Iowa Fire Marshal Defends 1996 Arson Investigation

Iowa’s fire marshal said Thursday that he has found “overwhelming evidence” that a man serving life in prison in a 1996 fire that killed his wife and child is guilty of arson, rejecting claims from defense experts that the case was built on faulty science.

Fire Marshal Ray Reynolds said he reviewed his office’s original investigation into the fire at the Springville home of Stephen Keyes after receiving a request from a defense lawyer earlier this year. He said his review of the 271-page document found no “gross mistakes” by investigators and that they used appropriate scientific methods in determining the fire was arson.

“I think this was a great investigation. I think the evidence is overwhelming,” he said in a phone interview with The Associated Press. “I can sleep easy on this one, and that’s a pretty firm term in my book. Integrity and justice are the foundations of the pillars in our office. I would not stand for an investigation that I later found out was wrong.”

Keyes is seeking a new trial based on claims by defense experts that his conviction was tainted by outdated fire investigation techniques. Prosecutors say Keyes set fire to his home on the day after Christmas in 1996 and prevented his wife and 2-year-old son from leaving by removing smoke detectors and putting a table in front of an exit. The two were found dead in a second-story bedroom. Keyes and two other children escaped. A jury convicted Keyes in 1997.

Keyes’ attorney, Rockne Cole, had asked for the review after the fire marshal’s office implemented a policy requiring Reynolds to inform a prosecutor if he has reason to believe a conviction is based on invalidated science or methods.

Cole said Thursday he has not received the report of Reynolds’ findings, which were recently shared with the prosecutor but have not been made public.

“We remain confident that the court will agree that in fact … the original arson determination was made upon now-discredited indicators of arson,” he said.

The understanding of arson has changed significantly in recent years, and groups such as the Innocence Project have made it a top priority to push for higher investigative standards to prevent wrongful convictions and review old cases. The effort intensified after the case of Cameron Willingham, a Texas man who was executed in 2004 for a deadly fire that some experts now believe was accidental.

Reynolds said the review of Keyes’ case was not done under his new policy, which would have been the first such review, but simply out of “an abundance of caution.” He said a claim of a wrongful conviction by a defense lawyer by itself isn’t enough to trigger a review.

Gerald Hurst, a prominent defense expert from Texas who worked on the Willingham case and others, has concluded in a report for Keyes’ defense that the investigation was based on techniques “long relegated to the category of old wives’ tales.” Hurst claims an investigator jumped to the conclusion the fire was arson before ruling out potential electrical causes and improperly used a sniffing dog to confirm his suspicion that gasoline was used as an accelerant.

The arson determination was based on since-discredited beliefs that cement cracks – called spalling – and low burn patterns are arson indicators, Hurst’s report says.

Reynolds said the agents did use terms such as spalling and low-burning that raise red flags today, but only to describe what they saw, not as proof of arson. He said he can understand why those words were questioned by defense experts, but he said they were taken out of context to try to get Keyes out of prison.

“In fact, the more we learn about the request, we are disappointed an ‘expert’ can take five snippets of testimony and attempt to discredit two retired agents,” he said. “We’re not going to shy away from investigating allegations. But we’re also not going to stand by and let people throw mud at our office and our work.”