Kirk Odom spent more than two decades in prison for a crime that was later proven he didn't do.

"It was something that was worse than a nightmare," he recalled from his D.C.-area home.

While in prison, he was sexually assaulted. He contracted HIV. He broke down, and at one point, attempted to take his own life.

"It was just a complete mess," he said. "Mentally, it just tore me down. It put more fear in me than I ever experienced or ever heard of."

A big part of Odom's fate, determined in the spring and fall of 1981, and those years of trauma in prison had to do with hair. And more specifically, the techniques that had long been used to compare a suspect's hair to those found at the scene of a crime.

The only problem was, those techniques and their interpretation were often misconstrued and oversold in the courts.

In the largest post conviction review of evidence, the FBI is now revisiting federal cases involving comparative hair analysis, many of which date back decades. The agency is also encouraging state and local crime labs to do the same.

That's because in its review of hundreds of cases thus far, they're discovering a lot of errors in the way this forensic tool had been applied.

But it doesn't stop there.

The review has been followed by calls by some in the science community for more evidence and standards when it comes to other comparative tools that have long been used, and are still used, in criminal investigations.

Comparative hair analysis: an approach used for decades

Analyzing hair as part of criminal investigations dates back at least to the early 1900s.

"By the twentieth century it was not common, but it was certainly being attempted," said Skip Palenik, a senior microscopist at Microtrace LLC.

Palenik says one can find this mirrored in detective fiction of the time, like Sherlock Holmes.

Chris Fabricant, director of strategic litigation at the Innocence Project, has been tracing the origins of hair comparison in courts, and found an early example that involved eyeballing a hair.

"One of the earliest recorded cases that involved the use of hair microscopy was a murder of a plantation owner in Mississippi in the late 1800s," said Fabricant. "And you can kind of see the probative value of a hair that appeared to be the same color as the victim's, and it was bloody and it was found in a noose...You could just take a look at it and decide for yourself how relevant that was to the prosecution."

By the mid twentieth century, the FBI's crime lab had begun training experts in comparative hair techniques. This specifically involved examining strands of hair, side by side under the microscope, and then presenting those findings in court.

Keep in mind, this was also before DNA testing was introduced in the late 1990s.

The practice was used in cases like Kirk Odom's.

Odom recalls how during the spring of 1981, a local D.C. police officer approached him because he thought he looked similar to the sketch of a suspect from a recent crime.

They brought Odom in and took hair samples.

"Still, I didn't know what was going on, what I was being charged with," said Odom.

Turns out he was being charged with breaking into a woman's apartment and raping her at gunpoint. Odom was 18 years old at the time.

During the trial itself, an FBI-trained analyst testified that the hair found at the scene was microscopically similar to Odom's, and that this similarity was really, really rare.

As Odom listened, he couldn't understand it.

"I was saying to myself, I knew it wasn't me, it wasn't my hair because I know I was not there," he said.

Still, he was convicted.

Errors identified

In 2013, FBI began its post conviction review in collaboration with the National Association of Criminal Defense Lawyers and the Innocence Project. By 2015, it had mined through hundreds of cases and identified errors in 90 percent of the testimony given by experts during trial.

Fabricant, with the Innocence Project, says one problem stems from the hair analysis being presented in misleading ways.

"These hair comparison presentations are not just verbal testimonies, but are often accompanied by elaborate visual demonstrations, that if you look at the hair at issue and the suspect's hair, they appear to match," said Fabricant.

In nine of these cases, the FBI reported that defendents had already been executed. 

Fabricant says hair errors don't necessarily alter the outcome of a case. Oftentimes other evidence supports a conviction, but the reviews have already reversed the outcomes of some cases.

Last month, two cases were reopened in Arkansas as a result of the review identifying flawed testimony from experts.

Beyond the federal post conviction review, Fabricant's group has found that about a quarter of the some 300 wrongful convictions that have ever been overturned by DNA evidence have involved faulty hair evidence.

A case of 'overconfidence'

For Skip Palenik, who analyzes hair and other fibers under the microscope for a living, he says a lot can be gleaned from hair, such as whether its from an animal, and its types of pigments and colors.

"You can certainly tell a lot from hairs," he said.

He often uses the structure of a pencil as an analogy for describing the microscopic anatomy of a hair, with its different cellular coatings and layers.

But part of the problem, he says, of the FBI hair testimony, is that as much as he loves the science of hair and what it can reveal under the microscope, "trying to say if two hairs came from an individual, you can't do that so well. And the reason you can't is there are only a limited number of characteristics that hairs have in first place."

In other words, hair is just not as unique as one might hope.

Palenik says these days, comparative analysis can help narrow down which hairs are the best candidates for mitochondrial DNA testing, but he thinks that "at some point along the way, some people at the FBI and in other public sector labs developed what I think of as overconfidence in what can you tell from hairs."

A growing chorus of scientists and people in the criminal justice world agree.

In reviewing cases, Fabricant and others found some common errors: an examiner would testify that the hair belonged to the suspect at the exclusion of everyone else. Or an expert would conflate the statistical weight of the similarities and the probability that the hairs originated from the same source. He worried about unintended, subjective biases filtering into the analysis too.

Comparative techniques put under the microscope

In 2009, a National Academy of Sciences Report on Forensic Science added to the increasing evidence against the technique. It reported that microscopic hair comparison methods are imprecise and problematic and that other comparison disciplines like fingerprints, bite marks, and tool marks also lacked standards and research to back them up.

"It's not that what we're doing is definitely wrong, it's that we don't know," said Karen Kafadar, chair of the department of statistics at the University of Virginia. She was involved in that report.

Kafadar says the FBI and the government are now doing more research and are trying to develop better standards, but this September yet another report came out. This one was from the President's Council of Advisors on Science and Technology, or PCAST.

Like the 2009 report, it called into question not just hair, but other comparison disciplines and their accuracy in drawing connections between suspects and crimes. Kafadar says, as it stands, a lot of the evidence supporting these techniques is weak.

"We're not expecting everything to be as precise as DNA," she said. "We're just trying to figure out what we have."

The group recommended regular assessments, better standards and more federal funding to study these techniques. Still, even without further study, it seems there's already agreement that relying on comparative hair analysis to link a suspect to a crime just doesn't work.

"I can tell you that we haven't used it [comparative hair analysis] in my county in decades," said William Fitzpatrick, the longtime district attorney in Onondaga county, New York, and chair of the National District Attorney's association.

But he does turn to other comparative techniques, like fingerprints, in assessing evidence of a crime. Fitzpatrick is concerned that these latest reports are creating uncertainty in these other areas. And while he accepts that more research is a good thing, he doesn't want these approaches to be thrown out in the meantime.

"I think it'd be nothing short of a disaster. You'd have hundreds of thousands of guilty defendants exonerated because the prosecution would be unable to present scientifically valid evidence that points to an individual's guilt," he said.

He also believes it's the role of judges to determine the reliability of forensic evidence.

For his part, Chris Fabricant does not have the same level of faith in the court system.

"The status quo as it exists today in forensic science is that scientific validity and reliability is determined by scientifically illiterate judges, that are presented by scientifically illiterate prosecutors, that are defended by scientifically illiterate defense attorneys, and are evaluated by 12 scientifically illiterate lay jurors," said Fabricant. "And the result is a lot of bad science getting into court. And that's happening right now, today, as you and I are speaking."

A vindication decades later

Back to Kirk Odom, in 2011 his case caught the attention of the public defenders service in Washington D.C, after another longtime conviction, hinging on hair comparison evidence, had been overturned thanks to advances in DNA testing.

Odom had already spent more than two decades in prison and nearly 10 years on parol. He was registered as a sex offender and constantly under watch, he said.

"That put a lot of stress and strain on me, you know still trying to find a job," he said.

But in reopening his case, DNA testing was able to rule Odom out as the perpetrator of that 1981 crime.

His conviction was overturned, and he was exonerated. He got the call in July of 2012. It was the day he turned 50.

"Oh my goodness, that was one birthday present I'll never forget," said Odom.

The FBI credits Odom's case, along with two others, in helping to trigger that big federal review of cases involving hair analysis.

On the one hand, Odom says faulty science has forever changed his life, but it's also the power of science, and things like advances in DNA testing even on hair, that still give him hope.

"As you can see, the justice system is not always right. It takes time to do everything," said Odom. "And for what I've been through it took a long time, but what science is now is not what it used to be, it's even better with the DNA testing and all."

The science, he says, is coming around.

"And that's a good thing," he said. "That's a real good thing."