Tracking an arsonist

It's 3 a.m. on Bogard and Ashe streets, dead center in the arsonist's hunting zone. It's quiet except for the hum of air conditioners and a distant voice. I move closer and see a young man on a porch. He's talking on his cell phone and smoking a cigarette. Ahead, a fan of palmetto fronds blocks the glow from a street lamp; I walk into its shadow, my footsteps scrape the sidewalk.

An arsonist has walked these same blocks, perhaps for as many as ten years. He passed the porches of these tightly packed wooden homes, some of which date to the Civil War. He moved through this same time of stillness, a time after the bars close and before the delivery trucks roll and early risers rise.

At some point, his footsteps turned toward the porches of these homes and crossed that invisible line between public and private. Then came the sound of a match or the click of a lighter. Did his heart beat faster as the flame grew in his hands? Did his hands shake as they moved the flame to a couch cushion or rag?

I've lived near these streets for 15 years, worked in a building blocks from homes that burned, watched how this unique area has been transformed by college students, wondered how one thing remained constant amid this change: the mystery of these fires. How has the arsonist been able to set them undetected all these years? Who is this person? Which house will burn next?

A city task force is trying to put this jigsaw puzzle together but knows it can't be done alone. "We need the community's help," Charleston Mayor Joe Riley said in a news conference after the most recent fire.

With this plea in mind, I've searched for pieces that might fit, looked for clues in police reports, talked with experts about why arsonists do what they do. In the process, I learn that at least 83 suspicious fires have been set in this area since 2002, 27 more than police and fire investigators have so far acknowledged.

Each fire generates data, and because so many fires have been set, a large amount of information is public. These details can reveal new clues about patterns and fires that investigators may have missed.

Police, for instance, have said that the arsonist hasn't killed anyone yet, but a review of old reports turned up a case in which an elderly man died, a case that investigators originally dismissed as an accident but in retrospect has the hallmarks of other intentionally set fires.

While discussing this with investigators, I see the determination in their faces, hear the frustration in their voices that their manhunt so far has failed.

It's also clear that they're sidetracked by the daily crush of crimes and fire calls, and that their system of storing records hamstrings them from building databases on serial crimes, even though past crimes are crucial to understanding when an offender might strike next.

But it isn't until I spend hours early one morning in the arson zone that the mystery sheds some of its veils, particularly why arsonists are so hard to catch in an old city like Charleston, a place of many shadows.

Fire task force members investigate the most recent suspected peninsula arson — at 54 Cannon St. this past July. This is the second time in the past year that there has been a fire at this address. A series of arson-related fires began in 2002 around Line Street. Year after year, more fires were set in this area, baffling police and terrorizing residents. Many of these fires seemed to come in waves. In all but three years, these rashes began in the spring and tailed off in the fall.

The beginning. That's when building puzzles is tough, when all the pieces are scattered about and make little sense. A puzzle is even more difficult when you don't know what it will look like when you're finished. That's what's happening with these arsons. Police don't know if the arsonist is young or old, tall or short; they're not distributing profiles or sketches because they say they don't have enough data and leads.

In this vacuum, residents float theories during neighborhood meetings: The arsonist is a street person, or someone angry about gentrification, or a firefighter, or a newspaper delivery person.

But as the fictional detective Sherlock Holmes warns in A Scandal in Bohemia, "It's a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts."

Over the years, the case of the Crosstown arsonist had generated volumes of data; police and fire reports contained basic information about when and where the fires were set. Newspaper reporters interviewed eyewitnesses. But this wealth of information hadn't been compiled in one place so it could be analyzed for patterns and other clues.

To fill this gap, The Post and Courier asked the city for copies of all reports of suspicious fires and arsons since 2002, citing the state's Freedom of Information Act. Two weeks later, Susan Herdina, a city attorney, said members of the arson task force would have to sift through boxes to find older reports.

Investigators said later that some reports weren't archived in order. Why wasn't this basic information already at their fingertips? Herdina said the city didn't have a computerized system in place until 2007.

The city eventually provided most of the reports. Into a spreadsheet went dates, times, addresses and other details police and firefighters had gathered. To this were added bits from county tax records, the National Weather Service and other information sources.

This new information showed that some fires on the newspaper's previously published lists and maps weren't suspicious and likely had accidental and other causes. Those fires were removed from the paper's list. But the new data also revealed many fires that weren't on the paper's old roster or the city's list, both of which identified 56.

The new tally contained 83 suspicious fires in all.

Putting it together

With this data in one place, some puzzle pieces seemed to fit -- and prompt even more questions.

First, the fires seemed to have a seasonal pattern; they typically started in the spring and tailed off in the fall. This pattern happened every year but 2003 and 2004. Did the arsonist leave town during these periods of inactivity? Or was this what crime experts sometimes call a "cooling off" period?

Second, the fires happened in a surprisingly compact area in downtown Charleston -- the neighborhoods immediately north and south of the Crosstown Expressway. Plotted on a map, all but two fires were within 5,000 feet of each other, a relatively short walk. Why, of all the city's neighborhoods, were the fires happening here?

Third, nearly all the fires involved homes, not businesses. The arsonist targeted older houses; on average the homes that burned were 105 years old, and some had been standing since 1835. And 9 of 10 were rental properties, many occupied by students. Why did the arsonist focus on homes?

Fourth, the arsonist zeroed in on entrances and exits. Four of five fires were started on porches or stairs. Was he trying to trap people?

Fifth, the arsonist set fire to certain types of materials over others. In one of three fires, a couch or chair was set on fire. In one of four, clothing, blankets and other fabrics were used. Paper, debris and mattresses also were popular ignition targets. Was he choosing these materials simply because they were convenient?

Finally, one of the most striking patterns was the time these fires broke out: More than half erupted between 2 a.m. and 5:30 a.m., and nearly eight of 10 were set after midnight. What did this say about who he might be?

Sometimes, non-patterns also are useful clues. Many serial arsons are done for fraud or profit, but a look at tax records revealed no common ownership threads.

Were there any other patterns hidden in this data?

Martin Jones is a mathematics professor at the College of Charleston, a lean and frequently half-shaven man who tosses around terms like "medoid" and "density curves" as if they were something you might overhear in the grocery checkout line.

Jones was particularly motivated to find the arsonist; he lives on Cannon Street next to a house where three students were injured when it was set on fire in 2006. Among other things, Jones and a colleague have used complex algorithms to analyze data on NFL games. (One finding: In 2007, it was a good idea to bet on teams that scored more than 21 points.)

Jones said human beings often behave in predictable ways. When asked to write down random numbers, people tend to do so in patterns anyway, a phenomenon the IRS uses to identify taxpayers who make up numbers on their forms.

"We're not good random number generators," he said. "So the arsonist may be doing patterns that even he's not aware of."

But Jones also warned that data can lead people down false paths, that people often see patterns where none exist, like looking up at the sky and seeing a horse-shaped cloud. Incorrect information about a particular fire could skew the analyses one way or another.

Keeping these limitations in mind, Jones plugged the data into computer programs that tease out clusters and other patterns. One analysis showed that the arsonist had a 42 percent probability of striking again between 3 a.m. and 5 a.m. "This assumes that the arsonist's pattern will be the same as it has been in the past," he said.

Jones analyzed whether the fires shifted over time and found they were centered in the same area year after year, a surprisingly stable pattern. Using GIS data, he also identified the longitude and latitude of the center of the arson cluster, a point near the intersection of Bogard and Ashe streets.

This data shed some light on when the fires were set, and how. But why?

Inside the mind

The era of modern criminal profiling began in 1957 when police asked a psychiatrist named James Brussel for help in New York City's Mad Bomber case. Brussel studied crime-scene photos and mail from the bomber to newspapers and suggested that police look "… for a heavy man. Middle aged. Foreign born. Roman Catholic. Single. Lives with a brother or sister. When you find him, chances are he'll be wearing a double-breasted suit. Buttoned." Brussel was correct on every detail but one: the arsonist lived with two sisters, not one.

After Brussel's coup, criminal profiles became an important investigative tool, and profilers gained national attention for their deductions. One of the most famous is John E. Douglas, former head of the FBI's Behavioral Science Unit and author of "Mind Hunter" and "Anatomy of Motive."

In his books, Douglas writes that the most effective way to solve a crime was to think like a criminal. To do this, he and other researchers interviewed hundreds of serial murderers, arsonists and sex offenders, including David Berkowitz (Son of Sam) and Ted Bundy. As he learned how these twisted minds ticked, Douglas came up with two ways to describe killers and arsonists: Organized and disorganized.

Organized serial offenders usually were intelligent and felt a warped sense of superiority to those around them. They carefully planned their crimes to avoid being caught.

Disorganized killers and arsonists didn't plan their crimes in detail; victims were people who crossed their paths. They typically used materials on hand, such as matches and lighters. They were loners who usually abused alcohol or drugs and set fires near their homes. They were often unattractive and had a disability.

Is the Charleston arsonist organized or disorganized? Some fire scenes were more organized than others. In 2009, for instance, residents called 911 when they heard a smoke alarm in a home at 39 Carolina St. When firefighters arrived, they found a jacket burning in a bedroom. They extinguished it and looked at the breaker box; all of the breakers were on, except the one connected to the smoke detectors. Why had the alarms gone off anyway? The arsonist had failed to remove the alarms' back-up batteries. But with the exception of this lapse, this fire seemed well-planned.

Other fire scenes seemed more disorganized; in one fire after another, the arsonist ignited couches, chairs, trash, mattresses and other items that happened to be on porches or next to houses.

Then again, was the arsonist simply setting fires when he spotted an easily accessible target, as a disorganized offender would do? Or was he carefully planning these fires in advance?

Predator on the loose

More clear is that the arsonist likes to manipulate people. "Arson is often an attempt to gain control and power and attain a feeling of success in their lives," Douglas writes. "Look at all the people an arsonist gets to manipulate and control; the victims of the fire, firefighters, police and other figures of authority, the media and even the community in general."

In an arson case, one of the first things investigators study is whether the structure was occupied. "If he's targeting unoccupied buildings, you've got more of a nuisance-type offender, regardless of the cost of the damage," Douglas says. If he sets fires in occupied structures, then he's like any other angry predator who feels slighted "either for perceived injuries of some sort or simply for not paying proper attention."

No one but the culprit knows exactly when the Charleston arsons began, but 2002 is a useful starting point. That year several fires broke out in homes on and around Line Street. Most were vacant, and vagrants were known to live in some of them; a gasoline jug was found in one charred home. "We think it's just one person," Thomas Reynolds, an assistant fire chief, told a reporter at the time.

Over the next few years, a mix of vacant and occupied homes were set on fire, often with devastating effects on residents. One in 2008 on Sires Street forced two students to jump from second-story windows. A third student ran through flames to escape and had to be airlifted to a burn center in Augusta for treatment.

Seven homes were torched at least two times. In 2008, for instance, a college student at 13 Perry St. reported waking up when she heard an unlocked gate open outside and footsteps on the leaves. She said she looked out the window and saw a white man flick a cigarette through a hole in her window and onto her bed. Three days later, an arsonist set fire to a porch door; two weeks after that, someone started another fire on the porch.

People also have been targeted in the latest rash. Early July 28 at 54 Cannon St., an arsonist set a fire at the base of the steps to the second-floor porch, trapping a College of Charleston student and his dog. Neighbors gathered and held a blanket as flames consumed the house. The student tossed the dog into the blanket and shimmied down a column to safety.

The Cannon Street fire was similar to an attempted arson a month before at 563 Rutledge. In that case, James Burkette, a resident on the second-floor apartment, found a charred blue and white blanket on the porch. Like the Cannon Street fire, it was at the foot of the stairs leading to his unit. "It's freaky because I don't think it was a random place to start the fire," Burkette said. It was the third time this home had burned in the past decade.

Thrills and revenge

In trying to understand the motivations of serial arsonists, criminologists have further broken offenders into subtypes. Some do it for profit or fraud, others are juvenile delinquents. But two of the most dangerous subtypes are revenge-seekers and thrill-seekers.

"If you have fires set for revenge, the pattern for those fires will look different than those by a thrill-seeker," said Dian Williams, head of the Center of Arson Research.

Williams said those who do it for revenge tend to be solitary fire-setters. Often the victim has no idea that someone is upset with them. She cited Timothy McVeigh, the Oklahoma City bomber, as an example of a revenge-seeker.

A revenge-seeker's goal is to get even. "As soon as they set fire to someone's home, they're done and they go to the next person," she said.

Thrill-seekers are among the most misunderstood subtypes, she said. "When I lecture around the world, people and investigators often think the culprit is motivated by a fascination with fire. But when you sit down face-to-face with them and do a comprehensive assessment, you find that what they're really fascinated with is getting away with a crime."

Thrill-seekers relish seeing their exploits in the news; sometimes they collect photos and news articles to show how clever they are, she said. This often puts news organizations in a difficult spot; the arsonist feeds off the news coverage, but if journalists work with police to deliberately under-report what's happening, arsonists sometimes set bigger and more destructive fires.

Thrill-seekers might case a neighborhood during the day, looking for items to ignite -- furniture on a porch, debris; they also tend to use accelerants. "They like fast, hot fires, which demonstrate their superiority over firefighters," she said. And they often set fires in the middle of the night. The possibility of being discovered increases the adrenaline rush.

She and other criminologists add that serial arsonists leave behind clues every time they set a fire, even when flames consume evidence, and that merely studying a crime's location can sometimes help detectives find their marks.

"There is a relationship between where a crime takes place and where a person lives," said Kim Rossmo, a former Canadian detective who heads Texas State University's Center for Geospatial Intelligence and Investigation.

Rossmo theorizes that many serial offenders live close to their victims, but not too close. He created programs to identify a killer's "hunting area," along with a "buffer area" around the killer's residence.

Serial arsonists tend to be good candidates for these geo-mapping profiles. "The predictive power of the model is related to the number of crime sites -- the more locations, the better the performance," he wrote in a book "Profilers."

Rossmo declined to study the Charleston arsons; he works with law enforcement agencies. But with Jones' help, the geometric center of the arsons could be identified. What would an arsonist see as he moved about this area?

It's 4 a.m., and I walk near Ashe and Line streets, the center of the arson zone. A man strides slowly on the sidewalk, appearing and disappearing as he moves through the shadows. He's thin and wearing a striped long-sleeved shirt even though it's humid and warm. "Got a cigarette?" he asks. Moments later, a man in his late teens or early 20s rides by on his bicycle. With him is an older woman in a short dark skirt. Looking north on Rutledge, the Crosstown's six-lanes look like an open field compared to the narrow streets south. A taxi drives by, and I move across the Crosstown toward a house that may have been the scene of a homicide.