Texas Tries Again

For all the dismal stories to come out of Texas's justice system in recent years—defense attorneys falling asleep during death-penalty trials, thousands of rape kits left untested for years while perpetrators went free, and the nation's highest number (48) of wrongly convicted people who've been released on DNA evidence—the most disturbing may have come a few years ago when credible evidence emerged that the state had executed an innocent man. Cameron Todd Willingham, a Corsicana resident convicted of starting a house fire in 1991 in which his three young daughters died, was given a lethal injection in February 2004, even after arson experts had debunked nearly all of the physical evidence in the case. When David Grann published a 9,000-word exposé in The New Yorker in 2009, it created a national outcry. But Gov. Rick Perry stood by the execution, calling Willingham a "monster" and scuttling a state investigation.

By refusing to investigate the Willingham case, Texas was turning its back on revolutionary advances in the field of fire science over recent decades. Methods traditionally used to investigate arson and win convictions turned out to be, in the words of leading arson expert Gerald Hurst, little more than "old wives' tales." For instance, burn patterns long thought to show that gasoline had been used to intentionally ignite a fire are now known to be false indicators. But as The Texas Observer reported in 2009, nearly 800 Texans remained in prison on arson convictions based in part on such evidence. One of them was Ed Graf, a Waco accountant serving a life sentence for a fire that killed his two stepsons in 1986. His conviction, according to Hurst and other fire scientists, rested on even shakier evidence than Willingham's. Even so, it seemed unlikely that Graf or the many others convicted on flawed arson evidence would ever have a second chance to clear their names.

But on Oct. 6, Graf will be the first convicted arsonist in Texas to be retried—and could soon become the first to be freed from prison following the Willingham scandal. The retrial is the result of a sea change that began in 2011, when the Texas Legislature removed the Perry-appointed head of the state's Forensic Science Commission—the man who stalled the Willingham investigation—and the commission issued a report that termed the evidence leading to the execution outdated and unreliable. The following year, a new state fire marshal, Chris Connealy, asked six arson experts, along with the Lubbock-based Innocence Project, to reexamine his office's old arson cases. The state that had once stood by an almost certainly wrongful execution was becoming the nation's unlikely leader in pursuing justice for wrongly convicted arsonists. The forensic scientists found evidence of wrongful convictions in five cases, including Graf's, and found two others inconclusive for lack of evidence.

Graf's story is in many ways a textbook example of how arson investigations can go awry. It began on a hot Tuesday afternoon in August 1986, when he left work early and picked up his 8- and 9-year-old stepsons, Joby and Jason, from day care. About 10 minutes after Graf and the kids got home, neighbors saw smoke pouring from the wooden shed behind the house. Flames ripped through the windowless structure in minutes. The boys were found dead inside.

The investigation was flawed from the beginning. Firefighters bulldozed the shed on the day of the tragedy; investigators had to examine the remnants in the town dump. But they nonetheless claimed to know how the fire started, theorizing that Graf had drugged the boys, dragged them into the shed, poured gasoline on the floor, ignited it, and locked the door. While there was also circumstantial evidence that made many in Waco believe that Graf, never a popular man, was guilty—including a $50,000 insurance policy he'd taken out on the boys two months earlier—it was the forensic evidence that sealed his fate.

In 2006, Graf's Innocence Project attorney contacted Hurst, who has worked to exonerate dozens of people across the country wrongly convicted of arson. He saw right away, he says, that the evidence against Graf was bad. The burn patterns that investigators thought indicated gasoline had been poured on the floor of the shed in fact showed no such thing. Graf could not have locked the boys in, Hurst found, because the door of the windowless shed had to have been open—otherwise, the fire would have died out from lack of oxygen. Moreover, autopsies showed no evidence of drugs in the boys' bodies. Hurst couldn't say for sure how the fire started, but he theorizes that the kids started it themselves and it got out of control.

Doug Carpenter, another nationally recognized fire expert, studied the evidence closely and agreed with Hurst's assessment. He also found one more piece of vital evidence: Toxicology reports showed that the boys had died from carbon-monoxide poisoning. That indicates they passed out and continued to breathe in the fire's toxic fumes, evidence that likely rules out the presence of gasoline in the shed. That's because, as Carpenter testified in a 2013 hearing challenging Graf's conviction, victims in gasoline fires typically die of heat exposure from the fast-moving flames; they stop breathing and have low levels of carbon monoxide in their blood.

In March 2013, the Texas Court of Criminal Appeals overturned Graf's conviction. He expected to be released. But then Waco prosecutors announced they would retry him for murder. This time, they will have to rely purely on circumstantial evidence—the insurance policy Graf tried to cash in, the fact that he'd insisted that the boys keep tags on their recently purchased school clothes and then returned the clothes after the fire, and family members' claims that Graf, who'd studied arson cases as an insurance-claims adjuster, had mentioned to them that the origins of fires are difficult to pinpoint.

"With Graf, you've got a case where the man isn't very popular," Hurst says. "People don't like Graf." That, he says, led many to suspect him right away. But Graf's attorney has pointed out that the insurance policy Graf bought actually covered his entire family, and that, as a tightfisted sort, it wasn't surprising that he kept tags on his kids' clothes in case they didn't fit or the boys didn't wear them. And just because someone works in insurance and knows something about arson doesn't make him capable of killing his two stepchildren. "OK, you want to retry him," Hurst says, "but what's your basis?" If there was no arson, there was no crime.

Can prosecutors convict a man of murder without any physical evidence that murder was committed? If they do, prosecutors in other Texas counties—and in other parts of the country as well—may be emboldened to retry more people whose convictions rested on faulty arson investigations. If they don't, it will be a significant victory for innocence advocates—and for those whose convictions rest on now-discredited science.

Dave Mann is editor of The Texas Observer.

This article appears in the October 4, 2014 edition of National Journal Magazine as Texas Tries Again.