NIJ Conference 2010: Summary of Plenary "Rising from the Ashes: What We Have Learned From the Cameron Todd Willingham Case"

July 2010


  • Itiel Dror, Cognitive Neuroscientist, University College London
  • David Grann, Staff Writer, The New Yorker, New York
  • John Lentini, President and Principal Investigator, Scientific Fire Analysis LLC, Big Pine Key, Fla.
  • Michael Logan Ware, Chief, Special Fields Bureau, Dallas County District Attorney's Office
  • Moderator: Mary Lou Leary, Principal Deputy Assistant Attorney General, Office of Justice Programs, U.S. Department of Justice, Washington, D.C.

On Feb. 17, 2004, Texas executed Cameron Todd Willingham for setting the fire that killed his three daughters in December 1991. The trial evidence that led to his 1992 conviction in their murder was based largely on findings from local fire investigators' examination of Willingham's house. However, in the weeks before Willingham's execution, independent fire investigators began re-examining the claims, refuting each one with scientific evidence. This new information did not persuade the state's review board or the governor's office. Subsequent reviews have confirmed the view that the original conclusions drawn from the fire investigation were based on outdated thinking and myths about the behavior of fires and the signs they leave.

The Willingham case is “a microcosm of many of the most challenging issues facing our criminal justice system today — the use of forensic science, crime scene investigation, eyewitness testimony [and] indigent defense, among others,” Principal Deputy Assistant Attorney General Mary Lou Leary, in the Department of Justice Office of Justice Programs, said when introducing the June 14 panel of four experts.

The work of the panelists — a journalist, an arson expert, a neuroscientist, and an attorney — aligns with a renewed focus on science as a priority of the current administration, Leary emphasized. In turn, they discussed the science of fire investigation, cognitive and psychological blind spots in conducting a criminal investigation, and one county's efforts to review capital convictions. In the area of DNA analysis, Leary noted, forensic science has provided the evidence for 254 exonerations nationwide. With a priority on science, DOJ is working “to build a criminal justice system that has no room for guesswork,” she said.

Investigative reporter David Grann focused national attention on the Willingham case with his September 2009 overview in The New Yorker. On the panel, he summarized the proceedings, from witnesses' initial reports of the incident to the clemency appeal sent to the governor's office, and Willingham's final words. The evidence presented at trial rested heavily on the findings of two men who investigated the burned-out one-story house in Corsicana, a city of 20,000 people located 55 miles northeast of Waco. The city's assistant fire chief — a certified arson investigator — and the deputy fire marshal had decades of experience.

Among the 20 points of evidence from the fire investigation that were submitted at trial, two were central to the prosecutors' arguments:

  • Crazed glass. Windows in Willingham's house had fractured in a distinctive crackle pattern the investigators attributed to a particularly hot-burning blaze, one they argued must have been fueled by lighter fluid or a liquid fuel.
  • Low burning and pour patterns. Heat rises. Yet, the investigators said, they found char marks along the base of the walls, on the floor, and under the beds in the girls' room, which pointed to a purposely set blaze. They also traced char patterns in the remains that looked like puddles — evidence that someone had poured a flammable liquid along the floor and a sure sign of arson.
  • Multiple sources. Burn marks that looked like the letter V appeared on more than one wall, pointing — literally — to several points where fires ignited. An accidental fire, investigators reasoned, would have a single source.

The investigators were confident of their analysis; one even testified that “a fire doesn't lie. But in the words of John Lentini, president and principal investigator of the Florida-based company Scientific Fire Analysis, “you can misinterpret what it tells you.”

Lentini learned about the Willingham case when he was contacted by a reporter from the Chicago Tribune, who asked him to review the trial evidence. His report debunked the myths behind each of the claims.

Fire investigators and those who study the science of fires often inhabit separate worlds, Lentini said on the panel. Fire investigators in many local jurisdictions joined the force after high school, or took individual training courses based on the accumulated — but uncorroborated — wisdom of senior colleagues. But even reputed sources spread misinformation. The idea that crazed glass is an indicator of rapid heating, for example, was codified in the widely used 1980 Fire Investigation Handbook, published by the National Bureau of Standards. Many investigators and their chiefs remain unaware that training and certification in scientific investigation techniques is available online at sites such as, a resource of the International Association of Arson Investigators.

The original investigators' knowledge of fire, Lentini found, drew on a number of myths. The crackle pattern they saw in the building's crazed glass resulted from rapid cooling — not a hotter blaze. As flames ignite and grow, heat rises — until it hits the ceiling. Smoke and gas collect there, and heat begins to radiate down. Within minutes, a phenomenon known as flashover can occur: As gases build up, they ignite, setting an entire room instantly on fire. Not only can a flashover set objects in the room alight — creating the V patterns investigators saw — it can also create char marks that are visually indistinguishable from pour patterns. Where gas collects under furniture, fire can burn there. Flames can scorch one wall, travel across the ceiling and touch down on the other side. In such cases, testing for the presence of a flammable liquid is often the only way to verify arson. Fire is dynamic, but investigations into its aftermath often don't account for its now well-documented behavior.

Collecting knowledge about the science of fire and disseminating that knowledge is a clear priority. But there was another problem, not related to forensic evidence, according to Itiel Dror, Ph.D., a cognitive neuroscientist at University College London.

“There is a much bigger problem that happens all the time: the psychology of violating evidentiary independence,” he said.

Investigators typically track many separate lines of evidence — eyewitness accounts, fingerprints, DNA, and more. Proving guilt rests on the mutual independence of the evidence.

The brain's tendency to find patterns can create bias — and contaminate evidence. Examiners are careful to avoid fouling biological samples, but contamination can be psychological, too. For example, to avoid the influence of suggestive body language, some argue that lineups should be double blind — ensuring that the identity of the suspect is kept from the eyewitness as well as the officer conducting the lineup. Research suggests that confessions can be tainted, too. Dror noted, for example, that one-quarter of the people who have been exonerated on the basis of DNA evidence, “confessed” to the crime.

Dror describes a British murder case in which detectives hit a dead-end when computer manipulation failed to decipher the patterns of letters and numbers in a fuzzy CCTB image of the killer's license plate. When police later arrested a suspect, however, the same examiners claimed they could pick out the pattern of the suspect's car plates from the fuzzy image.

Dror's published work also includes a study in which he and his research colleagues gave examiners fingerprints to study with certain contextual details about the case: that the prints were from a suspect who had confessed or been identified by an eyewitness at the scene of the crime. In fact, the fingerprints were from cases the same investigators had worked on in previous years, although the details were new. In several cases, the additional context changed the investigator's conclusions about the fingerprints.

These and other findings serve as a caution regarding routine practices in crime investigation. Dror argued that the lead investigator must use particular care in deciding what to reveal to medical examiners, fingerprint technicians and other investigators in order to maintain the independence and integrity of the evidence. To the extent such errors compromise that independence, police officers and detectives — often unconsciously — weaken their own case.

The example of Cameron Todd Willingham also highlights the importance of avenues for review when bias, error or other factors lead to a wrongful conviction. Interest in formally establishing a review process is building in a number of jurisdictions, but the practice was first established following the election of a new Dallas County, Tex., district attorney in 2007.

Prosecutor Michael Logan Ware, who leads the county's Conviction Integrity Unit, presented an overview of its work. Initially, the unit joined with the Innocence Project of Texas and the Dallas County Public Defender's Office to review more than 400 instances in which an inmate had requested a post-conviction DNA test but had been turned down.

By these and other avenues, 20 convictions have been overturned, with exonerations affirmed by the state Court of Criminal Appeals. Two additional defendants have been exonerated in capital murder cases using evidence other than DNA. “Prosecutors have a duty to seek justice,” Ware said, “not just protect convictions.”

“It is our position at the District Attorney's office that it is our obligation to protect the citizens of Dallas County,” Ware said. “If the wrong people are being convicted of violent crimes that are actually being committed by other people, we are not fulfilling our obligation to protect the citizens of Dallas County.”

Looking at common factors in the 20 cases, Ware said, suggests that improving the quality of defense provided to poor defendants is a priority — and an important lesson to be learned from the Willingham case. “Every one [of the 20 defendants] in these wrongful convictions was indigent,” Ware said. Improving the quality of indigent defense could have a positive effect.

The year after Willingham was executed, the state established the Texas Forensic Science Commission Exit Notice to investigate allegations of error and misconduct by forensic scientists, including those in the Willingham case. Lentini and others have submitted their findings to the commission.

The case of Cameron Todd Willingham suggests many ways in which science could be used to improve the system of criminal justice in the United States. Within their area of expertise, these panelists suggested ongoing work to make such improvements, specifically the ways scientific findings and processes can be used to improve investigations, prevent error, avoid bias and re-examine old cases.

Date Created: July 9, 2010