On March 11, 2004, bombs exploded on commuter trains in Madrid, killing about 200 people. Spanish police found fingerprints on a plastic bag of detonators and sent them through INTERPOL. One of them, Latent Fingerprint 17, reached the FBI Laboratory in Quantico.
An FBI examiner ran it through the Bureau’s automated database of millions of prints. The computer returned 20 candidates. The one it scored fourth-highest belonged to Brandon Mayfield, a lawyer in Portland, Oregon. The examiner compared the two prints side by side and concluded, on March 19, that they came from the same finger.
Then the checks began. A second examiner, a retired FBI examiner working on contract, verified the match. A unit chief reviewed it and agreed. Under the procedures of the time, the verifier knew, when he was asked to check, that an identification had already been made.
On April 13, the Spanish National Police laboratory reported its own comparison: “negativo.” Two days later, before finding out why the Spaniards disagreed, the laboratory told colleagues it was “absolutely confident” in the match, according to the Justice Department’s Inspector General. On May 6, agents arrested Mayfield as a material witness.
A court appointed an independent expert, picked by Mayfield’s lawyers from a list the FBI Laboratory had recommended. On May 19 he testified that the print was Mayfield’s. Describing his error later, in talks the Inspector General quoted, he said there came a point when his mind “shifted toward agreement with the Bureau’s identification.”
That same day, the FBI learned that Spanish police had matched Latent Fingerprint 17 to someone else: Ouhnane Daoud, an Algerian. Mayfield was released to home detention on May 20. On May 24, the FBI withdrew its identification.
The Inspector General published his review in March 2006. The two prints, he found, were unusually similar, with as many as 10 points in matching positions. But the examiners had also reasoned “backward”: once they had Mayfield’s print in front of them, they began to “find” details in the Madrid print that were not really there, reading murky spots as points of agreement. Nothing required the examiner to record what he saw in the Madrid print before he looked at Mayfield’s, so it was impossible to know for certain which details had been clear to him first.
The Inspector General found no compelling evidence that the verification procedure itself had swayed the verifier, who declined to be interviewed, and noted that the court’s expert had reached the same answer. Still, the report’s fixes were about order and independence: write down what the crime-scene print shows before comparing it, and let verifiers work without knowing whether anyone has already declared a match.
When the laboratory finally re-examined the print, overnight on May 23, a unit chief brought in another unit chief. One reason he gave the Inspector General: that colleague had not been involved in the identification.
In November 2006, the government settled part of Mayfield’s lawsuit for $2 million and, his lawyer said, apologized again.