The dilapidated three-decker was ablaze in a matter of minutes. Trapped inside, children screamed. Onlookers were desperate to help, but the flames were too strong, the smoke too thick. One bystander, a young Puerto Rican man named Victor Rosario, shattered a front window with his right fist, but the blast of heat knocked him back.

It was 1 a.m. on March 5, 1982, and the deadliest fire in Lowell history was raging.

Four bodies, a mother and her three baby boys, were later discovered on the first floor; four more corpses on the third. Arson was strongly suspected, and the community cried out for justice.

Lowell police swiftly delivered.

Just 48 hours after the fire, Rosario, the 24-year-old bystander, was detained as the prime suspect. After five hours of questioning, he signed a statement that he and two friends threw Molotov cocktails into the building as revenge for a botched drug deal.

Within a year, Rosario would be convicted of arson and eight murders and sentenced to life in prison. The city was relieved and in awe of investigators’ determination and expertise.

“The investigating team has shown that Lowell has a capability second to none for dealing with arson cases,’’ The Lowell Sun editorialized.

But an investigation for the Globe by the New England Center for Investigative Reporting has found a series of grave shortcomings in the police inquiry and the prosecution of Rosario, findings that suggest Rosario was wrongfully convicted and that buttress legal efforts underway to reopen the case. Among the findings:

  • Lowell police quickly concluded that the fire had to be arson, but their investigation relied on assumptions about fire-scene evidence that have since been scientifically discredited. Burn patterns police took to be definitive evidence of arson are now considered common byproducts of house fires, whether set or not.“This fire is a classic case of what was wrong with fire investigation in the 1980s,’’ said Dr. Craig L. Beyler, a nationally renowned arson expert who reviewed the case files and evidentiary reports for the investigative reporting center.
  • No accelerant, bottle glass, or other physical evidence of the three Molotov cocktails was discovered at the scene. Modern fire experts view this as an extraordinary gap, which should have caused police to reconsider their theory.“If a fire is arson, it’s not a subtle crime,’’ said fire scientist John Lentini, retained as an expert witness by Rosario’s legal team. “Particularly a place that’s been set with three Molotov cocktails — you’re going to find this stuff, you’re going to find gasoline and you’re going to find Molotov cocktails — and they found none.’’
  • The translator who assisted in the police interrogation has made a dramatic reversal of the account he gave at trial. In a sworn affidavit provided to Rosario’s current attorneys, he says Rosario was delusional during the questioning and did not understand what he was signing.
  • Rosario’s original court-appointed trial counsel offered a flaccid defense, making no real effort to rebut the arson evidence. The lawyer was facing vehicular homicide charges in the same jurisdiction at the time of the trial — a circumstance that he admitted was a distraction from his courtroom duties and that legal ethicists today say should have forced his withdrawal.

Two Boston attorneys, Andrea Petersen and Esther Horwich, have taken on Rosario’s case and plan to file a motion for a new trial this summer to reverse what they say is a miscarriage of justice. Rosario remains adamant about his innocence: “I know in my mind, in my soul, I was never capable to do something like that,’’ he said in a recent prison interview.
But the lead Lowell investigator, Harold Waterhouse, has no doubt that the fire was arson and that Rosario was the culprit.

“I don’t care about the new questions that have surfaced; I don’t care about these experts,’’ Waterhouse, who retired in 1989, said in an interview. “I’ll go to my grave with Victor being guilty.’’

A suspect trail of evidence

The deadly blaze lit up a narrow block of the Acre, Lowell’s low-income Hispanic quarter.

The three-decker on Decatur Street had no smoke alarms and flames raged at the front and back exits. The two young families inside — Adelaida Ferrer and her three sons; Nancy Velasquez, Efrain Cortes, and their two sons — had little chance.

The Lowell Arson Squad rolled up to the fire scene at 2:30 a.m. Friday, snowflakes whirling around the still smoldering building. Given the speed of the inferno, Waterhouse and his partner, Lieutenant William Gilligan, immediately suspected foul play.

When they entered the first-floor apartment, they zeroed in on burn patterns that, to them, were obvious signs of a fire deliberately set.

“My investigation showed that there was two separate fires set,’’ Waterhouse told the grand jury. “One in the front hallway and one in the first-floor kitchen, front apartment, by the use of a flammable fluid.’’

The proceedings then moved quickly, according to police records and trial testimony.

The next day, Saturday, Waterhouse found Victor Rosario’s name in a report from the Red Cross, whose medics had treated his cut hand. That evening, Waterhouse interviewed a neighbor who said he saw a Hispanic man outside the home before the fire. By midnight, Rosario was at the station facing questions. By dawn, he had signed a confession saying he and two friends had burned down the house using Molotov cocktails made from “12 oz. bottles [Millers], rags, gasoline, and some other liquid.’’

But the confession was not supported by the physical evidence police retrieved from the fire scene.

State Police tests found no traces of gasoline or accelerant of any kind. Nor did FBI tests on broken glass from the scene reveal the presence of Miller beer bottles — which meant the shards were just as likely from broken glassware or windows. Molotov cocktail fires almost invariably leave such evidence.

“It’s hard to break a beer bottle; the neck almost never breaks because it’s small and compact, and the bottom is usually in one piece,’’ said Lentini. “If they were there, they would’ve found them.’’

Also, the FBI lab could not find Rosario’s fingerprints on any of the gas cans and materials police said were used to make the firebombs. Jurors at Rosario’s trial were not informed of the FBI lab results. And the prosecution hammered home its case for Molotov cocktail arson with only limited protests from Rosario’s lawyer.

At trial, Waterhouse and others testified the burning and char patterns at the scene were clear proof of arson: low burning on walls; burning between cracks in the floor; burning on the underside of a door; certain patterns in the charring; the fire’s speed; and two points of origin.

But arson experts who recently reexamined the trial record say investigators’ conclusions were driven by assumptions about fire evidence that are now considered little more than investigative folklore, or “junk science.’’ Almost all the evidence cited by police, the experts said, was also consistent with a fire born of accidental cause.

“None of the things that they pointed to are indicators of arson,’’ Beyler said. “The floor damage and the low burn stuff, as a generic proposition, you’re going to get in this fire no matter how you started it.’’

Similarly, the current National Fire Protection Association 921 manual, the bible for all fire investigations, indicates that the burn patterns Waterhouse and other investigators pointed to in 1982 can be found in both arson and accidental fires.

Because investigators were so confident in their analysis, they never seriously considered possible accidental causes, such as a space heater centrally located between the three rooms with heaviest burning.

When the heater was raised at trial, Waterhouse admitted he hadn’t noticed it was there.

“I don’t remember if there’s a heating unit in the living room or not,’’ Waterhouse testified.

Beyler underscored the heater as a possible fire cause: “From the physical evidence, it’s more likely than arson,’’ he said. “There’s no way you can exclude that heating unit.’’

Waterhouse, now 79, said he was unimpressed with any second-guessing of his work by experts. “I was an expert in my day, and my conclusion was that it was arson, and I will never change my mind.’’

Long night at the station

Waterhouse and Rosario were waiting in the office of the arson squad when Lieutenant Gilligan walked in with a translator, Ramon Nieves, 30, an Acre resident who had previously assisted Lowell police.

It was 11 p.m. and the foursome had a long night ahead. Rosario insisted he was innocent, saying he was next door buying drugs when he smelled the smoke and heard the cries for help. Three hours later, however, he signed a statement typed by Waterhouse, saying he acted as a lookout while a friend threw a Molotov cocktail.

The interrogation continued and, at 5 a.m., he signed a second confession that went further — that he and two friends firebombed the house. (The friends, brothers Felix and Edgardo Garcia, were not indicted. Investigators could never develop evidence against them beyond Rosario’s signed statement.)

During the more than five-hour police interrogation, Rosario was distraught and became hysterical — falling to his knees at one point, hugging the legs of a chair, weeping, wailing, and crying for God, according to police records and trial testimony. His mental state in making his confession would become a major issue at trial.

The prosecution argued Rosario was “calm and coherent’’ when he confessed and then, overcome with guilt, became agitated.

“He was somebody who was reacting in a normal, human way to terrible acts that he had done,’’ Thomas M. Brennan, the Middlesex County prosecutor, told jurors during his closing. “Emotionally upset, to be sure. Crying and guilt-ridden, to be sure. But not mentally ill.’’

That description was echoed at trial by Nieves, who testified that while Rosario may have appeared “very nervous’’ he understood Waterhouse’s questions and, most importantly, was calm and coherent when he signed the confession.

Nieves has recanted that account.

“Looking back on his behavior, it seemed as if he were hallucinating,’’ Nieves said in a sworn affidavit. “It was clear he was not in his right mind.’’

Nieves, who now lives in Puerto Rico, said Rosario was “incoherent’’ by the time he signed the second confession early Sunday and that “often he did not respond to any questions.’’ Throughout the police questioning, Nieves said, Rosario was speaking about the devil and of being the son of God.

“He kept rubbing his arms as if something like ants were on him. He kept saying ‘Get off me!’ ’’

Nieves, through Rosario’s attorneys, declined requests for an interview, and his affidavit offered little explanation for his change of heart about the testimony he gave at trial.

Certainly his job situation was in flux at the time. Nieves was working part time with a Latin American community group when he first served as Waterhouse’s translator in March 1982. Then, in the months leading up to the trial in 1983, things took a turn for the better.

Nieves applied during the fall to be a correctional officer at the Billerica House of Correction. He was hired six weeks later. On his job application, Nieves cited two law enforcement references: Inspectors Waterhouse and Gilligan, according to a source in the Middlesex sheriff’s office.

Nieves’s description of Rosario’s behavior during the questioning suggests to Dr. Judith G. Edersheim, a forensic psychiatrist hired by Rosario’s current attorneys, that Rosario may have been suffering from alcohol withdrawal — delirium tremens or the “DT’s’’ — during the interrogation.

Edersheim, in a recent interview, said that while reviewing Rosario’s medical records she found a long overlooked 1983 test showing Rosario had “preexisting liver damage’’ — a common indicator of serious long-term alcohol abuse — when he began serving his life sentence.

The possibility of alcohol withdrawal is also raised, she said, by the fact that Rosario had gone 48 hours without a drink as his questioning began. In that time, Edersheim said, Rosario’s symptoms descended “along an almost perfect timeline for delirium tremens.’’ By the time he was questioned, he was “suffering from profound mental dysfunction,’’ she said in an affidavit.

“A person in florid DTs,’’ she added in an interview, “you could put the Magna Carta in front of them and they would sign it.’’

Dr. Alison Fife, a forensic psychiatrist who teaches at the Harvard and Tufts medical schools, reviewed Rosario’s medical and psychiatric records for the investigative reporting center and said it was plain to her that the interrogation was out of bounds.

“This interview should have been stopped.’’ Rosario, accused of murder, was “saying things that sound, in lay language, crazy. . . . He’s out of control.’’

His murder conviction, she said, “deserves another look.’’

The prosecution’s psychiatrist, Dr. Martin Kelly, stands by his 1983 assessment at trial that Rosario did not have a “psychiatric condition.’’

But, in a interview, he said he is troubled by the lack of physical evidence of Molotov cocktails. If there were none, he said, how could Rosario confess to throwing one?

Witness to an inferno

Edward Evans lived in the Acre, and he was heading home as 1 a.m. approached on March 5, 1982, after a night out drinking beer.

As he walked down Decatur Street past a three-decker, he testified later, he noticed three men and heard a living room window break. Evans said he exchanged glances with the man nearest the window — whose left arm was raised, his hand empty — and then hurried home.

The next thing he saw was an inferno.

Police believed the man Evans saw was Rosario. But Evans twice failed to pick out Rosario from police photo arrays of Hispanic men. Only after Evans saw a picture of Rosario on the front page of the local newspaper, under the headline, “Three Arraigned in Fatal Fire. Drugs, Revenge Cited,’’ did he call police to say Rosario was the one.

There were errors from the start in how this evidence was handled. Police said Evans told them he saw a man with his right hand up — Rosario is right-handed. Waterhouse later testified that Evans said he saw the man hurl something into the building. Evans saw no such thing.

Both errors were eventually brought to light and corrected at trial.

“The man’s left arm was raised, not the right arm,’’ Evans,  who died in 2001, testified at an evidentiary hearing. “I did not see anybody throw anything through the window.’’

But other discrepancies went undetected and were not addressed at the trial.

The day after Rosario signed a confession, Lowell police sought a warrant to search the fire scene for a “metal container suitable for storing gasoline.’’ To convince a judge they had good cause for the warrant, the police affidavit said, “This is the same type of container that Victor Rosario referred to in the signed statement.’’

But the statement Rosario signed makes no mention of a “metal container.’’ Also anomalous was a search warrant application in which Waterhouse’s partner, Gilligan, said Rosario had provided extremely specific details about materials used in the alleged arson, including another liquid that could have been used as an accelerant, Red Devil paint thinner.

Yet none of those incriminating details was included in the written confession, and Waterhouse, in an interview, said he doesn’t know where the information came from.

Rosario “never told me that,’’ he said.

Gilligan died in 2005.

The third and perhaps strangest element of the investigation was the handling of a statement from an alleged eyewitness. According to a police report prepared four days after Rosario’s arrest, a Decatur Street neighbor said she saw three men, including one of the brothers named in Rosario’s confession, throwing beer bottles into the fiery building.

But this woman’s story never reached the ears of the jury: Elisa Quinones was never called to testify and her account was never mentioned in court, despite an internal memo from the prosecutor asking police to subpoena her.

The reason may have gone to the reliability of her statement.

Quinones has since renounced the police report. “I never saw three men throwing bottles,’’ she said in a 2004 sworn affidavit prepared for Rosario’s attorneys. “I do not know why the ‘Lowell Police Department Document’ is so incorrect, but I can swear that it is false.’’

John Guilfoyle, the Lowell officer who wrote the report, said in a recent interview that Quinones’ statement was accurately presented. He said prosecutors decided not to use her as a witness in the trial, but he would not say why and referred questions to the prosecutor.

The prosecutor, Brennan, declined to discuss the case. Quinones could not be reached.

A compromised defense

An alert defense attorney might have caught the inconsistencies in the fire scene analysis and more vigorously challenged the quality of the investigation and the search warrants.

But Rosario’s public defender, John R. Campbell, mounted a weak defense, dwelling on immaterial angles and sitting quietly on important ones, according to a review of the trial transcript and case materials.

One doctor’s report admitted into evidence noted, without citation, that Rosario “was involved in a legal situation when he was in New York because of robbing an automobile.’’

Relying on that, prosecutor Brennan portrayed Rosario as a convict and, when Rosario denied the charge on the witness stand, a liar. In his closing argument, Brennan upped the ante, referring to the prior record as an “armed robbery.’’

But Rosario had no criminal record in New York, according to state records examined for the investigative reporting center. Campbell never protested the false and prejudicial assertion.

When it came to challenging the prosecution’s arson case, Campbell also foundered. Most fire scientists in 1983 would have been able to testify that a Molotov cocktail almost always left trace evidence behind — accelerant, a wick, or thick bottle glass — experts interviewed for this story said.

But Campbell never called an expert witness for the defense to rebut the police account.

John Campbell had a lot on his mind before Rosario’s murder trial. On Oct. 7, 1982, Campbell, an admitted alcoholic, was driving drunk when he hit an elderly couple in Stoneham. Campbell survived but the pedestrians were dead at the scene. The next day he was charged with two counts of vehicular homicide by the Middlesex district attorney’s office, the same office he was soon set to square off against in Rosario’s case.

Legal ethics experts interviewed for this story say Campbell faced a conflict that should have forced him to the sidelines.

“This guy knows he is facing charges and now has some interest in keeping the prosecutors happy in some fashion, or at least it might be perceived as such,’’ Boston College law professor Paul Tremblay said.

Under state law, Campbell was required to disclose his troubles and obtain written consent to continue as Rosario’s lawyer— and to do it as soon as possible. But Campbell waited more than four months to ask Rosario to sign a waiver.

“That seems to be a problem,’’ Tremblay added, “because obviously it’s easier to get a waiver from your client if it’s three weeks from when the trial’s going to start.’’

Indeed, Rosario said he signed a waiver because he felt he had no other choice. Campbell, he said, was his only ray of hope in a legal world he did not understand.

“I started looking at him like he was my father.’’

After Rosario’s conviction, Campbell, who died in 2005, conceded his indictment weighed on him. In an affidavit prepared as part of an unsuccessful appeal by Rosario, he wrote, “I may have been distracted by my legal situations.’’

Years without freedom

Since the mid-March interrogation in 1982, Rosario, 52, has not felt freedom; he has now spent more time in jail than out.

State courts have twice denied his appeals, and the state Parole Board twice denied parole, citing Rosario’s insistence on his innocence.

“Mr. Rosario takes no responsibility for killing 8 people. He is an unrepentant killer with no remorse,’’ the board wrote in its unanimous 2002 decision.

Brennan, the prosecutor who tried the case, is now a district court judge. He declined to comment on the conduct of the trial, calling it inappropriate “for a sitting judge to comment on a case, on its validity or invalidity.’’

But he added, “If he [Rosario] has reasons to file a motion for a new trial then he should file one.’’

The office of the current Middlesex County district attorney issued a statement defending Rosario’s trial and conviction. But the district attorney at the time of Rosario’s conviction, L. Scott Harshbarger, said that a review of the case may be in order.

“We always strive for justice,’’ said Harshbarger, who later served as state attorney general and is now in private practice. “This new information may warrant a second look at the conviction, but that is a matter that must be determined, on the record taken as a whole, by the district attorney and/or the court.’’

Rosario, meanwhile, waits for his latest appeal to begin. In prison, he runs 76-lap marathons, leads Bible study classes, has married a prison educator who taught him to read, and expects to be ordained July 9.

Tremont Temple Baptist Church minister Eleftheria Sidiropoulou, who is leading his ordination, said it will be the first time in her 30 years that she has seen a prisoner ordained.

“He’s not only a decent man,’’ she said, “he’s a man of his word.’’

Today, the house on Decatur Street is gone, a dirt lot in its place. No one has ever built on the ground where eight people died and a young man was blamed.

This story was reported for the New England Center for Investigative Reporting at Boston University, where Lehr is a journalism professor and Nicas is a journalism major. Other contributors are graduate journalism students Jennifer Judson, Roxanne Palmer, Solomon Syed, Rachel Blumenthal, Sarah Favot and high school intern Akshat Pandey. Jack Nicas can be reached at  jnicas@globe.com . Read the story in the Boston Globe.