The fire burned quickly and ferociously in the early hours of July 29, 1989, nearly consuming the small cabin at a religious retreat in the Poconos.
Inside was 20-year-old Ji Yun Lee, dead.
Investigators soon concluded that the fire had been set. Han Tak Lee, a Korean American businessman who had taken his daughter there in the hope of finding help for her escalating mental problems, was arrested, tried, and convicted. He is 21 years into a life sentence.
Lee has steadfastly professed his innocence. His lawyers contend that he was convicted by junk science - that research since his 1990 trial has debunked many of the once-ironclad indicators of arson.
On Monday, a federal appeals court wrestled with Lee's case - specifically, whether he should be given a new hearing to present evidence about the changed understanding of how fires burn, and whether he should be freed outright.
Defense attorney Peter Goldberger argued that Lee had been convicted only because of the testimony of fire investigators in Monroe County, and that their findings would not hold up today.
"Every credible and competent scientist in the field says that what we thought . . . 20 or 30 years ago was wrong," Goldberger told the three-judge panel of the U.S. Court of Appeals for the Third Circuit in Philadelphia.
Monroe County Assistant District Attorney Jeremy Bolles acknowledged that fire science is evolving, but said it is still not as "definitive" as DNA.
The Lee case, Bolles said, could not be retried, because so much time has passed and physical evidence taken from the fire scene probably no longer exists.
Lee is not the only Pennsylvania inmate who has challenged an arson conviction because it was based on passé science. Others include Philadelphian Daniel Dougherty, on death row for setting a fire that killed his two young sons; Gregory Brown of Allegheny County, serving life for starting a blaze that killed three firefighters; and James C. Young of Westmoreland County, also serving life for setting the fire that killed his wife, stepson, and infant son.
Doubts about the tenets of arson science surfaced in the 1980s, when investigators began testing fires and watching what happened.
By 1992, the National Fire Protection Association had issued a new guide of professional standards. But almost a decade went by before the standards were endorsed by the U.S. Department of Justice.
Burn patterns on a floor, for example, were once considered a reliable indicator of an accelerant. Investigators now know not to read too much into those marks, for they are common when a room becomes engulfed - even in an accidental fire.
Finely cracked or "crazed" glass was viewed as a sign of rapid heating caused by an accelerant. Scientists have found that the effect results from rapid cooling when water hits hot glass.
Low burning at doors also was thought to point to arson. Now, it is deemed normal in rooms that are engulfed.
Investigators examining the aftermath of the fire that killed Lee's daughter found eight or nine burn patterns on the floor, and also noted crazed glass and low burning. But in a 1999 fire-science publication, a nationally known fire investigator, John J. Lentini, disputed the findings as "old wives' tales" and called the case the "ultimate triumph of junk science."
Lentini later submitted a 41-page affidavit on behalf of Lee to the defense lawyers.
In court Monday, Judge Dolores K. Sloviter asked the prosecutor whether any scientific evidence rebutted Lentini's conclusions.
"I don't think we ever got to that," said Bolles.
"Don't you think we should?" asked Sloviter. "A man is in jail for the rest of his life."
Bolles also told the panel that traces of accelerant were found on Lee's pants. Goldberger said the residue was hydrocarbons that could have come from materials that burned before Lee fled the fiery cabin.
Sloviter expressed surprise when Bolles said no one had checked whether any evidence still existed. "If you don't produce the evidence, shouldn't we presume it's not favorable to you?" she asked.
The court had been concerned about the changing nature of arson investigation when it considered another arson-murder case a few years ago, Sloviter said, but "we didn't have as much as we have in this case."
Even as the cabin smoldered, Lee seemed stoic to investigators.
Jurors heard testimony about his demeanor during his 1990 trial. Two jurors told The Inquirer in 1992 that they found the portrayal of Lee's impassivity as influential as the physical evidence of suspected arson.
On Monday, Kathleen Chung of Northeast Philadelphia was among a handful of Lee supporters who attended the oral argument. She said that Korean men typically do not show emotion, and that Lee's demeanor was misunderstood.
"I hope this oral argument will be a Christmas present to Han Tak Lee," said Kyung Sohn of New York City. Lee, he said, has been doing "nothing but praying" that he will be exonerated.
The panel, which also included Third Circuit Judge Thomas Vanaskie and U.S. District Judge Lawrence F. Stengel, gave no indication of when it would rule.