For welder William Arnett Jr., 41, death came in the form of a pneumonia-like condition that confounded doctors with its resistance to treatment.

For refinery dockworker Jesus Frias, 45, the killer was aplastic anemia, a rare and painful blood disease.

And it was heart failure that took the life of Lenzie Butler, a 49-year-old refinery operator.

It's entirely possible, despite their proximity to hazardous substances in the Houston-area petrochemical complex, that none of these men died of work-related causes. Certainly, none of their employers -- Belmont Constructors (Arnett), Lyondell-Citgo (Frias) and Shell Oil (Butler) -- reported the deaths to the Occupational Safety and Health Administration, as they must do when there is even a suspicion of work-relatedness.

But questions about these untimely deaths are being raised by the men's families and former coworkers:

Why did Arnett -- an exceptionally healthy man, according to his wife -- succumb Oct. 29 to a strange respiratory ailment he should have been able to fight off? Could it have had anything to
do with his presence in the Rohm and Haas Texas chemical plant on Oct. 15, when a still-unidentified agent sent 32 workers to the hospital with complaints of nausea, dizziness and breathing difficulties?

How could Lyondell-Citgo not have reported Frias' death as possibly work-related, knowing that he worked around the solvent benzene for 20 years and having access to medical literature
linking benzene exposure to aplastic anemia? OSHA's Houston South Area Office is investigating a complaint against the company, Area Director Ray Skinner confirmed last week, although he would not elaborate.

Is it a coincidence that Butler died on Jan. 28, 1993, three days after a large cloud of vaporized oleum -- fuming sulfuric acid -- drifted onto Shell property from the neighboring Lubrizol chemical plant?

OSHA rules require employers to report, within eight hours, any death that is or may be work-related, or any accident that causes three or more workers to be admitted to a hospital.

Bob Whitmore, chief of OSHA's Division of Record Keeping Requirements in Washington, said the agency expects employers to "err on the side of reporting" whenever a questionable death
occurs. He cited this recent case as an example: A worker in the steel industry with a preexisting heart condition was lowered into a pit for 15 minutes; when he was pulled up, he was dead.

"The company argued this clearly was not work-related because the guy had a bad heart," Whitmore said, but the death should have been reported because the victim's work environment may have caused or contributed to it.

Skinner said an employer "cannot make an arbitrary and capricious decision" about whether to report a fatality to OSHA, which uses the data to identify hazardous companies and industries.

But some employers are reluctant to report any but the most glaring fatal accidents -- crushings, falls, electrocutions -- because of the many repercussions such fatalities bring. Apart from OSHA investigations, they often draw bad publicity, lawsuits and scrutiny from insurance companies. They make workers uneasy and blemish managers' safety records, possibly jeopardizing this year's bonus or next year's promotion.

Sometimes even obvious cases go unreported. In February, OSHA's Lubbock Area Office cited the Copan Corp., a small oil field service company in the Panhandle, for failing to report two deaths -- one by heart attack on July 3, 1990, and the other by asphyxia on Nov. 5, 1993. The company did not contest the citation and paid a $4,500 fine.

Erratic reporting is one reason the federal government's occupational health and safety database is inadequate and preventive efforts have had limited success, especially in industries such as construction.

In a 1987 study, a National Academy of Sciences panel noted: "There is no single agreed-upon estimate of the number of occupational fatalities in the United States." The panel found that in 1984, the estimated number of work-related deaths ranged from 3,740 (U.S. Bureau of Labor Statistics) to 11,700 (National Safety Council).

"The panel found it rather startling than an agreed-upon method has not been devised to estimate a phenomenon as basic as traumatic death in the workplace," the study says. Those who have followed the issue since 1987 say there is no reason to believe things have improved markedly.

Without good data on workplace deaths, injuries and illnesses, "everybody's kind of shooting in the dark," said Stephen Newell, director of OSHA's Office of Statistics. "How are you going to prevent something unless you know it exists?"

Although OSHA has taken no action against the employers of William Arnett, Jesus Frias and Lenzie Butler, their former coworkers and survivors believe the deaths were work-related.

Arnett, an ironworker and welder with 20 years' experience, had been with Belmont Constructors only three months when he got sick at the Rohm and Haas plant in Deer Park on Oct. 15.

He came home that evening complaining of chills and "shaking real bad," said his wife of 22 years, Belinda. Although he felt very ill, he went to work the next three days because the plant was in a maintenance shutdown, and work was behind schedule, his wife said. "My husband was always labeled a hard worker. He always did more than he had to."

Finally, on the evening of Oct. 18, Arnett admitted he was too sick to work. "At this point he was scared," his wife said. "He had chills, high fever, he started coughing real bad."

Arnett saw a doctor on Oct. 19 and was told he probably had pneumonia and should get to a hospital immediately. His wife took him -- "shaking violently" -- to the Ben Taub emergency room, and by the next morning he was on life support. He never regained consciousness and died Oct. 29.

The doctors at Ben Taub at first diagnosed pneumonia, Belinda Arnett said. "As time went on they were stumped. They couldn't understand why he wasn't responding to anything because he was so young and healthy. This was the first time he'd ever been in the hospital. He'd never had the flu."

She authorized an autopsy, the results of which should be available soon.

Belmont officials did not return telephone calls from the Chronicle. After an investigation of the Oct. 15 incident at its plant, Rohm and Haas reported Dec. 5 that "there is no objective evidence for toxic exposure" among any of the sickened workers to either cyanide or ammonia, chemicals that were initially suspect.

But Belinda Arnett and her attorney, Jerry Swonke, said they will continue to investigate William Arnett's death.

"It was so unexpected," Belinda Arnett said.

In the case of Jesus "Jesse" Frias, the cause of death is no mystery. He died Aug. 9 of aplastic anemia, a rare ailment in which the red, white and platelet cells in the blood are reduced. He had worked for 20 years as an operator on the docks at the Lyondell-Citgo refinery, helping load and unload benzene-containing products.

What is a mystery is why, given the connection between aplastic anemia and benzene, Lyondell-Citgo didn't report Frias' death to OSHA. The American Medical Association's Encyclopedia of
Medicine, for example, says that while the disease sometimes develops for "no known reason," long-term exposure to benzene fumes "has been implicated as a cause."

Lyondell-Citgo spokesman David Harpole said, "The great majority of blood-related illnesses are not related to chemical exposures, and we have no information to tie (Frias') illness to a work-related exposure."

Frias' wife, Virginia, said she had never heard of aplastic anemia until her husband was diagnosed with it last spring. The first clue that something was amiss came when Jesse Frias bled excessively after having a tooth pulled.

"The next thing I know he's in the cancer center at Bayshore Hospital," Virginia Frias said. "One day you're OK, and the next day they're telling you you're gonna die."

Frias' death -- along with the premature deaths of at least six other Lyondell-Citgo dockworkers in the past decade, two additional cancer cases on the docks, and a string of cancers in the refinery's "coking" area -- prompted a union workers' committee to ask the company for an independent health survey. The company refused.

"When I see nine people (on the docks) in 10 years come up with these kinds of afflictions, then the red flag is up," said committee chairman David Taylor, a member of Oil, Chemical and Atomic Workers Local 4-227. "I'm saying (to Lyondell-Citgo), "Maybe you're missing something."'

The union, which represents nearly 800 refinery workers, has the right to seek a third-party health investigation under terms of its contract, Taylor said. But Lyondell-Citgo spokesman Harpole said the company has no health data "that would indicate any reason for us to bring in a third party."

Virginia Frias has sued Lyondell-Citgo, Arco (part-owner of the refinery) and suppliers of the benzene-containing products, alleging they "knew for decades the dangers associated with the chemical benzene" but failed to warn Jesse Frias and other dockworkers and failed to give them protective equipment.

The suit claims the defendants "entered into a conspiracy" by withholding medical information from workers, deleting damning material from scientific studies and altering or terminating studies "that would prove unfavorable to them."

Virginia Frias' attorney, Catherine Baen, said that Jesse Frias' death was caused by benzene exposure and should have been reported to OSHA.

"Even the most conservative hematologists and oncologists will agree that benzene is a cause of aplastic anemia," Baen said.

For several years, workers at Shell's Deer Park refinery have complained about chemical releases from Lubrizol, Shell's neighbor to the east. The releases have become so commonplace that Shell, in company memoranda, typically lists its plant injury and illness rates "excluding Lubrizol events."

Some of the most alarming incidents have involved oleum, a liquid that can cause fatal respiratory disease in large doses and can cause eye and skin burns and nose, throat and lung injury in
lesser amounts.

Lenzie Butler was working as an operator the morning of Jan. 25, 1993, when a cloud of vaporized oleum -- created when a Lubrizol tank ruptured and plant workers doused it with water -- wafted onto Shell property. Butler died three days later, and Shell moved quickly to quell rumors that chemical exposure had precipitated his heart failure.

"This employee was not, repeat was not, exposed to oleum," said one company communication obtained by the Chronicle. "We at first had a statement or rumor that he was, but that is not the case."

In a prepared statement, Deer Park refinery manager Steve Reeves said that "based on our knowledge of (Butler's) medical history and the coroner's death certificate furnished to Shell Oil Co., it is clear to us that this particular employee did not suffer a work-related death. As such, the death does not qualify to be reported to OSHA as a work-related death."

Butler's survivors and 12 Shell workers have sued Lubrizol, however, alleging that carelessness with oleum contributed to Butler's death and gave the others lingering respiratory problems and other ailments. Plaintiff's attorney Keith Hyde of Beaumont declined to comment on the suit, and Butler's relatives could not be reached for comment.

David Reel, general manager of Lubrizol's three Houston-area plants, said he does not believe the company was negligent or that the oleum cloud had any major effect on the Shell workers.

"We had an incident that was unexpected, and we had safeguards in place that should have kept that from occurring," Reel said. Four Lubrizol officials testified in depositions last month, however, that the oleum tank that ruptured did not have a pressure-relief valve. It does now.

Reel said Lubrizol has "done a great deal" since January 1993 to prevent off-site chemical releases and has largely succeeded, although "we've had some nuisance odors that have been noticed by some of our Shell neighbors" since that time.