When the Science Was Never There: Junk Forensics, Unethical Investigations, and the Institutions That Kept Using Both

The Problem Is Structural, Not Incidental

The standard account of forensic fraud treats it as aberration. A bad analyst. A corrupt lab. An outlier. That account is wrong, and the evidence that it is wrong has been available for decades.

In 2009, the National Academy of Sciences released a report on the state of forensic science in American courts. The findings were direct: with the exception of nuclear DNA analysis, no forensic discipline had demonstrated, through rigorous empirical research, that its methods could reliably produce accurate results or that practitioners could consistently apply those methods the same way. The report identified bite mark analysis, microscopic hair comparison, blood spatter interpretation, tool mark analysis, and shoe print comparison as disciplines in which the scientific foundation was, at best, contested and, at worst, invented.

The courts had been admitting this testimony for years. They continued admitting it after the report was published.

What Bite Marks Actually Demonstrate

Bite mark analysis is premised on two claims: that human dentition is unique, and that skin reliably preserves the impression of that dentition in a way that allows comparison. Neither claim has been validated by controlled research. The claim of dental uniqueness is an assumption, not a finding. The claim of reliable skin impression is contradicted by research on how human skin deforms, decomposes, and changes shape after death.

The Innocence Project has documented multiple exonerations in which bite mark testimony was a central element of the conviction. Ray Krone spent ten years in an Arizona prison, two of them on death row, before DNA evidence excluded him and identified another person. The bite mark testimony at his trial had been presented by an expert who described his methodology as reliable and his conclusions as certain. Neither characterization held up.

The American Board of Forensic Odontology, the professional body that certifies bite mark analysts, has periodically updated its guidelines. It has not produced the controlled error rate research that would be required to justify the certainty with which analysts testify. A 2016 report from the President’s Council of Advisors on Science and Technology concluded that bite mark analysis should not be admitted in criminal proceedings pending development of a scientific foundation. Courts have largely continued to admit it.

Microscopic Hair Analysis and the FBI’s Correction That Did Not Correct Much

For decades, FBI analysts testified that microscopic hair comparison could positively associate a questioned hair sample with a known individual. The technique was presented as a reliable scientific method. It is not. Hair lacks the individualizing characteristics that would allow a positive identification. Two hairs can be indistinguishable under a microscope and come from different people. The technique can exclude, but it cannot include with scientific validity.

In 2015, the FBI, in cooperation with the Innocence Project and the National Association of Criminal Defense Lawyers, acknowledged that hair analysts had overstated the significance of hair comparison evidence in at least 95 percent of cases reviewed in an initial audit. The affected cases spanned decades. Some defendants had already been executed. Some had died in prison. The audit covered a fraction of the cases in which this testimony had been offered.

The problem with hair analysis is not that individual analysts were dishonest, though some were. The problem is that the method was taught as reliable, certified as reliable, and presented in court as reliable for generations before anyone was required to produce the research that would justify that characterization. The scientific community did not validate the technique. Courts validated it by admitting it repeatedly, and repetition became its own form of authority.

Fred Zain, Annie Dookhan, and the Lab Fraud Architecture

Fred Zain worked as a serology analyst for the West Virginia State Police crime lab from 1977 to 1989, and subsequently for the medical examiner’s office in Bexar County, Texas. During that time, he fabricated or altered results in a number of cases that has never been fully determined. The precise scope of his fraud was never established because the documentation necessary to determine it was inadequate. West Virginia did not have a system that would have caught him.

The West Virginia Supreme Court of Appeals, reviewing Zain’s conduct in 1993, described his work as representing a systematic practice of fraud. The court found that supervisors had received complaints and had not acted on them, that external auditors had identified anomalies that were not investigated, and that the lab had operated for years without meaningful quality controls. The fraud was discovered because a defendant whose conviction was built on Zain’s testimony sought post-conviction DNA testing, not because the system identified a problem.

Annie Dookhan worked as a chemist at the William A. Hinton State Laboratory Institute in Massachusetts from 2003 to 2012. She confirmed the presence of controlled substances in samples without testing them, forged colleague signatures, and manipulated test results. Investigations following her arrest estimated that approximately 24,000 cases were potentially affected by her work. The lab had given her performance awards for processing a volume of samples that should have been physically impossible to test accurately. No one investigated why her numbers were so much higher than her colleagues’.

These are not stories about exceptional wrongdoers. They are stories about systems that created conditions in which wrongdoing was easy, difficult to detect, and, once detected, difficult to remediate. Zain operated in a lab where no one verified his work. Dookhan operated in a lab where throughput was treated as a metric of quality. In both cases, the lab existed within a prosecutorial ecosystem in which the expectation was that the analyst’s job was to confirm what investigators already believed.

Shaken Baby Syndrome and the Medical Diagnosis as Legal Proof

Shaken baby syndrome, now more commonly referred to by practitioners as abusive head trauma, is a medical diagnosis based on a triad of findings: subdural hematoma, retinal hemorrhage, and brain swelling. For decades, the presence of that triad was treated in courts as sufficient evidence to establish that an infant had been violently shaken, and the person responsible for the child at the time of the medical emergency was treated as the responsible party.

The scientific basis for that inference has been contested by a significant portion of the biomechanical and pediatric medical community. Research published in peer-reviewed journals has raised questions about whether shaking alone, without impact, can produce the force necessary to generate the triad, whether the triad can result from accidental falls or short-distance drops, whether medical conditions can mimic the triad, and whether the retinal hemorrhages previously treated as a specific indicator of abuse occur in other circumstances. The medical community has not reached consensus on these questions. Courts have frequently proceeded as if it had.

Kathleen Folbigg in Australia and a number of defendants in the United States and United Kingdom have had convictions examined or overturned in proceedings that turned on the contested science underlying this diagnosis. In each instance, the original conviction was built on expert testimony that presented the diagnosis as more settled than the underlying research supported.

Why Courts Keep Admitting What Scientists Have Rejected

The persistence of discredited forensic methods in courts is sometimes explained as a lag problem. Science advances, courts catch up slowly. That explanation is too generous. The 2009 NAS report was not the first time the scientific community raised concerns about these disciplines. Challenges to bite mark analysis appeared in the research literature in the 1970s. Challenges to microscopic hair analysis predate the FBI’s 2015 acknowledgment by decades. The lag is not a matter of information. It is a matter of institutional incentive.

Courts rely on forensic evidence. Prosecutors build cases around it. Judges who were once prosecutors learned to trust it. Defense attorneys, particularly public defenders operating under severe resource constraints, frequently cannot afford the expert witnesses necessary to challenge it. The infrastructure of the criminal legal system is built around the assumption that forensic evidence is reliable, and that assumption is not examined until a specific defendant, in a specific case, raises a specific challenge with sufficient resources to make it.

The Daubert standard, properly applied, should have filtered out methods without validated error rates and peer-reviewed scientific support. It has not, in part because the standard gives trial judges substantial discretion, and trial judges are not scientists. Judges evaluate whether a method has been generally accepted in the relevant scientific community, but the relevant scientific community for bite mark analysis is forensic odontologists, many of whom practice bite mark analysis. The community that accepts the method and the community that could validate it are not the same community.

What Reform Would Actually Require

The 2009 NAS report made specific recommendations. A significant number remain unimplemented. The report called for the creation of a National Institute of Forensic Science, independent of the Department of Justice, that would set standards, fund research, and oversee accreditation. Congress has not created that institution. The report called for mandatory accreditation of all forensic labs. Many labs remain unaccredited. The report called for development and publication of error rates for every forensic discipline used in criminal proceedings. That research has not been systematically conducted.

The gap between what the report recommended in 2009 and what exists in 2026 is itself a finding. It documents who benefits from the current system and who absorbs the cost of its failures.

The history of forensic accountability in the United States is a history of external pressure producing internal acknowledgment that does not produce systemic change. The FBI acknowledged the hair analysis problem after advocates forced disclosure. Massachusetts addressed the Dookhan fallout after defendants organized and litigated in volume. West Virginia reviewed Zain’s work because a specific defendant pursued a specific remedy. In each case, the system did not self-correct. It was corrected from outside, at substantial cost, by people with inadequate resources and considerable determination.

That is not a sustainable model for ensuring that the forensic evidence used to deprive people of their liberty is accurate. It is, however, the model that currently exists.