“My problems are simple,” Willingham wrote Gilbert in September, 1999. “Try to keep them from killing me at all costs. End of story.”

  During his first years on death row, Willingham had pleaded with his lawyer, David Martin, to rescue him. “You can’t imagine what it’s like to be here, with people I have no business even being around,” he wrote.

  For a while, Willingham shared a cell with Ricky Lee Green, a serial killer, who castrated and fatally stabbed his victims, including a sixteen-year-old boy. (Green was executed in 1997.) Another of Willingham’s cellmates, who had an I.Q. below seventy and the emotional development of an eight-year-old, was raped by an inmate. “You remember me telling you I had a new celly?” Willingham wrote in a letter to his parents. “The little retarded boy.… There was this guy here on the wing who is a shit sorry coward (who is the same one I got into it with a little over a month ago). Well, he raped [my cellmate] in the 3 row shower week before last.” Willingham said that he couldn’t believe that someone would “rape a boy who cannot even defend himself. Pretty damn low.”

  Because Willingham was known as a “baby killer,” he was a target of attacks. “Prison is a rough place, and with a case like mine they never give you the benefit of a doubt,” he wrote his parents. After he tried to fight one prisoner who threatened him, Willingham told a friend that if he hadn’t stood up for himself several inmates would have “beaten me up or raped or”—his thought trailed off.

  Over the years, Willingham’s letters home became increasingly despairing. “This is a hard place, and it makes a person hard inside,” he wrote. “I told myself that was one thing I did not want and that was for this place to make me bitter, but it is hard.” He went on, “They have [executed] at least one person every month I have been here. It is senseless and brutal.… You see, we are not living in here, we are only existing.” In 1996, he wrote, “I just been trying to figure out why after having a wife and 3 beautiful children that I loved my life has to end like this. And sometimes it just seems like it is not worth it all.… In the 31⁄2 years I been here I have never felt that my life was as worthless and desolate as it is now.” Since the fire, he wrote, he had the sense that his life was slowly being erased. He obsessively looked at photographs of his children and Stacy, which he stored in his cell. “So long ago, so far away,” he wrote in a poem. “Was everything truly there?”

  Inmates on death row are housed in a prison within a prison, where there are no attempts at rehabilitation, and no educational or training programs. In 1999, after seven prisoners tried to escape from Huntsville, Willingham and four hundred and fifty-nine other inmates on death row were moved to a more secure facility, in Livingston, Texas. Willingham was held in isolation in a sixty-square-foot cell, twenty-three hours a day. He tried to distract himself by drawing—“amateur stuff,” as he put it—and writing poems. In a poem about his children, he wrote, “There is nothing more beautiful than you on this earth.” When Gilbert once suggested some possible revisions to his poems, he explained that he wrote them simply as expressions, however crude, of his feelings. “So to me to cut them up and try to improve on them just for creative-writing purposes would be to destroy what I was doing to start with,” he said.

  Despite his efforts to occupy his thoughts, he wrote in his diary that his mind “deteriorates each passing day.” He stopped working out and gained weight. He questioned his faith: “No God who cared about his creation would abandon the innocent.” He seemed not to care if another inmate attacked him. “A person who is already dead inside does not fear” death, he wrote.

  One by one, the people he knew in prison were escorted into the execution chamber. There was Clifton Russell, Jr., who, at the age of eighteen, stabbed and beat a man to death, and who said, in his last statement, “I thank my Father, God in Heaven, for the grace he has granted me—I am ready.” There was Jeffery Dean Motley, who kidnapped and fatally shot a woman, and who declared, in his final words, “I love you, Mom. Goodbye.” And there was John Fearance, who murdered his neighbor, and who turned to God in his last moments and said, “I hope He will forgive me for what I done.”

  Willingham had grown close to some of his prison mates, even though he knew that they were guilty of brutal crimes. In March, 2000, Willingham’s friend Ponchai Wilkerson—a twenty-eight-year-old who had shot and killed a clerk during a jewelry heist—was executed. Afterward, Willingham wrote in his diary that he felt “an emptiness that has not been touched since my children were taken from me.” A year later, another friend who was about to be executed—“one of the few real people I have met here not caught up in the bravado of prison”—asked Willingham to make him a final drawing. “Man, I never thought drawing a simple Rose could be so emotionally hard,” Willingham wrote. “The hard part is knowing that this will be the last thing I can do for him.”

  Another inmate, Ernest Ray Willis, had a case that was freakishly similar to Willingham’s. In 1987, Willis had been convicted of setting a fire, in West Texas, that killed two women. Willis told investigators that he had been sleeping on a friend’s living-room couch and woke up to a house full of smoke. He said that he tried to rouse one of the women, who was sleeping in another room, but the flames and smoke drove him back, and he ran out the front door before the house exploded with flames. Witnesses maintained that Willis had acted suspiciously; he moved his car out of the yard, and didn’t show “any emotion,” as one volunteer firefighter put it. Authorities also wondered how Willis could have escaped the house without burning his bare feet. Fire investigators found pour patterns, puddle configurations, and other signs of arson. The authorities could discern no motive for the crime, but concluded that Willis, who had no previous record of violence, was a sociopath—a “demon,” as the prosecutor put it. Willis was charged with capital murder and sentenced to death.

  Willis had eventually obtained what Willingham called, enviously, a “bad-ass lawyer.” James Blank, a noted patent attorney in New York, was assigned Willis’s case as part of his firm’s pro-bono work. Convinced that Willis was innocent, Blank devoted more than a dozen years to the case, and his firm spent millions, on fire consultants, private investigators, forensic experts, and the like. Willingham, meanwhile, relied on David Martin, his court-appointed lawyer, and one of Martin’s colleagues to handle his appeals. Willingham often told his parents, “You don’t know what it’s like to have lawyers who won’t even believe you’re innocent.” Like many inmates on death row, Willingham eventually filed a claim of inadequate legal representation. (When I recently asked Martin about his representation of Willingham, he said, “There were no grounds for reversal, and the verdict was absolutely the right one.” He said of the case, “Shit, it’s incredible that anyone’s even thinking about it.”)

  Willingham tried to study the law himself, reading books such as “Tact in Court, or How Lawyers Win: Containing Sketches of Cases Won by Skill, Wit, Art, Tact, Courage and Eloquence.” Still, he confessed to a friend, “The law is so complicated it is hard for me to understand.” In 1996, he obtained a new court-appointed lawyer, Walter Reaves, who told me that he was appalled by the quality of Willingham’s defense at trial and on appeal. Reaves prepared for him a state writ of habeas corpus, known as a Great Writ. In the byzantine appeals process of death-penalty cases, which frequently takes more than ten years, the writ is the most critical stage: a prisoner can introduce new evidence detailing such things as perjured testimony, unreliable medical experts, and bogus scientific findings. Yet most indigent inmates, like Willingham, who constitute the bulk of those on death row, lack the resources to track down new witnesses or dig up fresh evidence. They must depend on court-appointed lawyers, many of whom are “unqualified, irresponsible, or overburdened,” as a study by the Texas Defender Service, a nonprofit organization, put it. In 2000, a Dallas Morning News investigation revealed that roughly a quarter of the inmates condemned to death in Texas were represented by court-appointed attorneys who had, at some point in their careers, been “reprimanded, placed on probation, suspended or banned from practicing law by the State Bar.” Although Reaves was more competent, he had few resources to reinvestigate the case, and his writ introduced no new exculpatory evidence: nothing further about Webb, or the reliability of the eyewitness testimony, or the credibility of the medical experts. It focussed primarily on procedural questions, such as whether the trial court erred in its instructions to the jury.

  The Texas Court of Criminal Appeals was known for upholding convictions even when overwhelming exculpatory evidence came to light. In 1997, DNA testing proved that sperm collected from a rape victim did not match Roy Criner, who had been sentenced to ninety-nine years for the crime. Two lower courts recommended that the verdict be overturned, but the Court of Criminal Appeals upheld it, arguing that Criner might have worn a condom or might not have ejaculated. Sharon Keller, who is now the presiding judge on the court, stated in a majority opinion, “The new evidence does not establish innocence.” In 2000, George W. Bush pardoned Criner. (Keller was recently charged with judicial misconduct, for refusing to keep open past five o’clock a clerk’s office in order to allow a last-minute petition from a man who was executed later that night.)

  On October 31, 1997, the Court of Criminal Appeals denied Willingham’s writ. After Willingham filed another writ of habeas corpus, this time in federal court, he was granted a temporary stay. In a poem, Willingham wrote, “One more chance, one more strike / Another bullet dodged, another date escaped.”

  Willingham was entering his final stage of appeals. As his anxieties mounted, he increasingly relied upon Gilbert to investigate his case and for emotional support. “She may never know what a change she brought into my life,” he wrote in his diary. “For the first time in many years she gave me a purpose, something to look forward to.”

  As their friendship deepened, he asked her to promise him that she would never disappear without explanation. “I already have that in my life,” he told her.

  Together, they pored over clues and testimony. Gilbert says that she would send Reaves leads to follow up, but although he was sympathetic, nothing seemed to come of them. In 2002, a federal court of appeals denied Willingham’s writ without even a hearing. “Now I start the last leg of my journey,” Willingham wrote to Gilbert. “Got to get things in order.”

  He appealed to the U.S. Supreme Court, but in December, 2003, he was notified that it had declined to hear his case. He soon received a court order announcing that “the Director of the Department of Criminal Justice at Huntsville, Texas, acting by and through the executioner designated by said Director … is hereby DIRECTED and COMMANDED, at some hour after 6:00 P.M. on the 17th day of February, 2004, at the Department of Criminal Justice in Huntsville, Texas, to carry out this sentence of death by intravenous injection of a substance or substances in a lethal quantity sufficient to cause the death of said Cameron Todd Willingham.”

  Willingham wrote a letter to his parents. “Are you sitting down?” he asked, before breaking the news. “I love you both so much,” he said.

  His only remaining recourse was to appeal to the governor of Texas, Rick Perry, a Republican, for clemency. The process, considered the last gatekeeper to the executioner, has been called by the U.S. Supreme Court “the ‘fail safe’ in our criminal justice system.”

  One day in January, 2004, Dr. Gerald Hurst, an acclaimed scientist and fire investigator, received a file describing all the evidence of arson gathered in Willingham’s case. Gilbert had come across Hurst’s name and, along with one of Willingham’s relatives, had contacted him, seeking his help. After their pleas, Hurst had agreed to look at the case pro bono, and Reaves, Willingham’s lawyer, had sent him the relevant documents, in the hope that there were grounds for clemency.

  Hurst opened the file in the basement of his house in Austin, which served as a laboratory and an office, and was cluttered with microscopes and diagrams of half-finished experiments. Hurst was nearly six and a half feet tall, though his stooped shoulders made him seem considerably shorter, and he had a gaunt face that was partly shrouded by long gray hair. He was wearing his customary outfit: black shoes, black socks, a black T-shirt, and loose-fitting black pants supported by black suspenders. In his mouth was a wad of chewing tobacco.

  A child prodigy who was raised by a sharecropper during the Great Depression, Hurst used to prowl junkyards, collecting magnets and copper wires in order to build radios and other contraptions. In the early sixties, he received a Ph.D. in chemistry from Cambridge University, where he started to experiment with fluorine and other explosive chemicals, and once detonated his lab. Later, he worked as the chief scientist on secret weapons programs for several American companies, designing rockets and deadly fire bombs—or what he calls “god-awful things.” He helped patent what has been described, with only slight exaggeration, as “the world’s most powerful nonnuclear explosive”: an Astrolite bomb. He experimented with toxins so lethal that a fraction of a drop would rot human flesh, and in his laboratory he often had to wear a pressurized moon suit; despite such precautions, exposure to chemicals likely caused his liver to fail, and in 1994 he required a transplant. Working on what he calls “the dark side of arson,” he retrofitted napalm bombs with Astrolite, and developed ways for covert operatives in Vietnam to create bombs from local materials, such as chicken manure and sugar. He also perfected a method for making an exploding T-shirt by nitrating its fibres.

  His conscience eventually began pricking him. “One day, you wonder, What the hell am I doing?” he recalls. He left the defense industry, and went on to invent the Mylar balloon, an improved version of Liquid Paper, and Kinepak, a kind of explosive that reduces the risk of accidental detonation. Because of his extraordinary knowledge of fire and explosives, companies in civil litigation frequently sought his help in determining the cause of a blaze. By the nineties, Hurst had begun devoting significant time to criminal-arson cases, and, as he was exposed to the methods of local and state fire investigators, he was shocked by what he saw.

  Many arson investigators, it turned out, had only a high-school education. In most states, in order to be certified, investigators had to take a forty-hour course on fire investigation, and pass a written exam. Often, the bulk of an investigator’s training came on the job, learning from “old-timers” in the field, who passed down a body of wisdom about the telltale signs of arson, even though a study in 1977 warned that there was nothing in “the scientific literature to substantiate their validity.”

  In 1992, the National Fire Protection Association, which promotes fire prevention and safety, published its first scientifically based guidelines to arson investigation. Still, many arson investigators believed that what they did was more an art than a science—a blend of experience and intuition. In 1997, the International Association of Arson Investigators filed a legal brief arguing that arson sleuths should not be bound by a 1993 Supreme Court decision requiring experts who testified at trials to adhere to the scientific method. What arson sleuths did, the brief claimed, was “less scientific.” By 2000, after the courts had rejected such claims, arson investigators increasingly recognized the scientific method, but there remained great variance in the field, with many practitioners still relying on the unverified techniques that had been used for generations. “People investigated fire largely with a flat-earth approach,” Hurst told me. “It looks like arson—therefore, it’s arson.” He went on, “My view is you have to have a scientific basis. Otherwise, it’s no different than witch-hunting.”

  In 1998, Hurst investigated the case of a woman from North Carolina named Terri Hinson, who was charged with setting a fire that killed her seventeen-month-old son, and faced the death penalty. Hurst ran a series of experiments re-creating the conditions of the fire, which suggested that it had not been arson, as the investigators had claimed; rather, it had started accidentally, from a faulty electrical wire in the attic. Because of this research, Hinson was freed. John Lentini, a fire expert and the author of a leading scientific textbook on arson, describes Hurst as “brilliant.” A Texas prosecutor once told the Chicago Tribune, of Hurst, “If he says it was an arson fire, then it was. If he says it wasn’t, then it wasn’t.”

  Hurst’s patents yielded considerable royalties, and he could afford to work pro bono on an arson case for months, even years. But he received the files on Willingham’s case only a few weeks before Willingham was scheduled to be executed. As Hurst looked through the case records, a statement by Manuel Vasquez, the state deputy fire marshal, jumped out at him. Vasquez had testified that, of the roughly twelve hundred to fifteen hundred fires he had investigated, “most all of them” were arson. This was an oddly high estimate; the Texas State Fire Marshals Office typically found arson in only fifty per cent of its cases.

  Hurst was also struck by Vasquez’s claim that the Willingham blaze had “burned fast and hot” because of a liquid accelerant. The notion that a flammable or combustible liquid caused flames to reach higher temperatures had been repeated in court by arson sleuths for decades. Yet the theory was nonsense: experiments have proved that wood and gasoline-fuelled fires burn at essentially the same temperature.

  Vasquez and Fogg had cited as proof of arson the fact that the front door’s aluminum threshold had melted. “The only thing that can cause that to react is an accelerant,” Vasquez said. Hurst was incredulous. A natural-wood fire can reach temperatures as high as two thousand degrees Fahrenheit—far hotter than the melting point for aluminum alloys, which ranges from a thousand to twelve hundred degrees. And, like many other investigators, Vasquez and Fogg mistakenly assumed that wood charring beneath the aluminum threshold was evidence that, as Vasquez put it, “a liquid accelerant flowed underneath and burned.” Hurst had conducted myriad experiments showing that such charring was caused simply by the aluminum conducting so much heat. In fact, when liquid accelerant is poured under a threshold a fire will extinguish, because of a lack of oxygen. (Other scientists had reached the same conclusion.) “Liquid accelerants can no more burn under an aluminum threshold than can grease burn in a skillet even with a loose-fitting lid,” Hurst declared in his report on the Willingham case.