(V.Richard Haro/Associated Press)

At laboratories throughout the United States — some large enough to contain a three-story house — researchers have been lighting rooms and houses on fire and analyzing the results with the kind of scientific scrutiny that has upended several deeply entrenched misconceptions about how fires behave. The upheaval is more than academic. For generations, arson inspectors have used outmoded theories to help indict and incarcerate many suspects. But as new science is brought to bear on old cases, it is becoming clear that over the past several decades, dozens, perhaps hundreds, of people have been convicted of arson based on scant research and misguided beliefs. Many of those people are still in jail, hoping that someone will take up their cause.

One of the most famous cases was back in 1991, when Cameron Todd Willingham was charged with setting his house on fire and killing his family. As David Grann’s famous New Yorker story details, the prosecution brought in fire experts who pointed to a whole bunch of telltale signs of arson: pour patterns of flammable liquid, charred wood on the flood. Willingham was eventually found guilty and sentenced to death.

Only later, in 2004, did a new fire investigator, Gerald Hurst, point out that most of this evidence was flawed “in light of current knowledge” about arson. (Those telltale signs were perfectly consistent with an accidental electrical fire.) But Willingham’s appeal was denied, and he was executed anyway. As it turns out, this isn’t an unusual occurrence — as Starr details, courts have been slow to catch up to the rapidly evolving science of arson.