In the early hours of September 14, 1982, an intruder broke into a house on Warwick Boulevard in Newport News, Virginia, beat a man unconscious with a crowbar, and raped his wife while their three children slept down the hall. He bit her legs more than once, leaving marks sharp enough to photograph. Detectives and naval officials screened the dental records of hundreds of sailors off the aircraft carrier docked nearby; two dentists examined Keith Harward's teeth for a closer look and cleared him. Six months later, with no suspect and two U.S. senators writing to the Navy, detectives learned Harward had been accused of biting his girlfriend in an unrelated argument. They shipped his dental molds to Dr. Lowell Levine, a New York forensic odontologist.
Levine concluded that Harward had inflicted the bites. A second dentist the prosecution brought in, Dr. Alvin Kagey, agreed. Told of Levine's finding, the two dentists who had originally cleared Harward reversed themselves to match it. At trial in 1986, Levine told the jury that, with reasonable medical certainty, Harward had caused the bite marks on the victim's leg, walking them through a canted tooth, a hooked edge, a chipped corner that lined up with the photograph. Harward's own lawyer called no dentist to contest any of it. The jury convicted him of first-degree murder. He was sentenced to life.
Nothing about that testimony was consciously dishonest. Forensic odontologists had been comparing bite marks to dental casts in American courtrooms since the 1970s, under guidelines a board the profession had chartered for itself. None of the four dentists' conclusions, on either side of the reversal, rested on a premise anyone had actually tested: that human teeth leave a pattern on skin unique enough to trace back to one mouth, through tissue that stretches, bruises, and heals before anyone finishes photographing it. A National Research Council committee that reviewed the whole of forensic science reached the opposite finding in 2009, more than two decades after Harward's trial: "The uniqueness of the human dentition has not been scientifically established." No study of a large population had ever tested that claim. The same committee named the failure that had just reversed two dentists' own exclusions: when police supply the suspect for comparison, it found, an examiner faces outsized pressure to match the bite to him, and the guidelines gave no way to measure or correct for it.
The actual killer was identified in January 2016, when the state's own forensic laboratory, authorized to run new DNA tests on evidence collected the night of the crime, excluded Harward as a contributor and matched the profile instead to another sailor off the same ship, Jerry Crotty, who had died in an Ohio prison in 2006 on an unrelated charge. Virginia's Supreme Court granted Keith Harward a writ of actual innocence on April 7, 2016. He walked out of prison the next day, after thirty-three years, and the state later paid him just over $1.5 million for them.
A jury cannot measure certainty; it can only hear how certain a witness sounds. Two dentists had already said Harward's teeth did not match the bite, until they were told what the answer was supposed to be, and then they agreed.