Fingerprinting and analysis of hair fibers and marks made by weapons are familiar forensic tools to those of us who love crime shows, never mind to criminal defendants on trial and those who say they were wrongly convicted by evidence based on those techniques.

So you may be surprised to learn that none of those methods—which comprise the majority of what most real-life labs do—have been scientifically validated, and of the techniques commonly used in the nation's forensic labs, only DNA analysis has been rigorously proved to match a suspect to a crime.

Those are the conclusions of a new report released yesterday by the National Academy of Sciences (NAS). "In terms of the reliability and accuracy in making individualization conclusions, it is fair to say that, with the exception of nuclear DNA analysis, there is a lot we do not know about other forensic disciplines," said the NAS panel's co-chair, Constantine Gatsonis, director of the Center for Statistical Sciences at Brown University, in a statement.

In the report, the NAS recommends that Congress establish a National Institute of Forensic Science to study the non-validated techniques, and to create and enforce standards for forensic scientists and the labs they work in. Those labs should work independently of police departments and prosecutors, the panel that assembled the report added.

After the findings were released, Sen. Jay Rockefeller said he'd introduce legislation that would "address the need for standards, including best practices and certification and accreditation of forensic professionals."

Between 5 percent and 10 percent of a crime lab's analysis involves DNA testing; the rest encompasses other techniques, according to the New York City–based Innocence Project, which works to exonerate convicts through DNA analysis. (The Human Genome Project has a detailed explanation of what happens to DNA in a forensics lab.)

In the case of tool mark analysis, there aren't studies of large populations that would show how many of the weapons such as knives or wood share distinguishing characteristics. Scientists lack information about how much one person's fingerprints vary with each impression — and how much prints differ across a population. And there's no evidence that analyzing hair fibers is a reliable way to identify a specific person, said the report, which Congress commissioned two years ago.

About half of the 232 people the Innocence Project has helped exonerate were originally found guilty by "unvalidated or improper" forensics, the group says. Its co-director, Peter Neufeld, called the NAS findings "unprecedented."

"This report is a major breakthrough toward ensuring that so-called scientific evidence in criminal cases is solid, validated and reliable," Neufeld, who testified at NAS hearings on the issue and shared data with the panel, said in a statement. "For too long, forensic science professionals have not had the support or management needed to identify the real strengths and weaknesses of different assays and techniques. This report provides the roadmap for rectifying that problem."

Image by timsnell via Flickr