CORSICANA, Texas -- As a key witness in the capital murder trial of Cameron Todd Willingham, then-Deputy State Fire Marshal Manuel Vasquez testified that the fire that killed the defendant's three daughters was "a crime of arson."
"In your opinion, could this fire have been started accidentally?" Vasquez was asked.
"No sir," he replied. "It was not accidental."
Seventeen years later -- and five years after Willingham was executed in connection with the case -- nine fire experts have raised the possibility that the blaze may have indeed been accidental, citing revised standards in arson investigations.
Some circumstances in the Willingham case were similar to those in the case of Ernest Ray Willis, who was convicted of setting a 1986 fire that killed two sleeping women in the small West Texas town of Iraan. Willis, who spent 17 years on death row, was released in 2004 after then-Pecos County District Attorney Ori White dropped the murder case after an inquiry that strongly suggested that the fire was an accident. Authorities were also accused of concealing evidence and giving Willis anti-psychotic drugs that left him dazed during his trial.
"The science of arson investigations was very different in the late 1980s than it was in the late 1990s," said White, who is now Pecos County attorney. "The scientific evidence exonerated him, and that's why I dismissed the case."
The fire marshal's arson finding was the dominant element in the prosecution's case against Willingham, an unemployed mechanic who denied setting the blaze. Retired Judge John H. Jackson, who, as assistant district attorney at the time, prosecuted the case, said the finding unquestionably showed that the fire was deliberately set, despite "paper experts" who are now challenging the findings. Waco attorney David Martin, who represented Willingham and has since said he believes that his client was guilty, says an expert for the defense also concluded that the fire was arson.
But Craig Beyler, a fire scientist hired by the Texas Forensic Science Commission to review the case, has concluded that an arson finding "could not be sustained," saying Vasquez, who died in 1994, used concepts that "are often inconsistent with modern fire science."
The arson investigation into the Willingham case was conducted before the 1992 publication of what has since become the standard for arson science: NFPA 921, a Guide to Fire and Explosion Investigation, published by the National Fire Protection Association. Vasquez and Douglas Fogg, then assistant fire chief for Corsicana, cited more than 20 indicators to conclude that the fire was intentionally set. But several of those indicators have since come into question in light of the newer standards in NFPA 921.
Among them:
"Crazed glass," a weblike pattern earlier assumed to be caused by use of a liquid accelerant. Now it is also attributed to the rapid chilling of hot glass by water from fire hoses.
"Pour patterns," "trailers" and "puddles," markings suggesting that someone dumped accelerant. A "post-flashover fire" -- which spread from single objects to engulf an entire room, reaching temperatures of more than 2,000 degrees -- can also produce floor burn patterns that cannot be distinguished from those caused by liquid accelerants.
Multiple points of origin, which can indicate that a fire was deliberately ignited. But other causes, such as embers or drop-down burning, can produce the same effect, experts say.
Gerald Hurst, an Austin arson expert, also challenged the fire marshal's investigation in a report that Willingham's appeals attorney submitted to Gov. Rick Perry in 2004 in an unsuccessful bid to delay Willingham's execution.
"On first reading, a contemporary fire origin and cause analyst might well wonder how anyone could make so many critical errors in interpreting the evidence," Hurst said. "However, when the report is looked at in the context of its time and in light of a few key advances that have been made in the fire investigation field in the last dozen years, it becomes obvious that the report more or less simply reflects the shortcomings in the state of the art prior to the beginning of serious efforts to introduce standards and to test old theories that had previously been accepted on faith."
Corsicana, in a response to Beyler's report, noted that NFPA 921 wasn't published until after the fire and Willingham's trial, and may not have been "universally acknowledged" until the late 1990s.
"Therefore, it is not remarkable that the investigators did not employ a methodology that was not yet published or accepted," said the report by Corsicana Fire Chief Donald McMullan, who was not in office when the fire occurred. "Having said that, it may very well be that the fire investigators did use many or some of the principles stated in NFPA 921, since some of those specific principles were known in 1991."
Fogg acknowledged during cross-examination that accelerants aren't the only cause of a puddle pattern.
Vasquez investigated the case over eight days, interviewing witnesses and examining the charred 50-year-old frame house. In his Jan. 24, 1992, report, he determined that the fire was set in three places and eliminated natural gas or electricity as the cause. He also reported finding a burned can of charcoal lighter fluid near the east end of the porch. A sample of wood debris at the base of the front door tested positive for the combustible liquid accelerant kerosene.
Hurst said the positive testing for "mineral spirits of kerosene" could have been the charcoal lighter fluid washed across the porch toward the doorway from the fire hoses. He also challenged Vasquez's suggestion that brown spots on the porch could have been accelerant, saying they may have been rust.
Walter Reaves of Waco, Willingham's appeals attorney, said the prosecution would have been hard-pressed to bring a case against his client if the current standards had been in effect. "I don't even think he would have gone to trial," Reaves said. "How can you take somebody to trial if you can't even say it's arson?"
White, the former Pecos County district attorney, hired Hurst and another expert, Kendall Ryland, fire chief for Effie, La., to conduct an independent inquiry in the Ernest Willis case after questions arose about the evidence that convicted him.
In the original arson investigation, White said, pour patterns made it appear that "someone had walked through the house with a gas can pouring accelerant out."
But with the development of more advanced scientific techniques in the 1990s, he said, investigators determined that the extreme heat from post-flashover burning "can create marks that look like accelerant pour patterns."
"That was one of the mistakes that the initial arson investigators made," he said. "In the late '80s, they did not have a full grasp of the science of arson investigations."
Ryland, a former fire instructor at Louisiana State University, was also one of three experts who examined the Willingham case for the Chicago Tribune, which published an investigative report in 2004 questioning the arson findings that led to Willingham's execution. In looking into the case, Ryland said he could not re-create the conditions described by the original investigators.
"I really did get sick to my stomach after I reviewed that case," Ryland, a fireman for more than 25 years, told the Fort Worth Star-Telegram this week. "He could have done it, but he probably didn't."
___
(Fort Worth Star-Telegram researcher Cathy Belcher contributed to this report.)
___
(c) 2009, Fort Worth Star-Telegram.
Visit the Star-Telegram on the World Wide Web at .
Distributed by McClatchy-Tribune Information Services.
