How Do Fingerprints Identify People? A Comparison Made by a Human

Friction ridge skin on the fingers, palms and soles forms during fetal development under influences that are partly genetic and partly random, which is why identical twins have different fingerprints. The ridges form patterns classified into broad groups, principally loops, whorls and arches, which are useful for sorting and not for identification. Identification relies on the finer detail, the points where ridges end, split into two or form short independent segments, which are collectively called minutiae. Comparison examines the type, position and relative orientation of these features, and at higher magnification the shapes of individual pores and ridge edges. Fingerprints are also durable, remaining unchanged through life apart from scarring, and they are deposited constantly through sweat and skin oils, which is what makes them practical evidence rather than merely distinctive.

Examiners work through a documented sequence intended to impose discipline on a subjective judgement:

• •Analysis, assessing whether the mark recovered from a scene contains enough clear detail to be worth comparing at all, which is a decision about quality made before looking at any candidate
• •Comparison, examining the mark against a known print feature by feature
• •Evaluation, reaching one of three conclusions: identification, exclusion or inconclusive
• •Verification, an independent second examiner repeating the process, which historically was frequently not blind, meaning the second examiner knew what the first had concluded
• •Database searching using automated systems that rank candidates by similarity, which produces a shortlist for human comparison rather than making the identification itself
• •A point standard was used in some jurisdictions, requiring a set number of matching minutiae, and has largely been abandoned in favour of holistic judgement, which improved flexibility and removed an objective threshold

The confidence collapsed publicly in 2004 when American authorities identified a lawyer in Oregon as the source of a fingerprint recovered from the Madrid train bombings, on the basis of three examiners plus an independent expert appointed for the defence. Spanish authorities disagreed, the mark was eventually matched to an Algerian national, and the man was released after two weeks in custody. The subsequent review found that the initial database search produced a candidate whose print was genuinely similar, and that once examiners believed they had a match, contextual bias led them to interpret ambiguous detail in ways that confirmed it. That mechanism has since been demonstrated experimentally: examiners given the same prints alongside different contextual information reached different conclusions, and some examiners disagreed with their own earlier conclusions when shown the same prints again without knowing it. A 2009 report by the American National Academy of Sciences examined forensic disciplines broadly and concluded that fingerprint comparison lacked a demonstrated error rate and that claims of absolute certainty were not scientifically supportable.

The response has been substantial and incomplete. Error rate studies were finally conducted, with a large black box study finding false positive rates well below one percent but not zero, and false negative rates considerably higher, which gives the discipline the empirical grounding it had lacked. Testimony standards changed, with examiners in several jurisdictions now prohibited from claiming zero error rate, absolute certainty or individualisation to the exclusion of all others. Blind verification, in which the second examiner does not know the first conclusion, has been adopted in some laboratories. Context management procedures restrict what information examiners receive about a case. Statistical approaches expressing the strength of evidence as a likelihood ratio, as DNA evidence does, are being developed. The broader lesson reached beyond fingerprints, since the same review questioned bite mark comparison, hair microscopy, tool mark analysis and arson indicators, several of which have since been substantially discredited and have featured in exonerations.

Fingerprint identification compares minutiae, the points where ridges end or divide, and the patterns are genuinely highly variable, including between identical twins. What was never established until recently was the error rate of the human comparison, and the Madrid bombing misidentification in 2004 showed how contextual bias leads examiners to confirm a match they already believe. Studies now put false positives below one percent rather than at zero, and claims of absolute certainty are no longer permitted in several jurisdictions.