Forensic Explorations Below Ground: Profile James E. Starrs

Among the titles of the articles that Starrs has written or published in his quarterly are:

• "Ink Aging's Expert Derailed in Clickety-Clack Track Litigation"
• "The Death of Napoleon Bonaparte was Homicide: Yes, No, or None of the Above"
• "Facial Reconstruction: Is it Worth the Effort?"
• "A Battle of Blood Spatter Experts and the Shenanigans of a Texas Prosecutor"

A tribute to his Irish humor, Starrs features a column on "Frolicking with Forensics" and generally includes odd little stories from around the world, such as the debate over whether belly button lint has forensic significance. The Review is intended as a way to keep forensic science and legal proceedings accountable to the high ideals of science. If someone is attempting to pass off a method or practice as scientific which has not been tested and found to be so, Starrs will expose it. He also reviews books and combs through them with the same ambition. If a case ought to be examined, he will write up the details. If an injustice has been done or an expert has made a serious lapse in judgment, these, too, get some treatment. In short, Starrs has set for himself the task of pointing out errors and demanding corrections as a way to keep the applied field of forensic science free of junk science or fraudulent claims. But he also gives praise where he feels praise is due.

Judge Louis H. Pollak

Thanks to his public criticism, he has been called into several significant court cases on certain forensic methods that he believes have not been sufficiently tested to pose as science. Among these is the technology for fingerprint analysis. After nearly a century of admissibility in courts across the country, one judge started asking some hard questions early in 2002. In a pretrial ruling for a multiple murder case, Judge Louis H. Pollak, a senior federal judge in the Eastern District Court of Pennsylvania, entertained a motion to evaluate fingerprint evidence against U.S. Supreme Court standards set in 1993 in Daubert V. Merrell Dow Pharmaceuticals. The resulting standards gave judges guidelines to bar junk science from the courtroom. It wasn't the first such attempt.

In 1923 in Frye v. United States, while evaluating a device that was proposed as an accurate measure of blood pressure during deception, the court decided that a scientific claim must be "sufficiently established to have gained general acceptance in the particular field in which it belongs." Across the years, criticism of this standard inspired revisions and in the Federal courts decades later, the Daubert guidelines replaced Frye. In short, when scientific evidence is presented, judges must determine if the theory is testable, its potential error rate is known, it's been peer-reviewed, and it is generally accepted.

A long history of fingerprinting admissibility did not impress Judge Pollak, or the fact that experts often check one another. He found that the technique failed Daubert on three points, notably that the error rate has not been quantified, the "peers" are not scientists, and the millions of prints on record have never been analyzed for possible duplication. Starrs had testified on these points in a prior case that became relevant for this hearing.

"I testified as to what I know about forensic science and science in general and what I can say about the individual problems with respect to fingerprinting." He has long noted this problem, particularly with regard to partial or smudged prints: "There is no scientific basis for fingerprinting as it is presently utilized," he says. "They haven't developed a factual basis for the assertions they make."

Pollak's decision could have had a significant impact on the future of forensic testimony, even in areas like ballistics, documents analysis, and handwriting comparisons, since they too rely on technological interpretation. His ruling focused on whether latent prints, which are generally partial, could be accurately matched to inked prints, which are complete. In a 49-page opinion, he discussed the lack of scientific standards controlling the technique. He wrote that he would allow experts to show how comparisons are made and even to say that no two people have the same print, but not that a specific latent print was made by a specific person. But until fingerprint interpretation becomes a proven science, he said that juries not experts should decide the ultimate issue. If fingerprint comparison is indeed scientific, this report implied, then with sufficient measures enacted, it will pass the test. If it's not, then the technician's testimony in a criminal case ought not to be viewed as definitive. (Judge Pollak subsequently backed off this strong stance, which allowed the other side to relax a little.)

Starrs agrees with the thrust of Pollak's initial position. He has seen experts testify that they found a certain number of points of comparison between a latent print and the prints taken from a suspect - a common practice based on tradition - but he could not see the scientific nature of deciding to just stop in their interpretation at a given number of points, say, 8 or 12.

"Everything is completely at the discretion of the fingerprint expert," he bemoans. "There is no such thing [in fingerprint analysis] as a standardized routine as there is with DNA. You've got a protocol that is established by the FBI and everybody follows it. But just because you can take a fingerprint doesn't mean that you can positively identify it. Just because you can classify it as being A, B, or C doesn't mean you can positively identify it. One doesn't flow from the other. I have no problem with the reliability of the taking of fingerprints and I have no problem with the classifying, but it's in the positive identification that I have a problem." He would like to see a number of methodologies that pass as science in the courtroom but that rely on technological skill be subjected to rigorous testing before being admitted.

Yet he does not just criticize from an academic perspective. He has done some of this testing himself.