ADMITTING ERROR AND ADMITTING FINGERPRINTS:
Is There A Sound Factual Basis For The Law?

Earlier this year, federal district judge Louis Pollak barred the government from presenting expert testimony that a criminal defendant's fingerprints matched fingerprints found at the crime scene. The ruling in United States v. Plaza ("Plaza I") was careful and learned, which is hardly surprising. Judge Pollak, who was formerly a law professor and Dean of both the Yale Law School and the University of Pennsylvania Law School, is among the most respected members of the federal judiciary.

What is surprising is that last week, in another careful and learned opinion, Judge Pollak reversed his own earlier decision. In his second opinion in United States v. Plaza ("Plaza II"), Judge Pollak decided that, on second thought, the art of matching fingerprints was sufficiently reliable for the law's purposes after all.

The Plaza decisions show our justice system working at its best. First, a judge showed sufficient courage and independence to rule fingerprint evidence inadmissible, despite the fact that it has been accepted by courts for over a hundred years. Then, faced with new evidence, the judge took the highly unusual step of admitting that he was wrong. Judge Pollak deserves considerable praise for both decisions.

At the same time, however, the underlying issue exposes a well-kept secret of our legal system: It is pervaded by unjustified assumptions. It turns out that reliance on fingerprint evidence is justified. However, as a few other examples will illustrate, much of the law is based on nothing more than guesswork.

The Standard for Admitting Expert Testimony

Federal Rule of Evidence 702 governs the admission of expert testimony in the federal courts. The rule authorizes a trial judge to prevent the jury from hearing scientific or technical expert testimony that is not based on "reliable principles and methods."

The rule was modified to reflect the Supreme Court's rulings in the1993 case of Daubert v. Merrell Dow Pharmaceuticals, Inc.and the 1999 case of Kumho Tire Co. v. Carmichael. These decisions stress that reliability is the touchstone in decisions whether to admit expert testimony, and help trial court judges decide what counts as a "reliable" principle or method.

They set out a number of factors that go into determining reliability, making clear that before admitting expert testimony, judges should ask questions such as the following: Has the methodology been tested? Have papers validating it been published in peer-reviewed journals? What is the error rate of the methodology? Such questions enable the trial judge to distinguish legitimate scientific or technical expertise from junk science.

Plaza I: Rejecting Some Fingerprint Evidence

Is using fingerprint evidence a "reliable" method? We have always assumed so. However, in his Plaza I opinion, Judge Pollak observed that despite its long acceptance by American courts, fingerprint evidence had never been scientifically tested.

The government argued that the courts' century of experience with fingerprint evidence itself counted as a sufficient test of its reliability. However, Judge Pollak rightly rejected this claim. That is not the kind of testing that the Supreme Court had in mind in its Daubert and Kumho Tire opinions.

After all, prior to the Thirteenth Century, English society resolved legal disputes through trial by ordeal--in which those accused of crime could clear their names if, for example, they could carry red-hot irons without singing their flesh. (Another variant of trial by ordeal is hilariously memorialized in a scene in Monty Python and the Holy Grail.) The fact that this form of "testimony" persisted for centuries in no way assures its reliability.

Judge Pollak's Plaza I opinion did not, of course, suggest that fingerprint evidence is no more reliable than trial by ordeal. Nor did it call into question one familiar type of fingerprint evidence: evidence to the effect that two sets of "rolled" fingerprints--that is, fingerprints taken by a police officer or other professional under controlled circumstances--are a match.

Rather, Judge Pollak's Plaza I opinion only addressed efforts to match rolled fingerprints to "latent" fingerprints. Latent fingerprints are sometimes left behind when a person touches an object (such as a gun). Typically, they are less complete and less clear than rolled fingerprints.

Moreover, the Plaza I opinion did not even bar all testimony regarding latent fingerprints. It allowed expert witnesses for the government (and the defense) to testify to the similarities between a latent print and a set of rolled prints. It only barred experts from further testifying that a particular set of prints was or was not a match.

Plaza II: A Change of Mind

Even so limited, however, the Plaza I opinion would have severely limited the ability of the government to obtain a conviction based upon latent fingerprint evidence - which might sometimes be the only type of evidence left at a crime scene by the perpetrator. Jurors who are not themselves experts would have wondered why the actual experts were not testifying as to the existence of a match. If the expert can't even say this is a match, a reasonable juror might have thought, how can I base a conviction on this evidence?

Accordingly, the government asked Judge Pollak to reconsider his ruling. Plaza I had been decided based upon the transcript of a similar motion in a different case. For Plaza II, however, Judge Pollak listened to the live testimony of some of the world's leading fingerprint experts in order to resolve the government's motion for reconsideration.

In the end, Judge Pollak changed his mind because the live testimony revealed that the FBI had in fact conducted a considerable number of tests to determine the reliability of fingerprint evidence. Evidence of the results of these tests allowed Judge Pollak to answer two of the basic Daubert questions in the affirmative: The method had indeed been tested, and its error rate was indeed low.

Indeed, the tests showed a very low error rate, with the errors invariably being "false negatives" rather than "false positives." A false negative means that someone whose fingerprints actually were left on a crime scene is falsely exonerated, whereas a false positive means that an innocent person appears to be guilty. Obviously, a defendant has little to complain about when the methodology at issue in his case produces occasional false negatives but few or no false positives.

In addition, where Judge Pollak's Plaza I opinion had criticized fingerprint matching methodology for its subjectivity, his Plaza II opinion took note of the fact that the FBI uses the same standardized safeguards for matching latent to rolled prints as used in the United Kingdom. Thus, consistent with Daubert and Kumho Tire, the FBI methodology is "generally accepted."

The Exceptional Character of the Plaza Ruling

It is heartening to see a federal judge taking great care to ensure that evidence introduced in court is reliable. Yet it is equally disheartening that Federal Rule of Evidence 702 is not typical. For the most part, the rules applicable in court have not been validated by anything like rigorous scientific standards.

For example, in homicide and civil cases, Federal Rule of Evidence 804(b)(2) permits, as an exception to the hearsay rule, the statement of a deceased witness concerning the cause of his death if the statement was made while the witness was dying. The theory underlying this rule--which codifies an ancient principle of the common law--is that no one would lie about who killed him as he is about to meet his Maker.

Is that true? Is it perhaps equally likely that the minds of people on the verge of death are cloudy? Or that some people would see the "dying declaration" (as the hearsay exception is known) as an opportunity to settle old scores by falsely accusing enemies of homicide? The truth is that this rule of evidence, like most others, is not based on any research, reliable or otherwise.

Nor is this phenomenon confined to rules of evidence. Decisions of constitutional law are often based in substantial part on untested factual assumptions--for example, the assumption by champions of a robust First Amendment that laws restricting speech have a "chilling effect" beyond their immediate impact. This assumption may be true or false, but the courts that formulate doctrinal rules based on it (and many other assumptions) do not ordinarily demand empirical evidence before issuing their pronouncements.

Of course, much of law is based on normative rather than factual determinations. But even normative propositions often turn on factual ones. Whether the benefits of affirmative action justify stigmatizing its beneficiaries is a normative question. Whether affirmative action in fact stigmatizes its beneficiaries, and how powerful or weak, how widespread or limited, the stigma may be, are at least partly empirical questions. Both should be relevant to deciding whether, and under what circumstances, affirmative action is lawful.

For all we know, there may be solid empirical evidence supporting many of the policies written into our law. But until something like Federal Rule of Evidence 702 is routinely applied by courts and legislatures to questions beyond the matter of admitting expert testimony, our law-giving institutions will continue to operate more or less in the dark.