Keith Harward spent 33 years in a Virginia prison for a 1982 murder and rape he did not commit. The evidence that convicted him, and nearly got him sentenced to death, was the testimony of six forensic dentists that his teeth matched bite marks on the surviving victim. There was never any science behind that claim. When the Innocence Project finally tested the preserved DNA in 2016, it pointed to a shipmate who had died in prison. Harward's case is one of dozens in which "junk science" has sent an innocent person away, and it is the centerpiece of a conversation about why courts keep letting it in.
In this TalksOnLaw interview, recorded in 2022, Joel Cohen speaks with M. Chris Fabricant, Director of Strategic Litigation at the Innocence Project and author of Junk Science and the American Criminal Justice System, about the origins of bite-mark evidence, the human cost of unvalidated forensics, and the double standard he calls "poor people science."
What Junk Science Is
Fabricant defines junk science as "subjective speculation masquerading as scientific evidence" and explains why the pattern-matching disciplines, bite marks, footwear, hair, firearms and toolmarks, are so vulnerable: they rest on an examiner's eye, not on validated methods. He describes the FBI's 2004 misidentification of Brandon Mayfield's fingerprint in the Madrid bombing and the experiments showing that even experienced examiners change their conclusions when told different facts about a case. Most crime labs, he notes, sit inside police departments and prosecutors' budgets, and there is no "FDA for forensics"; the only gatekeeper is a judge consulting precedent rather than the scientific literature.
How Bite Marks Took Over
Fabricant sued the Justice Department for the archives of the American Board of Forensic Odontology and traced bite-mark evidence to a 1970s Los Angeles case in which a judge admitted a dentist's testimony about a bite on a nose because, he said, he could see the match himself. The precedent "spread like a virus." The 1979 Ted Bundy trial, with little other physical evidence and a hypnotized eyewitness, turned bite marks into the "Bundy signature" and the dentists who testified into celebrities. One of them later testified against Robert DuBoise, who sat near Bundy on Florida's death row and was exonerated decades later.
Poor People Science
The Supreme Court's 1993 Daubert decision came from a civil case about a prescription drug, pushed by corporate defendants tired of junk science in product-liability suits. Fabricant explains the research showing that judges tightened their gatekeeping in civil cases afterward while criminal admissibility barely changed, even after the National Academy of Sciences (2009) and the President's Council of Advisors on Science and Technology (2016) found that most pattern-matching disciplines lacked scientific validation. The defendants in criminal cases are disproportionately poor, Black and brown, and, he argues, that is why nobody cared. The Innocence Project's original intake rule, DNA-testable cases only, is what revealed how often "infallible" forensics, eyewitnesses and confessions were wrong: faulty forensic evidence appears in nearly half of DNA exonerations.
The Harward Case and the Problem of Finality
Fabricant recounts the "dental dragnet" of sailors aboard the USS Carl Vinson, how Harward went from excluded to suspect after biting his girlfriend in a fight, and how a paralegal's chance reading of an old appellate opinion brought the case to the Innocence Project. He explains why Harward's exoneration was possible only because DNA happened to survive, and why clients convicted on bite marks or toolmarks without DNA, like Eddie Lee Howard in Mississippi and Steven Chaney in Texas, face newly-discovered-evidence rules that courts read narrowly for fear of "opening Pandora's box." As of the recording, no forensic technique in his book had been ruled inadmissible in any state.
What to Know Now
The gatekeeping rules have tightened on paper. Federal Rule of Evidence 702 was amended effective December 1, 2023 to make explicit that the proponent must show by a preponderance of the evidence that expert testimony meets every requirement of the rule, and that an expert's opinion must reflect a reliable application of the method to the facts; the advisory committee's note targets overstated forensic "match" testimony in particular. In 2022–23 the National Institute of Standards and Technology published a scientific foundation review of bite-mark analysis finding no support for its basic premises: that human dentition is unique, that it transfers accurately to skin, and that examiners can reliably interpret it. In 2026 Oregon enacted a law designating bite-mark analysis, microscopic hair comparison and comparative bullet-lead analysis as discredited disciplines for purposes of post-conviction relief, and the first conviction was vacated under it that June.
The cases Fabricant mentions have moved. James Genrich, convicted in Colorado on toolmark testimony, won a new trial in 2023, and the charges were dismissed in April 2026, the first toolmark conviction overturned on reliability grounds. Charles McCrory, convicted in Alabama in 1985 on bite-mark testimony that the dentist himself later recanted, was granted federal habeas relief in September 2026 by a judge who called bite-mark identification "rank speculation"; the state must retry or release him. In Texas, the Court of Criminal Appeals stayed Robert Roberson's execution in October 2025 and sent his "junk science writ" claim, based on shaken-baby evidence, back to the trial court, where it remains pending. The National Registry of Exonerations counted 3,767 exonerations through 2025; of the 97 recorded in 2025, 40 percent involved false or misleading forensic evidence. Bite-mark evidence remains admissible in every state.
