Disheartening science.

Reason Magazine - Breaking Up the Forensics Monopoly:

"The most recent comprehensive study of crime lab proficiency, published by the Journal of Forensic Sciences in 1995, analyzed the tests administered by the Forensic Sciences Foundation and Collaborative Testing Services as a part of the accreditation process.

...The best-performing group of disciplines, which included “finger and palm prints, metals, firearms, and footwear,” had error rates at or above 2 percent. The first item on that list is especially important: False fingerprint identification usually leads to a false conviction, because of the prestige of fingerprint evidence and its undeserved reputation for infallibility. With 238,135 requests for latent fingerprint comparisons in 2002 alone, a false positive error rate of 2 percent implies up to 4,800 false convictions or guilty pleas made in hopes of a lighter sentence each year in the U.S., 1,700 of them in felony cases. (The number of improperly matched fingerprints is not completely clear. A 2005 study of fingerprint analysis suggests that the false positive rate may now be as low as 0.8 percent. But another recent study suggests it could exceed 4 percent.)"