The Library Book

by Susan Orlean

  The city now contended that it had uncovered a motive. Harry had gone to the library with no ill intentions, investigators believed, but he’d become upset when the security guard turned him away, and he’d started the fire in a fit of pique. The theory had some logic. But Harry’s interaction with the guard seemed more trifling than provocative, and Harry didn’t seem the type to react so strongly to a minor scold. But Harry had mentioned in his interviews that the guard who’d stopped him at the door was African American. Was it an idle comment, or did it imply something more? According to a 2015 survey, fewer than four percent of the people in Harry’s hometown of Santa Fe Springs were black. When he was growing up, that number was probably smaller. Nothing else I ever heard suggested that Harry was racist, but I had noted that he repeatedly mentioned the guard’s race.

  If he were angry, it would have been easy to slip into one of the library’s nooks and niches and then strike a match. Maybe Harry did it as a small defiant gesture, nothing more. Maybe he touched the match to a book without giving much thought to what might follow. Early in the investigation, Harry told agent Thomas Makar that he thought the person who started the fire didn’t mean for it to get so big. Perhaps Harry wasn’t the kind of person to set a building like the library on fire—but might he be the kind of person who struck a match when he was mad? I could imagine this: that Harry might have felt peevish because of the guard, and then each time a librarian stopped him, he got more affronted. He might have fingered a matchbook in his pocket with nothing specific in mind. But perhaps he found himself in a secluded spot, all alone amid the creaking piles of books and papers; Harry Peak, actor, always on the brink of being noticed but really always out of view, his image of himself getting a little more threadbare every day, his buoyant optimism unraveling, nothing happening quite as he’d imagined it and not at all like the version he bragged about to the people around him and even to himself. Maybe he let himself tear a match out of a matchbook and then flicked its rough orange head over the striker, and suddenly, he was holding the spurt of flame and feeling exhilarated by his daring, remembering in that moment the kid he had been, who always pushed things further and made more people admire him, and without picturing what would happen in the next minute or the next seven hours, he acted in that instant, almost in a disembodied state. Once he saw the flame take a bite out of a book and he realized it was out of his control, I can picture him rushing away, scrambling, the way you scramble when you’ve broken your mother’s favorite vase—not because you feel bad about it but because you know you’ll have hell to pay.

  Judge Chaney preferred the theory that Harry started the fire intentionally because he wanted attention. In that regard, he was insatiable. But his usual tactic to get attention was to boast about something glittery, like saying he had gone drinking with Cher. He wanted to present a version of his life populated by famous people, superstars. Starting a fire in a public building didn’t have the effervescence that his brags usually did. The fire wasn’t glamorous. It was a gritty, ugly crime, decried by everyone in the city. Saying he started the fire may have placed him at the center of a news event, but it also would have made him appear, to many people, contemptible. Would he really have wanted that kind of attention? As Demitri Hioteles told me, “Harry got off on making people happy.” The fire didn’t fit that; it was too dismal, too real.

  But Harry did tell people he’d started the fire. He repeated the confession to investigators. If it was just a fib, why had he stumbled through his excuses, crisscrossing his alibis again and again? Why had he failed his lie detector test? Where was he, really, on the morning of April 29, 1986? If he wasn’t at the library, how did he know details about that morning? And if he didn’t do it, who did?

  Some years ago, I read a story in The New Yorker that got under my skin. “Trial by Fire,” by David Grann, was about a case in Texas in which a man named Todd Willingham was convicted of setting a fire in 1991 that killed his three children. The key evidence against Willingham was patterns left by the moving fire—what arson investigators call “burn marks”—in the family’s house. A long-standing belief among arson people is that fires burn hottest where they originate. The charring on the house’s wooden floors was darkest and deepest under the children’s beds. There was nothing under the children’s beds that could have started a fire spontaneously, so investigators believed that someone had set the fire intentionally. The only person in the house that night besides the children was Willingham, who claimed he had been asleep at the time the fire began and that he had done everything possible to save his children. Eventually, Willingham was convicted, largely because the burn marks were interpreted as proof that the fire started under the beds. He was sentenced to death. After losing all of his appeals, he was executed in 2004.

  Struck by Willingham’s insistence on his innocence, a prominent scientist and fire investigator named Dr. Gerald Hurst had been asked by Willingham’s family to reexamine the case shortly before the date of execution. Hurst began by trying to determine whether the fire truly was arson. Hurst believed that the analysis of where the fire started was wrong. Despite the heavy burn marks under the children’s beds, he didn’t believe the fire had started there. He inspected the house again. When forensic science was applied to all of the evidence, it showed that the accelerant on the front porch was probably from a can of lighter fluid used to start a small charcoal grill that had been knocked over by firefighters entering the house. A faulty space heater or wiring had probably started the fire inside the house. Flames had raced down the hall into the children’s bedroom. The extreme burn marks under their beds indicated only that the fire had settled there for a while. Hurst’s analysis was too late to change the outcome for Willingham, but it succeeded in raising great concern about the reliability of what we assume about fire.

  As long ago as 1977, forensic scientists warned that the principles of arson investigation were mostly myth. If windows in a burned building were greasy, investigators assumed that accelerant must have been used and left residue on the glass. But modern buildings are filled with petroleum-based products that could leave residue on windows if they burn. Extremely hot fires were presumed to have been supercharged by accelerants, indicating arson, but scientists now know that the temperature of a fire is unrelated to its cause and to whether it was intentional or accidental. Burn patterns, which were central to Willingham’s conviction, are more misleading than they appear. Scorch marks don’t indicate where the fire started; they indicate only that at some point, the fire lingered there. Areas of the most extensive burning are not necessarily where the fire began.

  The first scientifically based report on how to examine fires was published in 1992, six years after the fire at Central Library. Released by the National Fire Protection Association, the report debunked many assumptions about arson. It took special exception to the legal principle known as “negative corpus,” often used as proof that a fire was set deliberately. The term “negative corpus” means, literally, lack of a body. It posits that an event is a crime if there is nothing to prove that it wasn’t a crime. In the case of a fire, negative corpus means that if accidental sources are eliminated, the fire is deemed arson, even when there is no affirmative proof that it was arson. If there is no evidence of how the fire was started, it is assumed that the source of ignition was a lighter or matchbook that was then removed from the scene. It’s like finding a dead body, ruling out the obvious causes of death like a heart attack or stroke, and then declaring it murder even though there is no positive proof that it is murder. This ignores the possibility that the death was caused by something natural that hasn’t been detected.

  Legal scholars and forensic scientists have challenged negative corpus for years. Shaken-baby syndrome is another theory that relies on negative corpus, with disastrous results. The logic behind shaken-baby syndrome works in a reverse spiral, just like it does with arson. If an infant dies and there appears to be no obvious natural causes, police used to a
ssume that someone must have killed the baby by violent shaking, which leaves few discernible signs. The mysterious death would be attributed to an invisible method of killing, rather than to the possibility that babies are fragile and might die for biological reasons we don’t always understand or might need a long time to discover. In the past, a number of parents and caregivers have been convicted of killing infants based on the illogical logic of negative corpus. Ten years ago, medical journals and legal analysts began to dispute the thinking behind shaken-baby syndrome and the legality of negative corpus. Many pediatricians and medical examiners who used to testify for the prosecution in the cases are now testifying for the defense, and many shaken-baby convictions have been overturned.

  The National Fire Protection Association’s report emphasized the danger of misinterpreting where a fire started, especially because point of origin is the key to any fire investigation. Every building contains items that could start a fire. If an investigator declares that the fire started in, say, the center of a warehouse floor or in the middle of a sparsely furnished living room—far from anything flammable—it leads naturally to the conclusion that someone started the fire.

  But reaching this conclusion depends on knowing for sure where the fire began. In the majority of arson convictions that have been overturned, the point of origin was incorrectly identified. In the case of Todd Willingham, the difference between thinking the fire started under the children’s beds and thinking it started on a porch next to a charcoal grill was the difference between life and death. In a 1995 case in Illinois, a man named William Amor was accused of setting a fire that killed his mother-in-law. The fire was so hot that it burned in a state of complete flashover for over ten minutes. Investigators still believed they could identify the point of origin in the room, even though what was left of it was a few sticks of wood and a burned floor. Amor was convicted of first-degree murder and aggravated arson and sentenced to forty-five years, based on the arson team’s statement that Amor must have intentionally dropped a lit cigarette on the floor in order to start a fire. Eventually, his case was reviewed using more exacting science. In controlled studies, pinpointing a fire’s origin in a blaze as hot as the Amor fire had an accuracy rate of between six and ten percent, suggesting it would have been nearly impossible to correctly establish a point of origin. Another study showed that a lit cigarette couldn’t have started the kind of fire that destroyed the apartment. When studied more rigorously, the evidence used to convict Amor fell apart. After twenty-two years of incarceration, he was released in 2017.

  Central Library had bad ventilation and ramshackle floor fans and sizzling light sockets and an extremely high “fire load,” which is the measure of flammable contents per square foot. All of these possible causes of fire were dismissed because investigators determined the point of origin to be a small section of one of the stacks’ bookshelves. Nothing on the bookshelf could have ignited spontaneously, so investigators concluded that the only possible cause of the fire was “an open flame, held in a human hand.”

  But what if the fire at Central Library didn’t start where investigators thought? In 2011, a former firefighter and arson investigator named Paul Bieber founded the Arson Research Project, an organization modeled on the Innocence Project, that examines what it believes are wrongful felony convictions. The Arson Research Project does the same kind of work but concentrates its efforts on arson cases, especially ones in which someone was killed. Bieber likes to call himself a “forensic science nerd.” His skepticism about arson investigation began when he worked on a 1997 case in which a man named George Souliotes was accused of triple-homicide arson. Investigators labeled smudges on the floor as pour marks from an accelerant, even though chemical analysis didn’t find any evidence of accelerants in the house. Souliotes was convicted and sentenced to life in prison. Sixteen years later, after an examination based on the National Fire Protection Association’s new recommendations, the marks were ruled out as evidence of an accelerant; there was no scientific basis for them to have been designated as such. No cause was ever found for the fire, and Souliotes was released.

  “Souliotes was the case that exposed me to the errors in fire conviction and testimony,” Bieber told me recently. “The point of origin was testified to with a certainty that science just didn’t support.” Bieber began to believe that many arson investigators’ testimony consisted of nothing more than well-intentioned professional hunches. He didn’t think investigators were willfully misrepresenting the information, nor were they wrong all the time. He believed the real problem was that they had the wrong foundation for their interpretations. Without science backing their findings, Bieber felt that firefighters’ testimony should be considered ordinary observation rather than expert testimony, which is supposed to be analysis based on repeatable scientific methodology and is viewed by juries as having a special kind of authority.

  Bieber suspected that many arson convictions were based on flawed investigations. The Willingham fire was one of the Arson Research Project’s case studies. Bieber and his staff have reviewed dozens of other arsons since then. When they used scientific methods rather than the old dogma of arson, two thirds of the fires they studied turned out not to be arson. Many convictions had been made in error.

  The statistics on misidentifying arson are alarming. Nationally, the rate is the same as what the Arson Research Project discovered in its cases: About two thirds of the fires that have been reexamined turned out not to be arson. The Exoneration Registry collects statistics about overturned felony convictions beginning with cases that were tried in 1989. So far, it lists fifteen hundred instances of overturned sentences. Thirty of those exonerations were arson convictions. Ten fires were arsons in which the wrong person was convicted. In the other twenty, scientists proved that the fire was caused by something ordinary, like a malfunctioning space heater. In these cases, someone was convicted of a crime that had never happened.

  Bieber told me that many fire investigators think he discounts arson too quickly and is too critical of investigators’ techniques. He understands that it is hard to find good evidence in fires. “It’s very difficult to get around the site of a fire,” he said to me. “And even when investigators get to the source, it’s too hot to really get near it. Then water is dumped on it. Then shelves and furniture collapse, and the area is strewn with rubble. And you try to find evidence in there! It’s crazy to expect that, and I’m telling you, people have been sent to prison for decades based on this kind of information.” Bieber is on the extreme edge of arson theory, but his distance from the mainstream is shrinking as more investigators become convinced that the old way of studying arson is, as Bieber puts it, “bullshit.”

  I compiled a thick file about the fire at Central Library, including reports from the Los Angeles Fire Department and the National Fire Protection Association. The reports described the path of the fire almost minute by minute. Reviewing files couldn’t compare to actually examining the library, and Bieber warned me that it was impossible to draw conclusions from a study, but I still was interested in his opinion. Ever since I’d first spoken to him, I’d wondered about the investigation of the fire at Central. The fire took place in 1986, six years before the definitive NFPA guidelines were published. Ever since, the field has moved away from many of its long-standing suppositions and toward the hard-line, science-based methods the NFPA advised. The conventions of analyzing arson—about how to look at burn marks and fire temperature and concrete spalling and whether no obvious cause proved an obvious arson: assumptions that were passed on from one generation of arson investigators to the next—were upended. Jail doors were swinging open for people who, it turned out, really hadn’t lit their houses on fire. Arson investigation had become a changed field in the time since the library burned.

  I convinced Bieber to read the files on Central; I had been living in the world of this fire and in the enigma of Harry Peak for over four years, and now, if I believed Bieber’s analysis, there
was the possibility that something would help it make sense. A few weeks later, he wrote me a long email: “Under the circumstances described in the report, an area of origin . . . more specific than the general area of the second floor of the Northeast stack is not reasonable. Suspicions based on a more specific area of origin are unreasonable.” He said that from what he could tell, isolating the exact place where the fire started would be impossible, especially because it burned for almost seven hours, incinerating everything in reach. Bieber said he thought it reasonable to say the fire started somewhere in the northeast stacks, where firefighters first detected smoke, but determining an ignition point more specific was unrealistic. Within that large area were a number of items that could have easily started a blaze on their own without any human involvement. In conclusion, Bieber wrote, “After burning for two or three minutes past full-room involvement, where thousands of gallons of water have been pumped into the compartment, cement walls have been breached by jack-hammers, and the book shelves have collapsed on top of themselves in a big pile . . . figuring out exactly where a fire started is a fool’s errand. Not to say people don’t claim they can do it, but they can’t.”

  Once investigators suspected arson, Bieber added, they began looking for evidence to support that suspicion and probably stopped looking at possible accidental causes such as wiring and coffeepots. They believed the fire was started by “an open flame, held in a human hand,” and looked to confirm that. “At the end of the day, I have no idea what started the fire at the L.A. library back in 1986,” Bieber wrote, “but neither do [the investigators].” When I told some of the investigators what Bieber had said, they dismissed it. “Only the gatherer of all the facts is in a position to opine as to the cause of the fire,” said Ron Hamel, the former fire captain who had described the eerie colorlessness of the library fire, adding that anyone who didn’t have access to witness statements and hadn’t examined the scene wouldn’t be able to have a professional opinion.