial assault -- unlikely to be replaced by DNA profiling for criminal identification, says Cornell researcher

Media Contact

Media Relations Office

Fingerprint identification, which recently was ruled by a Philadelphia federal judge to be scientifically flawed as evidence, is unlikely to be replaced by DNA profiling in the courts, says a Cornell University researcher. The main reason, he says: Police have come to rely on fingerprint analysis so heavily in presenting evidence.

"There are so many cases in which there are no evidentiary equivalents, including DNA profiling. Practical reasons militate against wholesale rejection of fingerprinting, and I expect that the FBI and other organizations will try to upgrade its scientific credentials," says Michael Lynch, professor of science and technology studies at Cornell.

Lynch and Simon Cole, a Cornell visiting scientist, have been awarded a $144,000 grant by the National Science Foundation to conduct a yearlong study comparing the scientific histories of DNA profiling and fingerprinting and the sociological implications of the two techniques in their use as evidence in placing criminal defendants at crime scenes. In 1995 Lynch made a study of the first national database of DNA profiles, set up by the British government. Cole is the author of the book Suspect Identities: A History of Fingerprinting and Criminal Identification, based on his 1998 Cornell doctoral dissertation. He also is affiliated with the John Jay College of Criminal Justice in New York City.

Fingerprint classification is based on analyzing the ridge characteristics on fingertips, such as places where ridges split or stop, or even tiny details on individual ridges. DNA profiling, which was introduced in the mid-1980s, is used to genetically compare bodily traces, such as blood and hair, found at a crime scene with samples taken from suspects. State and national DNA databases are under construction worldwide. The two researchers' study is under way as fingerprint evidence is under judicial assault. On Jan. 7 U.S. District Judge Louis H. Pollak ruled that fingerprint evidence, which has been irrefutable in courtrooms for nearly a century, does not meet standards for scientific testimony, and that analysts cannot testify in a trial that a suspect's prints match those found at a crime scene. The judge called into question the techniques for matching fingerprints as being subjective. The ruling is being viewed by legal experts as a landmark opinion.

However, cautions Lynch, DNA profiling itself is not error-free. He adds, "Although DNA profiling probably is improving, it is unknown how many times errors appear. Because it is practiced now by so many different organizations, questions arise about labs handling samples correctly and whether police labs know what they are doing."

Lynch believes that "the courts are confusing the issue by making the identification with science so important. Whether fingerprinting is science or not is beside the point. The question is, is it good evidence?" On the other hand, he concedes that DNA profiling, springing as it does from biology and biomedical labs, has firm scientific credentials, whereas fingerprinting "is home-grown police science."

Cole describes fingerprinting as a technique totally lacking in statistics, and he agrees with Judge Pollak's decision that fingerprint analysis is unscientific. DNA profiling, on the other hand, is based on statistics, or probabilities: Analysts attach a probability that a bodily trace collected at a crime scene matches a sample taken from a suspect. But, he says, because fingerprint analysts don't have such numbers to use, they don't admit to the likelihood of error and thus assert that the error rate is zero.

Says Lynch, "Unlike DNA profiling, where procedures for probability have been established, fingerprinting has no such probabilities established. The assumption has been that no two fingerprints are alike. But since it's impossible to compare the fingerprints of everyone in the world, this assumption still stands. Thus, saying that fingerprinting is error-free is ambiguous."

Cole wrote in The New York Times last year, "…the relevant question isn't whether fingerprints could ever be exactly alike -- it's whether they are ever similar enough to fool a fingerprint examiner. And the answer, it's increasingly, unnervingly clear, is a resounding yes." The two researchers' study will not make recommendations on whether probabilities should be developed for fingerprinting but will document why fingerprinting has avoided probability studies over the decades and the consequences of this resulting lack of data. They also will interview participants in The Innocence Project at the Cardozo School of Law in New York City, an ongoing effort using DNA profiling to reopen criminal convictions obtained in the pre-DNA era.

However, Lynch warns, "Fingerprinting probabilities could result in a nightmare of argument that the courts are not equipped to deal with, yet the arguments now seem to be going the way of quantitative backing for fingerprinting."