In 1992, 24-year-old Lorie Lee Lance was killed in a house fire in Old Hickory, Tennessee. Claude Francis Garrett, Lance’s boyfriend, was accused of setting the fire. Although he swore he was innocent, an arson investigator claimed that a set of large, irregular burns in the living room represented a “pour pattern” indicating the use of a liquid accelerant. A large container of kerosene was found in the home, which Garrett claims was used in a kerosene heater.

The arson investigator’s testimony helped get Garrett convicted not once, but twice. Garrett was originally convicted in 1993 and was retried in 2003. At the second trial, evidence was introduced that the so-called “pour pattern” evidence was junk science. In fact, most of the arson techniques used before the mid-1990s have been debunked. The arson investigator stood by his original findings, however, and the jury believed him.

A veteran fire investigator who testified for the defense in the 2003 trial is now one of Garrett’ most passionate advocates. He calls Garrett’s case “a classic example of injustice.” In 1992, the National Fire Protection Association promulgated new, science-based rules for arson investigators that debunked the traditional methods used in the Garrett investigation. If the new rules had been followed, the defense expert says, no investigator would have concluded it was arson.

For one thing, the so-called pour patterns would have been tested for actual kerosene residue in the flooring samples. Later tests revealed there was none. For another, we now know that fires caused by ignitable liquids simply don’t cause the kind of charring found on the floor at the Garrett residence. The original arson investigator’s testimony was demonstrably wrong.

Over the past few years, several other people have been exonerated in arson cases that relied upon now-debunked conclusions about pour patterns and the behavior of fire. Others have had their convictions overturned or their sentences reduced. Many spent decades in prison before their cases were reexamined.

Yet despite it becoming clear that a lot of old arson techniques are invalid, meaning that people are serving time for wrongful arson convictions, most states aren’t systematically reexamining old cases. The wrongfully convicted must rely on groups like the Innocence Project stepping in or investigative reporters writing about their cases.

This week, Garrett will go before the Tennessee Board of Parole. It has been 25 years since his initial conviction, and he has now served his minimum term. By all accounts, he has been a model prisoner, and he has numerous character witnesses to present. Two scientific analyses of the fire scene have been submitted to bolster his claim of innocence. Will he be released?