In 2016, a jury convicted Audrey Cannon of burglary after police said she broke into a house in Northeast Portland.
Forensic investigators with the Portland Police Bureau linked Cannon to the crime scene based on a fingerprint recovered from a liquor bottle they analyzed.
“She was charged and convicted of that crime based on the fingerprint," said Marc Brown, Oregon's chief deputy public defender for appeals.
According to court documents, the latent fingerprint is the only evidence that connects Cannon to the crime scene.
“It’s a problem because we like to have evidence that’s reliable," he said.
On Friday in Salem, the Oregon Court of Appeals is set to hear Cannon's appeal. The case could change the way fingerprint evidence is presented in Oregon's courts, or whether it should be allowed at all.
For decades, fingerprints have been used during trial as irrefutable evidence linking a person to a crime. But recently, science has found the way fingerprints are collected and analyzed is subject to error.
Cannon's case is unique because the conviction relies heavily on the fingerprint evidence. It forces Oregon's appeals court to consider the quality of fingerprint evidence and how it’s presented in court.
Fingerprints are used all the time to identify people, whether for a background check or a gun permit. Those fingerprints are done in a controlled setting and are widely accepted.
The issue Cannon's case raises involves latent fingerprints — those left unintentionally.
“Think about the dynamics of touch," said Liz Flannery, a supervisor at the Oregon State Police crime lab.
“How does the finger actually contact the surface? Is it during course of movement?" she posed. "If so, there’s the potential that you’re creating distortion in the fingerprint.”
Most latent fingerprints, Flannery said, aren’t completely clear.
Brown said the Multnomah County judge who presided over Cannon’s trial should not have allowed the jury to hear the fingerprint evidence.
“There’s no appellate case that says this is valid science," Brown said. "Most states have an appellate decision from way back when that says these are valid."
By all accounts, scientific evidence is powerful at trial. It’s only become more so because of TV crime shows that depict forensic investigators clinching convictions.
During Cannon’s trial in 2016, the Portland Police Bureau criminalist testified that she linked the fingerprint found in the burglarized home to Cannon.
The Portland Police Bureau declined to comment because of the pending litigation.
"'In Portland, over 70 years of identifying prints, we have never [had] an erroneous or a mistake in an identification that has been out of our office,'" the Bureau's criminalist testified in 2016 according to appeals court documents. "' ... The Portland Police Bureau has never misidentified a person.'"
Brown said it’s an incredible statement; one that he said is not scientifically plausible, unprovable and highly unlikely.
“When a juror’s sitting there and here’s this fingerprint that we found at the scene belongs to that person — because it’s drilled into us that fingerprints are the gold standard of identification — once people hear that it’s game over," Brown said.
In 2004, the FBI arrested a Portland man in connection with a terrorist attack in Madrid that killed nearly 200 people riding commuter trains.
The FBI used a fingerprint to link Brandon Mayfield to the bombings. Agents searched Mayfield’s house and law office in the Portland suburbs and he was held in jail for two weeks. Weeks later, the FBI and Spanish National Police determined the fingerprint belonged to an Algerian national.
Shortly after being cleared and released, Mayfield spoke publicly about his experience.
“I said right away, 'I’m not going to plea anything on this. I’m not guilty of this. I didn’t do it,'" Mayfield told the Portland City Club in 2004. "But I also said, No. 1, this is not my finger print; No. 2, if it is I don’t know how it got there ... that was always my feeling.”
University of Oregon Law Professor Carrie Leonetti teaches about criminal procedure and evidence. She told OPB’s "Think Out Loud" that when it comes to latent fingerprints, forensic experts are trained to testify that they have a zero percent error rate.
"Which would make any real scientist cringe," Leonetti said.
"They’re starting to do studies and, lo and behold, they do have an error rate. It’s not appalling," she said. "But obviously, if you said to a jury it’s a match, but we have a 10 percent error rate, I don’t think you can convict somebody only on a 10 percent error rate.”
Like Brown, Leonetti agrees that fingerprint evidence shouldn’t be allowed into court until more research is done.
“If it’s validated, then two or three years from now, you’re testifying, but you’re testifying honestly to the limitations of your field and its reliability problems," she said.
Since Mayfield’s case, two key studies have looked into latent fingerprint identification.
One study was done by the FBI and the other by the Miami-Dade Police Department. Both found cases where latent fingerprints were falsely identified.
OSP's Flannery concedes their work isn’t always perfect.
“I think any good scientist would tell you that there’s always a possibility that there’s an error," she said.
The Oregon Department of Justice is arguing the case for the state. The agency declined an interview, citing its policy about not discussing pending litigation.
In court documents, Peenesh Shah, an assistant attorney general with the Oregon DOJ, wrote that they’re not aware of any court rejecting fingerprint evidence because it lacks scientific validity.
Shah wrote, "The relevant studies do not conclude that the error rate is unreasonably high." But he also noted in his brief that it’s higher than the general public would believe.
"The solution is for defendants to educate jurors about the real error rate, either through cross-examination of the state's expert or through testimony of a defense expert," Shah wrote.
Oral arguments in Cannon's appeal are set to begin at 9 a.m. Friday.