The Arson Evidence Doesn’t Hold Up. Florida Is About to Convict Her for Murder Anyway.

Megan Wallace had just been booked at the St. Johns County Jail in St. Augustine, Florida, when she started hearing gossip about its most notorious resident. Michelle Taylor, then 34, had allegedly set fire to her own house in 2018, killing her 11-year-old son. The motive was insurance money. Everyone at the jail seemed disgusted by her. “The guards treated her like shit,” Wallace said.

A mother herself, Wallace vowed to stay away from Taylor. But after a couple of months, Taylor was moved out of solitary confinement and into her cellblock. “The stories I’d heard didn’t add up to how she was in real life,” Wallace said. Taylor was withdrawn and heavily medicated. Other women at the jail were openly cruel toward her, but she didn’t lash out. “She slept all day and wouldn’t get up for breakfast or lunch.”

Wallace knew how it felt to be judged by people who didn’t have all the facts. She had spiraled into addiction after the sudden death of her husband, culminating in her arrest two days after Christmas in 2022. Prosecutors accused her of drug trafficking, which she insisted was bogus. As Wallace fought her own charges, she started to feel sorry for Taylor. “All she did was cry about her son,” Wallace said. “She was like, ‘I don’t want to live.’”

Wallace eventually opened up to Taylor about losing her husband. They formed a bond that strengthened over time. As Wallace got to know Taylor, she seemed less like a monster and more like a grieving mother who had suffered unspeakable trauma. David wasn’t the only child Taylor had lost. Her middle child, Natalie, who was born with cerebral palsy, died in a tragic accident five years before the fire. News reports about Taylor mentioned her daughter’s death, leading to callous comments online and lurid rumors at the jail. “People said she drowned her daughter in the bathtub and locked her son in a closet and took off the door handle,” Wallace said.

Taylor didn’t talk about the fire in jail. But she’d always sworn she had no idea how it started — she barely escaped herself. Although Wallace had no way to know the truth, it seemed obvious to her that Taylor had loved her children and her home. By the time Wallace saw her own charges dropped in the summer of 2023, she felt certain that the fire had been an accident and that Taylor had been wrongly accused.

Back home, Wallace started reading everything she could about arson cases. She learned about people who had been wrongfully convicted based on junk science. And she discovered that the Florida state fire lab, which examined the evidence in Taylor’s case, had once lost its accreditation after misidentifying gasoline in numerous cases. One name came up over and over again: John Lentini, a renowned Florida fire scientist who had helped exonerate people all over the country. In October 2023 she wrote him an email with the subject line “Please help.”

As it turned out, Lentini had been contacted about the case before, by a defense attorney who no longer represented Taylor. At that time, Lentini was skeptical he could help; there appeared to be overwhelming evidence of arson. According to the lab, a dozen fire debris samples taken from Taylor’s home contained gasoline.

Although Lentini was a fierce and vocal critic of the lab, he found it hard to believe that it would produce a report containing so many false positives. In a call with Taylor’s new lawyer, he offered to examine the underlying data from the fire debris samples — but he was doubtful it would change much. “If there really was that much gasoline in the house,” he told Wallace, “there is nothing I can do.”

But on January 4, 2024, Wallace received an email from Lentini. “Michelle is not guilty,” it read. “The lab report that says they found gasoline is bullshit. Every part of the state’s case rests on that.”

Today the state of Florida is prepared to convict Taylor for killing her son, despite the fact that the only direct evidence of arson has been thoroughly discredited.

More than a year after Lentini raised red flags about the state’s case, writing in an expert report that it is based on “unreliable methodology and incorrect opinions regarding the presence of ignitable liquid residue,” several other leading fire experts have agreed with his conclusions. They include two different lab directors who are also veteran forensic chemists. One of them examined the data from the state lab, while the other retested the carbon strips used to analyze fire debris from the scene. Both recently submitted reports to Taylor’s defense attorney saying there is no evidence of gasoline. 

The lab’s conclusions have also been contradicted by two forensic chemists with the federal Bureau of Alcohol, Tobacco, Firearms, and Explosives, who reviewed Lentini’s expert report and agreed in depositions last April that the gasoline findings were unsupported.

The lab has not responded directly to these claims. But in January, the analyst who examined the fire debris evidence in Taylor’s case submitted an amended report backtracking on some of her original findings “due to the re-evaluation of the data.” Four of the samples she previously said were positive for gasoline were now determined to be negative for any ignitible liquid. A spokesperson for the Fire Marshal’s Office did not respond to a detailed list of comments in time for publication.

Despite the ongoing dismantling of evidence supporting its arson case, prosecutors with the State Attorney’s Office for Florida’s 7th Judicial Circuit have refused to drop the charges against Taylor. They insist she is a habitual liar and a fraud — a mother so diabolical she was willing to set fire to her house with her children inside to “further her lifestyle.” They point to a paper trial that proves her willingness to commit arson “in an attempt to fraudulently collect insurance funds. Her 11-year-old-son … died as a result.” Yet the investigation carried out on behalf of Taylor’s homeowner’s insurance company did not find evidence of arson either. Fire debris analysis conducted at a private lab in the aftermath of the fire revealed no gasoline or other ignitable liquid.

To Lentini and others who have worked on wrongful convictions, the case against Taylor is a version of an all-too-familiar story. In the absence of reliable forensic evidence, prosecutors put together a circumstantial case that can convince a jury a defendant is capable of murder, even if the science does not add up. In cases where a child has died, which are especially emotionally charged, it does not take much to cast parents or caretakers in a suspicious light. Mothers who escape a fire without their children are often judged harshly for that fact alone — and those with a checkered past are easier still to demonize, especially when the state can show that they were guilty of fraud. In the case of Angela Garcia, a Cleveland woman tried three times for killing her daughters in a fire, prosecutors seized on evidence of financial fraud to win a conviction and life sentence despite the fact that there was no reliable evidence of arson. Her sentencing judge accused her of treating her daughters “like coins in a slot machine.”

In Taylor’s case, prosecutors have made clear that they plan to tell the story of a bad mother who senselessly sacrificed her only son for material gain. But that theory eclipses a different story, one with frightening implications for anyone who survives a fire in Florida. According to Lentini, Taylor’s case is the sixth time he has personally seen a misdiagnosis of gasoline by chemists at the state fire lab — and there is reason to believe there are many more. A review of the lab in 2016 by the nation’s leading accrediting body for crime labs found that analysts were using flawed methodology to identify gasoline in fire debris evidence. Out of 26 cases in which the lab had identified gasoline, more than half did not stand up to scrutiny.

Yet rather than reconsider its case against Taylor, prosecutors have sought to suppress any reference to the lab’s problematic history when the case goes to trial. “The State respectfully requests that no party or witness in the case be allowed to comment on the prior loss of accreditation by the State Lab,” Assistant State Attorney Jennifer Dunton wrote in a pretrial motion last spring, arguing that it was “not relevant.”

With a trial date set for June, Taylor and her attorney were reluctant to speak on the record about the case. But the likeliest scenario is that neither side will have a chance to tell their story to a jury. After long refusing to consider a plea negotiation for a crime she insists she did not commit, Taylor has been forced to confront the risk of going to trial against prosecutors intent on proving their case with or without reliable evidence of arson. In Florida, a first-degree felony murder conviction carries a mandatory life sentence. With her next court date set for April 2, a guilty plea to a lesser charge and years in prison may be the best of a bad set of options.

The first 911 call on the night of the fire consisted mostly of harrowing screams. “My house is on fire!” cried Taylor’s 18-year-old daughter, Bailey, wailing that her brother was inside.

The fire was at 1041 Lee St., just east of downtown St. Augustine. On a recording of the call, which was placed at 9:42 p.m. on October 23, 2018, the screaming continues while a 911 operator tries to get information. After two minutes, Bailey says she needs to call her dad and hangs up.

The second 911 call came two minutes later from Heather Anderson, who lived across from the Taylors — the last two houses on a dead-end street in a residential neighborhood just a few miles from the beach.

Taylor had shown up at her door begging for help. There was soot on her face, and she was hysterical. Anderson’s husband raced to the house in search of David. But now the fire was spreading, Anderson said, her voice trembling. She saw the Taylor’s small chihuahua Milo run out of the house. But David was nowhere to be seen.

Taylor had fled the house without her phone or keys. Witnesses described her frantically trying to get back inside. A St. Augustine police officer who was first on the scene tried to ask her questions to no avail. “Due to the emotional state of Michelle and Bailey, I was unable to gather any information in regards to how the fire started or why David was not able to get out,” he later wrote in a report.

Firefighters arrived at 9:49 p.m. Ten minutes later, they found David’s body, lying face down and covered in debris. He was unrecognizable.

In Florida, fatal fires are investigated by the state fire marshal’s Bureau of Fire, Arson, and Explosives Investigations, which dispatches personnel around the clock. The bureau also sometimes calls upon the ATF to provide additional resources — “an extra shovel,” as the supervising lieutenant said in a deposition in Taylor’s case. That night, the BFAEI called ATF Special Agent Kristie Calhoun, who drove out to the scene from Jacksonville. She would be in charge of determining the cause of the fire.

The Taylors lived in a modern, one-story home with a stone facade. It had an open floor plan, with the kitchen to the right of the entrance and a hallway leading to four bedrooms on the left. Toward the back was the living room and the area most heavily damaged by the fire, which investigators would label the area of origin. The drywall was gone from the walls, and “the entire drywall ceiling had fallen to the floor, exposing the wooden roof structure,” Calhoun wrote in her report. What was left of the insulation was all over the ground.

Around 1:20 a.m., Mason Patrou, a rookie detective with the St. Johns County Sheriff’s Office tried to interview Taylor at a local hospital, where she and Bailey were being treated for minor burns and smoke inhalation. “Michelle was very hesitant to speak with us and clearly displeased with our presence,” Patrou later wrote.

Until the fire, the evening had not been out of the ordinary. Taylor, who’d recently lost her job as a custodian for the local school board, had a meal with her husband Dennis and their children at Texas Roadhouse. Upon coming home, Taylor briefly ran out to buy lottery tickets. Dennis got a call from a friend who’d shot a deer and wanted help retrieving it; he left around 8 p.m. Taylor and David went to her room to watch TV. At one point, Taylor said she went outside to look for a homework assignment David said he’d left in the car, then settled back into her bedroom, where they watched back-to-back episodes of the sitcom “Mom.”

It was during the second episode that Taylor said she heard the smoke detectors go off. She opened her bedroom door to find thick, dark smoke. “Michelle said her and David went out of her bedroom, banged on Bailey’s door and they all tried to get out the front door but they couldn’t get the door open,” Patrou wrote. “It was so black you couldn’t see anything.” They turned down the hallway “when David went back for the dog.” She lost sight of him. She and Bailey then ran to the spare bedroom and climbed out of the window.

Taylor said she had no idea how the fire started. But there were a few possibilities. “Michelle stated that there was a candle lit on the coffee table, near the couch,” Patrou wrote. Asked if there was anything flammable nearby, “she cited a white runner on the table and the dog, which runs loose and jumps on couch.” There was also an old electric recliner in the living room inherited from Taylor’s grandmother, which plugged into the wall and constantly malfunctioned. Dennis had recently tried to repair the electrical wires, splicing and reconnecting them using heat shrink wrap and a glue gun.

Finally, Taylor sometimes smoked in the house. Although it was unclear whether Taylor was smoking that night, a fire marshal’s detective would report that a lighter and cigarette butt had been found in the debris.

All of these scenarios were consistent with the area of origin. In the back corner of the living room, the couch and the recliner had been almost entirely consumed by the fire, with only their metal frames remaining.

Nevertheless, suspicions of arson had crept in by the time the sun came up on October 24. A K9 handler had swept the home with his accelerant-detecting dog, a black labrador named Fresca, who alerted in five different places. Fire debris samples were collected in each spot, to be examined at the state fire marshal lab, which would determine whether there was proof of an ignitable liquid.

In early November, three samples came back positive for gasoline.

The Florida State Fire Marshal’s office is governed by Florida’s Department of Financial Services, which oversees fraud and fire investigations alike. At the time of the fire, the Chief Financial Officer and fire marshal was Jimmy Patronis, whose most recent annual report boasted “an arrest rate for the crime of arson well above the national average.”

Florida is only one of two states in the country whose fire marshal has its own lab for fire debris analysis. Although numerous agencies are involved in fire investigation, most forensic evidence is processed at the bureau’s state-of-the-art facility in Havana, just outside Tallahassee.

In mid-November, Calhoun and a team of investigators returned to the scene of the fire. For three days they excavated the site, collecting fire debris for additional testing. Seventeen samples were placed into tightly sealed metal cans and sent to the lab.

Most fire debris analysis works roughly like this: Cans are heated in an oven. The resulting vapors are captured on charcoal strips suspended from the top of each can. The strips are then injected into a machine called a gas chromatograph/mass spectrometer. The former (GC) separates compounds in a given sample, while the latter (MS) detects the compounds, ultimately producing an electronic signature called a chromatogram. The resulting graph looks like an irregular combination of peaks and valleys, which mark the chemical compounds in the sample. The patterns are compared to graphs of known substances to see if there is a match.

The process is straightforward in theory. But in practice, interpreting the data can be fraught with uncertainty. Modern homes are replete with synthetic materials, which make up everything from shoes to rugs to couch cushions. Many are made from petroleum-based products that share some of the same compounds as ignitable liquids. When these products burn and break down in a process called pyrolysis, the compounds contribute their own peaks to a chromatogram. Because today’s GCMS technology is highly sensitive and capable of picking up trace amounts of material, chromatograms from a fire debris sample are often crowded with detail. One training video by the National Center for Forensic Science at the University of Central Florida says “finding an ignitable liquid residue in fire debris is like searching for Waldo in the ‘Where’s Waldo’ puzzles.” The difference is that, unlike an ignitable liquid, Waldo is always somewhere to be found.

Gasoline, which is the accelerant most commonly used to commit arson, is a complex mixture that is especially tricky to identify within fire debris. Lab analysts have long abided by specific parameters when examining a sample for gasoline, starting by ensuring there are five specific peaks on a chromatogram, which must appear at certain ratios. But the interpretation is ultimately highly subjective.

In a 2019 podcast by Chemical & Engineering News, ATF chemist Michelle Evans said the difficulty of gasoline identification had made her more cautious in her analysis over the years. She told the host she was working on a case where the data was a little too ambiguous. “I think it’s there, but I don’t think I have enough to call it,” she said. “Maybe when I was starting out I might have actually identified gasoline. But in looking at the data, I don’t feel comfortable having to testify to that.”

Among the new samples sent from Taylor’s home to the Florida lab were two different types of vinyl flooring found in the house. A pair of boxes containing the same flooring had been found in the garage, pieces of which were collected as comparison samples. Such samples are key to distinguishing compounds found in synthetic materials from those found in an ignitable liquid. “Comparison samples allow the laboratory to evaluate the possible contributions of volatile pyrolysis products to the analysis,” Calhoun wrote in her report.

Comparison samples, in other words, are supposed to be clean of any ignitable liquid. So when the second set of lab results came back, something was clearly wrong. The flooring found boxed up in the garage had come back positive for gasoline. A third comparison sample, described as “underlayment from hallway closet floor,” also came back positive for gasoline.

Calhoun’s report attributed the false positives on the tiles to “soot staining,” attaching a photo that showed evidence of smoke in the garage. Yet the boxes were barely visible in the picture, let alone the tiles inside. In a deposition, the lab analyst who examined the flooring said she did not recall seeing soot on the comparison samples. Nor did she think that light smoke on the boxes would necessarily contaminate the tiles inside. “I mean I’m not, like, an expert in soot,” she said.

Lentini would later argue that the results from the comparison samples should have invalidated all of the gasoline findings. The soot explanation did not account for the false positive on the third comparison sample, he wrote in his report. What’s more, “if ‘soot’ can cause a false reading of gasoline, that can be applied to all of the debris samples.’”

But Calhoun did not see it this way. In December 2018, she obtained new vinyl tiles matching one of the previous comparison samples and submitted it to the lab. In February 2019, that one came back negative.

By then, the St. Johns County Sheriff’s Office was convinced they had an arson case on their hands. Patrou, the lead detective, saw red flags in the family’s financial records. At the time of the fire, the Taylors had been behind in their mortgage payments. Their bank was threatening foreclosure if they did not pay by early 2019. Despite having enough money in the bank to cover the mortgage, financial records showed that six local churches “had made payments on behalf of the Taylors.”

But even more disturbing were interviews with Dennis’s side of the family, who immediately blamed Taylor. There was bad blood between Taylor and her mother-in-law; the two had almost come to blows on the night of the fire. Patrou wrote that Dennis’s mother told him she believed “Michelle deliberately burned the house down because Michelle didn’t have the money to do what she wanted.” Dennis’s sister agreed.

The investigation into Taylor stretched into 2020, only to be derailed by the Covid pandemic. Calhoun’s final 45-page report was submitted in 2021. It synthesized the findings from the state fire investigators and the St. Johns County Sheriff’s Office, concluding that the fire could not have been an accident.

It had all happened too fast, Calhoun concluded. A timeline from the night of the fire had been put together through surveillance footage taken from the neighborhood’s security cameras. It showed Taylor going out to her car at 9:33 p.m. Only six minutes later, at 9:39 p.m., video showed smoke coming from the back of the house. The first 911 call came three minutes later. There was too little time between Taylor’s last appearance and the signs of the fire captured on the tape. “An ignition source such as a cigarette or candle scenario would be a slow growing fire,” Calhoun wrote, and if a small fire was already underway before Taylor went to her car, she would have spotted it on her way back to her bedroom.

There were at least some scenarios in which an accidental fire might have quickly overtaken the house. A cluster of living room furniture was near two windows and a pair of French doors — potentially powerful fuel, Calhoun noted. But the gasoline findings rendered other explanations moot. As Calhoun eliminated each accidental cause one by one, she wrote that each was “inconsistent with the presence of gasoline as confirmed by the laboratory analysis.”

In August 2021, Taylor was arrested for felony murder.

Fire investigators once described their work as an art more than a science. The process of determining the origin and cause of a fire relied as much on an investigator’s experience and intuition as on the evidence at the scene. When it came to uncovering arson, many fire investigators falsely believed they could figure out what happened based solely on visual indicators like “pour patterns” — places where an ignitable liquid had supposedly been poured, which caused a fire to burn hotter and faster than normal.

Lentini, who began his career as a forensic chemist before moving into fire investigation, had been trained to look for such signs of arson too. But in 1991, he worked on a case in Jacksonville that dramatically transformed the understanding of fire behavior, revealing how quickly an accidental fire could reach “flashover” — a critical concept often summarized as the moment a fire in a room becomes a room on fire. In the right conditions, a smoldering fire can reach flashover in under four minutes.

Today fire investigations follow the scientific method, as set out by the National Fire Protection Association in NFPA 921: Guide for Fire Investigations. But for those sent to prison based on misconceptions about fire behavior, the advances in the field came too late. The most famous wrongful arson conviction is that of Cameron Todd Willingham, who was accused of setting a fire in Corsicana, Texas, that killed his three children. Despite a race by fire experts to prove that his case was based on junk science, Willingham was executed in 2004.

That same year Texas executed Willingham, a fire broke out in Palatka, Florida, killing a woman named Tscharna Hampton. Her boyfriend, Randy Seal, was accused of dousing her with gasoline and setting her on fire. As in the Willingham case, fire investigators relied on discredited arson indicators. And as with Willingham, the state announced it would seek the death penalty.

Like Taylor, Seal was prosecuted in the 7th Judicial Circuit of Florida. Lentini was hired as an expert for the defense. A lab analyst had identified gasoline in numerous samples. But when Lentini received the data, it was way off. “I looked at it and I thought, ‘This is not gasoline,’” he said. 

Lentini examined the carbon strips used to analyze the fire debris. He found no signs of gasoline. At the trial, he used a PowerPoint presentation to break down the science in the most accessible terms for the jury. “You go down to the gas station, you buy some gasoline, you evaporate it, you run it through your machines, you get a pattern,” he explained. In order to identify gasoline in a sample, “you’ve got to be able to match that pattern.”

“In the early days, the methodology was more or less: ‘The sample displays sufficient similarities to the standard for me to conclude that they are the same,’” Lentini went on. But this was entirely too subjective. Rules for identifying ignitable liquids in fire debris were developed by the American Society for Testing and Materials, which laid out criteria for gasoline, starting with a group of five compounds, whose peaks on a chromatogram had to appear in specific proportion to one another. “This is not a suggestion. This is not a guide. This says if you want to call it gasoline, you’ve got to have this stuff.”

While the Florida state lab purported to abide by this rule, known as ASTM E1618, the gasoline results showed otherwise. Not only were the ratios wrong, Lentini testified, but three of the necessary compounds were not present at all. In a deposition, a different expert later described the lab analyst’s approach as “peak pick.” If you zoom in close enough, the expert explained, virtually any chromatogram from a fire debris sample will contain patterns resembling gasoline. “It’s just, Let’s blow it up and let’s see what components show up,” he said.

With no real rebuttal to the chemistry, the prosecutor goaded Lentini on cross-examination, daring him to smell the blue jeans worn by Hampton when she died. “You’re not willing to do it, are you?” he asked. Lentini replied with obvious irritation. “They smell like burnt cloth,” he said. More importantly, “they don’t look like gasoline when analyzed with the gas chromatograph.” The prosecutor replied that jurors could use their “common sense” to conclude that the clothing smelled like gasoline.

Seal was found guilty but avoided the death penalty. A judge sentenced him to life in prison.

The Seal case was the first time Lentini saw the state lab analysts find what he has wryly called “Florida gasoline.” It would not be the last.

One year after Seal was convicted, a fire broke out inside an apartment complex outside Fort Lauderdale. Eleven-month-old Jada Reynolds was found in her crib, dead from smoke inhalation. Her mother, Sasheena Reynolds, had escaped the fire while carrying her 2-year-old son. Witnesses described her panic. “She kept saying, ‘My baby is inside the fire!,’” a neighbor told a reporter.

At first there was an outpouring of support. Hundreds of people went to the baby’s funeral and neighbors raised money for the family. Reynolds’s electricity had been cut off at the time of the fire; she relied on scented candles to illuminate the home, which were believed to have caused the blaze.

But everything changed when the fire marshal lab found evidence of gasoline. Reynolds was arrested and charged with first-degree murder. Investigators said they’d suspected the fire was no accident given “how quickly the fire spread and burned.”

Lentini was hired by Reynolds’s defense attorney. The lab results echoed what he had seen in Seal’s case. Rather than face trial, Reynolds agreed to a plea deal and was released from jail in 2013. By then, Lentini had seen false gasoline findings in two other cases.

In 2015, Lentini sat down to write a letter to the American Society of Crime Laboratory Directors/Laboratory Accreditation Board. “After several weeks of careful consideration, I have concluded that it is my professional obligation to bring to your attention a serious issue,” he wrote. He warned that the lab had “a continuing laboratory policy of identifying gasoline where none exists, based on a subjective and highly speculative thought process.” Although the lab claimed to follow ASTM E1618 in theory, in practice it deviated completely from the standard. “I suspect that erroneous identifications of gasoline happen on a routine basis.”

There is no federal agency that regulates forensic labs. Accreditation bodies like ASCLD/LAB, which has since been renamed, are private entities whose leadership is closely tied to professional forensic associations and which charge labs to receive accreditation. Although labs must demonstrate certain protocols and best practices to remain in good standing, there is little oversight of the work itself. Indeed, at the time Lentini filed his complaint, the lab’s accreditation had just been renewed.

In January 2016, a team from ASCLD/LAB traveled to Havana, Florida, to visit the lab. It interviewed lab analysts as well as its director, Carl Chasteen. Two months later, reviewers reported their findings. “The interpretation methodology being employed by the laboratory is an undocumented, unvalidated protocol that is not generally accepted in the scientific community,” they wrote. The numbers were alarming. “Of twenty-six(26) randomly selected cases in which gasoline or gasoline mixtures were reported, there were fourteen in which the interpretation of the data by the analyst resulted in an unsupported conclusion.”

The lab lost its accreditation as a result of the review. Chasteen appealed the results, writing in a 50-page letter that the review appeared “to repudiate all of our work over the years, our professional reputations, and our personal character.” A new panel of reviewers upheld the decision to revoke the lab’s accreditation.

Lentini considered the lab’s suspension a victory. But it was short-lived. After agreeing to a “corrective action plan,” Chasteen won back its accreditation. He also successfully sought accreditation from a second organization. By 2017, the lab had gone from being mired in scandal to boasting that it was the only public lab in the country dual-accredited by both organizations.

The saga had a critical impact in at least one case, however. The Florida Innocence Project filed a post-conviction appeal in Seal’s case based on the discoveries about the lab. New experts analyzed the evidence and concluded that the findings of gasoline had been wrong. In 2023, prosecutors offered Seal a plea deal and he accepted.

Before his release in 2024, Seal sent Lentini a handwritten note from prison. “I knew you were always right,” he wrote. “The lab was junk, still is. Thank you for what you did for me.”

Last spring I traveled to St. Augustine and drove to the site of the fire. The neighborhood is surrounded by longleaf pines, with modest homes set back from the road on green lots dotted by palm trees. Google Maps had captured the house after the fire, with boarded-up windows and piles of debris on blue tarps on the driveway. But when I arrived at 1041 Lee St., there was nothing but a concrete driveway leading to a gray slab.

Wallace told me that Taylor once described the house as her “dream home.” The family moved there in 2012 from a more rural area, mainly out of necessity. According to Dennis and Bailey, there was a man in the area who was a known pedophile; he had once tried to lure David into his car. “The cops said they couldn’t do anything about it,” Bailey told me. So they decided to move. “My mom and dad said it was to protect us.”

It took them six months to find a place they could afford, according to Dennis. The house on Lee Street was a foreclosure sale, which needed a lot of work. They painted and redid the floors and installed planters, a picnic table, and a bird bath outside. Because her daughter Natalie loved butterflies, Taylor took butterfly lawn ornaments she’d found at Dollar Tree and used them to decorate the entrance to the house.

I met Wallace at a Hampton Inn off I-95. She brought a small stack of papers: police reports and testimonials about Taylor from family and friends, which she had collected for an upcoming bond hearing. They described Taylor as loving and generous to friends and strangers alike. “My mother never got to grieve properly,” Bailey wrote. “She would never hurt anybody, please just let her come home.”

Wallace said Lentini had breathed new life into Taylor’s defense. But she was also frustrated — she could not understand how prosecutors could stick to their case. She’d recently learned that, back in January, Calhoun had forwarded Lentini’s report to two different forensic chemists at the ATF, both of whom had sat down for depositions. One of them took issue with Lentini’s claim that the state lab “routinely identifies gasoline where it does not exist.” But he did not disagree with Lentini’s conclusions about the chromatography data. “The patterns that I’m seeing in the data do not appear to be gasoline to me,” he said.

The other analyst was Michelle Evans, who gave the 2019 podcast interview in which she said she’d become more cautious in her gasoline findings over time. She concurred that the data did not show the patterns required to identify gasoline. “This strongly supports there not being gasoline there.”

Not long after my trip to St. Augustine, Taylor was released on bond after nearly three years in jail and reunited with her family. As her joy and relief began to fade, so did her hope that prosecutors would ever drop her charges. The fear of a life sentence has given way to anguished uncertainty about whether to plead guilty after all. More than six years after the fire, Wallace said, “she just wants it to be over.”

On March 3, which would have been David’s 18th birthday, Taylor and her family met at the cemetery where he is buried alongside his sister Natalie, under a large oak tree dripping with Spanish moss. They ate cake and wrote messages on Mylar balloons, which they released into the air.

Wallace shared some photos via text. In one, Taylor stands smiling with Bailey, her sole remaining child, over brilliant yellow flowers.

The group at the cemetery included Taylor’s mother, four sisters, and her nieces and nephews, some of whom she helped raise. In phone calls, friends and relatives told me the state’s case never made any sense to them. Taylor wasn’t seeking a lavish lifestyle. Her whole life revolved around her home and her children. “Wherever they went they took their kids,” her mother told me. “She loved the ground her kids walked on.”