(FORTUNE Magazine) – In real life, what could be called Erin Brockovich moments occur very rarely. Lawsuits by injured victims are common enough. But they almost never make it to trial. Rarer still is the tingling moment when the little guy gets to confront a giant corporation with the evidence of its conduct. That's why a Santa Clara, Calif., courtroom was filled with spectators, reporters--and expectations--on an early November day. The attendees had gathered to see opening statements in a trial in which two retired IBM workers are suing the $83-billion-a-year company for knowingly allowing them to be poisoned, years ago, by toxic chemicals in the company's semiconductor and disk-manufacturing facilities. That poisoning, they claim, caused them to develop cancer.

As one would expect, plaintiffs lawyer Richard Alexander inveighed against his opponent. He portrayed IBM as aware of its employees' incipient illnesses decades ago, but lacking the decency to share that information with the workers. Displaying an enlarged photocopy of an IBM health questionnaire filled out by one of the plaintiffs, he charged, "This chart in 1976 confirms that IBM knew that James Moore reported complaints consistent with chemical poisoning!" Alexander ticked off a list of symptoms cited by Moore on the questionnaire. They ranged from headaches to blackout and hot flash--"all signs," Alexander asserted, "of systemic chemical poisoning."

Soon after, defense lawyer Robert Weber took his turn. His rebuttal was all the more devastating for the low-key manner in which he delivered it. Weber returned to Moore's health questionnaire and pointed out pages that Alexander hadn't mentioned. There, in giant magnifications of Moore's handwriting, were his explanations of the symptoms. The headaches? They occurred "2-3 times a year." The hot flash? That came "when I got the tetanus shot in 1956"--ten years before Moore started working for IBM. The blackout? It, too, had occurred years before Moore joined IBM. Nothing, in short, was what it had seemed on first glance. "So much for that," said the defense lawyer.

Weber proceeded through Moore's IBM medical history. It turned out to be a short journey. Far from a litany of contemporaneous evidence of chemical poisoning, Moore's records didn't whisper a word of such a condition. Indeed, they suggested he had enjoyed robust health during his 24 years at the company. The records mentioned only one relevant visit to an IBM doctor. That took place in 1967, when Moore complained of "profuse nasal discharge" after exposure to solvents--what the defense lawyer referred to as "a runny nose"--for which he was given a decongestant. Other than a few drop-ins for a bad back, Moore never returned to IBM's medical department again before he retired in 1993.

By the end of Weber's presentation, even IBM critics in the audience were wondering about the plaintiffs' case. John Roberts, a former IBM worker who had organized a courthouse vigil to protest the cancer deaths of compatriots, put it this way: "I feel like we're on the Titanic."

It's too early to say the plaintiffs' case has fatally collided with an iceberg--the trial is expected to stretch at least into January, and juries are notoriously unpredictable. Whichever way the verdict goes, the stakes are extremely high. IBM is defending 250 similar cases in California and New York, and as the first to go to trial, this one is viewed as a harbinger. A victory will generate crucial momentum for the winning side.

Moreover, beyond the fear that a big verdict could inspire suits against every other manufacturer in the $70 billion-a-year U.S. semiconductor industry--a sector that employs 255,000 people--the cases bear tremendous symbolic importance. Semiconductor manufacturing in the 1970s, '80s, and '90s played a role not unlike coal-mining decades ago: It embodied America's industrial prowess. At a time when manufacturing was on the decline, the sterile "clean room" in which chips were made symbolized the new frontier of high technology, the pride of American industry. And it still does.

All the heat and dust over the prospects of the lawsuits, however, should not obscure the more fundamental question: Are semiconductor workers getting cancer from chemicals in the workplace? The answer to that question is a mystery--and one that no lawsuit can resolve.

With his bow tie, round glasses, and tousled hair, Dr. Martin Sepulveda has the distracted air of an academic. Sepulveda, 52, carries the opaque title of vice president of global well-being services and health benefits at IBM. He is the company's chief doctor, in charge of the health and safety of its 315,000 workers. Diffident and soft-spoken, Sepulveda comes close to showing emotion only when asked about the courtroom portrayal of IBM as deceitful and callous. "This was the part that made me really indignant," he says, "the allegations that there was medical fraud perpetrated." Perhaps there's a personal element to it. Sepulveda began his IBM career in the mid-'80s as a physician in the company's East Fishkill, N.Y., semiconductor plant, the operation that generated the first suits against IBM. As he recalls it, the workers he treated never showed signs of anything beyond normal health problems.

Sepulveda is proud of what he considers IBM's "sterling" safety record. "In a company like this, we take the materials we use extraordinarily seriously," he says. "We're maniacal about it." To build in a margin of safety, he points out, IBM's policy has always been to limit exposure to any chemical to 25% of the most stringent requirement set by public health agencies. Workers are trained in handling chemicals, endure periodic written tests on their knowledge, and are examined for signs of health problems. (It's worth noting here that the issue of semiconductor health risks is more and more a historical one. The process has become progressively automated, increasingly protecting workers' health, albeit at the expense of their jobs.)

Sepulveda readily admits that clean rooms are filled with potentially dangerous and even deadly chemicals. From his vantage point, the issue isn't the presence of the substances, it's how the company manages and limits exposure to them. For example, semiconductor manufacturing can involve the use of arsine gas, which can kill with one whiff. But as long as the gas is used in a way that prevents the worker from breathing it, he would argue, there's no problem.

There has never been a groundswell of worker dissatisfaction with clean room safety, Sepulveda says. In an employee survey from 1988, produced in the current lawsuit, the No. 1 employee complaint was factory noise. Next in order of prevalence were worries about chemicals. "At times very strong odors enter our clean room environment," a worker wrote. "This causes us great discomfort--headaches, dizziness, upset stomachs, and so forth."

Strangely, though, the third-most common complaint seems to contradict the second. Many employees groused that IBM was too concerned with caution: "Many safety rules are overburdening and unnecessary," wrote a worker, in a typical comment.

As Sepulveda sees it, IBM has long been seeking answers to the questions raised about its workers' health. Three times in the past 15 years the company has funded academic studies. The first, in the mid-'90s, linked a group of chemicals known as ethylene-based glycol ethers to elevated miscarriage rates among women at two IBM semiconductor plants. The company phased out the chemicals. The second study found that "brain tumor risk was not associated strongly or consistently with employment" in IBM's electronics manufacturing.

The third study, still in progress, is tackling the big question: It's assessing the rate of cancer, and cancer deaths, at the three IBM plants that have generated the lawsuits against the company. The results, which are expected next year, will be peer reviewed and published. It's the sort of material that could potentially hurt IBM in litigation. But, Sepulveda says, "in my own mind, there was enough concern that had been raised by former employees"--in the form of lawsuits--to justify the investment. He vows, "We're going to accept the results we get, and we're going to do the right thing based on those results."

Yet paradoxically, scientific studies can raise as many questions as they answer. Within the reports funded by IBM are hints that the health risks are both more prevalent and harder to pin down than the company suggests. Critics have questioned, for example, how researchers could conclude there was no excess brain cancer among IBM electronics workers even as the study found that there was a threefold elevation in brain cancer among programmers with ten or more years of work at the company.

Others wondered how a single chemical culprit could be isolated as responsible for birth defects, as happened in the IBM-funded study on glycol ethers, when there were so many other toxins present. The study's lead author, Dr. Adolfo Correa, then of Johns Hopkins University and now at the federal Centers for Disease Control and Prevention, defends the findings. But a 1996 presentation by him suggests both how difficult it was to identify the glycol ethers as the sole factor--and what a chemically intensive environment he was studying: "We actually examined about 28 chemicals specifically in this plant: chemicals that were used in high volume, that had known prior reproductive toxicity, and for which there was potential for exposure."

The list of chemical dangers in clean rooms is alarmingly long. Dr. Myron Harrison, a former medical director of IBM, catalogued them in a 1992 article entitled "Semiconductor Manufacturing Hazards." The risks include "gas leaks," "oil mists from pumps and tools," and "vapors from wet chemical spills," just for starters. "Sporadically and unpredictably," Harrison wrote, "a noticeable amount of chemical vapor is expelled into the worker's breathing zone."

Sepulveda likens the article to an insert in a drug bottle--it chronicles, in minute detail, all the things that could go wrong. In real life, he says, the vast majority of those problems don't occur--precisely because IBM is aware of the risks and takes steps to prevent problems.

Still, the article raised fundamental questions about chipmaking. Harrison commented on the "rapid pace of change" in manufacturing processes and "the fact that adequate toxicologic assessment of chemicals almost never proceeds [sic] their introduction into manufacturing settings." Most alarming was this frank admission: "Any large semiconductor facility uses several thousand chemicals. An attempt to review the toxicology of all these materials is doomed to be superficial and of little value." For all the dangers he had laid out, the IBM doctor suggested, there were potentially as many more that nobody yet even knew about.

To be fair, this reflects how we, as a society, deal with chemicals. We use first, ask questions later. Few businesspeople would agonize over such a question: Are you going to wait for ten years of chemical testing before you introduce a process that will enable you to make a cutting-edge chip? Not waiting is often eminently reasonable from a business perspective--yet frightening from the perspective of public health. "There are 100,000 commercial chemicals in use in the United States," says Dr. Bruce Fowler, a University of Maryland professor who is on leave as a senior research advisor at the federal Agency for Toxic Substances and Disease Registry. "Of that number, perhaps 500--certainly less than 1,000--have ever really been studied."

Even more confounding is our ignorance of the health effects of exposure to multiple chemicals. Fowler says scientists have barely begun to ponder which combinations multiply each other's effects, which are neutral, and which cancel each other out. In most cases, they simply don't know.

If employee surveys suggest workers weren't rising up in revolt over chemicals, interviews today reveal that the clean room atmosphere was not as pristine as IBM portrays it. "On paper, all the t's are crossed and i's are dotted," says John Roberts, who worked for nearly 20 years at IBM's facility in San Jose (the plant in which the plaintiffs in the current trial worked). "But then when you get to the reality of the situation"--pressure to get the job done--"shortcuts get taken." Roberts and four other workers interviewed for this story recall routinely bending deep inside machinery to clean it with solvents.

Chemical odors are a leitmotif. "It used to be a common joke here that once you leave, you go outside and choke on the fresh air," says Lee Conrad, who worked for 26 years in an IBM fab in Endicott, N.Y., and is now national coordinator for Alliance@IBM, a 5,000-member union. "We had a lot of gallows humor. But even the clocks on the walls--the faces were pitted from the fumes."

Fear of chemicals put managers in an uncomfortable position, recalls Beth Diesner-Gee. Now 60, Diesner-Gee was an administrative manager in several IBM facilities between 1979 and 1991. (She is a plaintiff in one of the suits against IBM. Her husband, Lynn Diesner, died of brain cancer in 1990, an illness she charges was caused by exposure to chemicals and video-display-terminal radiation during his 24 years as an IBM programmer.)

Diesner-Gee says the secretaries she oversaw constantly fretted about the chemical odors, which wafted even into office areas at the plants. They worried about what seemed to be high rates of birth defects among children of IBM workers. She says IBM brass instructed her to give the "standard response"--a sentence she can still reel off today in a gravelly monotone, like an announcer racing through a list of legal disclaimers in a TV ad: "There has never been a proven case against IBM that any of its employees have ever suffered any adverse effects from chemical exposure or working with video display terminals."

So concerned was Diesner-Gee that she brought in some higher-ups to calm her staffers' fears. In a tense meeting, she says, one of the IBM executives fled in the face of angry questions from the secretaries. "This is bullshit," she recalls them saying. "You're not giving us any answers."

Diesner-Gee says she never did get a satisfying response. The last time the subject came up during her years there, it was in a much more personal context. In 1990, shortly after her husband's brain cancer was diagnosed, she took him in his wheelchair to the medical department at IBM's San Jose facility. As they chatted with an IBM doctor, her husband asked how many employees had been diagnosed with cancer and how many had died. The doctor told them that IBM didn't track that sort of information. When she pressed him, she says, the doctor repeated the company mantra: There has never been a proven case that the IBM work environment causes cancer or any related illness. Diesner-Gee found herself on the receiving end of the line she herself had used. Shattered, she wheeled her husband out. He would die two weeks later.

If the search for answers on the dangers of the clean room has become a quest for some people, then Dr. Joseph LaDou is that group's prophet. LaDou is director of the International Center for Occupational Medicine at the University of California at San Francisco. He has been studying and agitating on the subject for more than two decades. Possessed of an austere countenance, LaDou is quick to preemptively assure a reporter, as he welcomes him to his modern home in Silicon Valley, that the redwood used to build it was recycled. "No trees were cut down to make this house," he says. LaDou seems to carry the burdens of worker safety on his shoulders.

In 1984, when the New York Times published a front-page story that publicized the health risks of clean rooms--it reported a "wave of legal actions"--LaDou was quoted ruing the lack of research and asserting that the young industry was just getting old enough for its workers to begin showing elevated cancer rates. Nearly 20 years later, the wave of legal action has never quite materialized--250 lawsuits against IBM is a substantial number but not an epidemic considering that some 100,000 people have worked in company clean rooms over the years--and LaDou finds himself still saying many of the same things. He is frustrated.

LaDou is the leading expert on the health risks of clean rooms. He has conducted research and edited a textbook on occupational medicine. He has preserved his credibility by refusing to be a paid expert for either side. And along the way he has developed an antipathy to IBM. ("Do I sound jaded?" he asks. "I've been in this for an awful long time. I should watch myself.")

But for all that, LaDou doesn't allow his feelings to shade his science. He candidly acknowledges that the subject of semiconductor manufacturing risks remains a mystery. "What we know is very little," he says. The most he'll allow himself to say is, "We have reason for concern."

LaDou's criticism of IBM--indeed, of the entire semiconductor industry--centers on what he says is its unwillingness to honestly confront the health issues. He points to a notorious 1998 episode in which he helped negotiate a deal for an industrywide study of worker safety. The research would have been conducted by the California Department of Health Services with a supporting grant from the U.S. Environmental Protection Agency. But ultimately the major chip companies, including IBM and Intel, refused to cooperate with researchers. (Among other things, the companies cited concerns over the privacy of employee records.)

Five years later the industry's main trade group has taken only baby steps toward solving the mystery of the clean room. The Semiconductor Industry Association has just begun to study the feasibility of doing a study on the cancer risks of chipmaking. "You go as fast as the science allows you to go," cautions SIA president George Scalise.

LaDou is not placated by the industry's willingness to underwrite research; he questions the legitimacy of such funded studies. But he remains convinced that the truth will eventually emerge. "We know enough about the carcinogens in the workplace that a number of these cases are going to be settled in favor of the plaintiffs," he says. "They're not going away."

Lawsuits may indeed keep pressure on the industry. But they may also discourage companies from research. And even when data are available, the adversarial nature of the legal process tends to cloud the complex evidence that emerges. Dueling experts and dense subject matter can leave one with the unsettling feeling that Nothing Can Truly Be Known.

For starters, it's often impossible to prove how an occupational disease is caused. One reason asbestos became the tort of torts is that it's a rare exception to this rule. If a person has mesothelioma, it's virtually certain it was caused by asbestos.

In the Santa Clara cases, the two plaintiffs have been stricken with, respectively, breast cancer and non-Hodgkin's lymphoma. Some 250,000 Americans were diagnosed with one of those two diseases last year, and the list of potential causes is endless. Each plaintiff has many risk factors, any of which could be implicated. Plaintiff Alida Hernandez has at least five of the most common risk factors for breast cancer, including having had hormonal replacement therapy. And how does one determine that the non-Hodgkin's lymphoma of plaintiff Jim Moore was caused by workplace chemicals and not by the genetic fact that his mother had lymphoma--or the fact that he smoked a pack a day for years, or that he was exposed to crop-dusting chemicals when he worked on a farm for more than a decade? Could his disease have been caused by chemicals? Perhaps. But which ones?

The plaintiffs and IBM don't even agree on precisely which toxins the workers were exposed to. For all the dread substances mentioned, the two that came up most often in the early days of testimony were isopropyl alcohol and acetone. Those chemicals are hardly esoteric; many people encounter them as ingredients in rubbing alcohol and nail polish remover, respectively. Moore used isopropyl alcohol at work, he testified--but he also used it at home to soothe leg aches and clean cuts. And if you think that's murky and convoluted, wait until the lawyers trundle out their experts on airflow to argue about how the ventila-tion systems 25 years ago succeeded--or failed--to remove the chemicals that were in the air.

And so one is left to watch the plaintiffs' case--this stuff is bad--and the IBM defense--where's the proof?--pass each other by like ships in the night. Expressed in those terms, both sides are right. But trials aren't about broad issues. They're about specifics: Was this particular individual's specific disease caused by the actual chemicals they were exposed to?

For all of the broad concerns, the specifics seem to favor IBM. As the Santa Clara judge bluntly told the lawyers in a discussion in his chambers, much of the plaintiffs' opening statement was "basically a general smearing of IBM, but not related to our case."

Indeed, what a lot of this boils down to is the chasm between law and reality. In the California cases the law requires that the plaintiffs prove not only that their cancers were caused by chemicals, which is hard enough, but also that IBM knew at the time that those specific individuals had been poisoned by chemicals, and that IBM hid the information from the employee, and thereby caused the illness to be aggravated.

Why this extraordinary burden of proof? Because such cases would normally fall under workers' compensation laws as opposed to the tort system. Workers' comp can provide lost wages or medical treatment to employees who are injured or ill. But the problem, if you believe your company poisoned you, is that it doesn't offer the possibility of, say, money for emotional suffering or anything resembling a big punitive damages payout.

The California courts have left open only the narrow exception that allows plaintiffs to use the tort system for egregious cases in which the company knowingly harmed workers and kept them in the dark about it. As a result the plaintiffs have twisted their cases into the legal equivalent of modern sculpture, trying to fit the exception. It's that challenge that makes it likely that IBM will triumph in these cases--if not at trial, then on appeal. Indeed, the court dismissed two other cases against IBM before the Santa Clara trial started. As the judge put it, "the plaintiff's own experts were unable to opine, to a reasonable degree of medical certainty, that [plaintiff] Suzanne Rubio had systemic chemical poisoning."

Though the California cases are getting all the press, another group of cases filed in New York State could pose a greater risk to IBM--because they don't have to overcome the workers' compensation hurdle. There, some 50 suits have been filed on behalf of the children of IBM workers who claim that chemical exposures during or before pregnancies caused catastrophic birth defects. Because the children never worked for IBM, there's no need for legal contortions--they can sue IBM directly for hurting them. IBM can be found liable, for example, simply for being negligent.

These cases also include an element that can completely overwhelm the unknowability of the science in the minds of jurors: the emotional impact of seeing maimed children. And given that the kids were born with their defects, nobody can blame them on the children's smoking, drinking, or using nail polish remover. That may explain why IBM settled one of the birth-defects suits three years ago. The company vows to contest the remaining cases. The science is as murky when it comes to birth defects as it is with cancer, and some of the plaintiffs' cases rely on unproven theories that damage to male workers' sperm caused their children's birth defects. Still, the suits clearly present the greatest threat that IBM will be hammered in court.

In the end, we're left with two distinct universes. In the legal realm the plaintiffs can't prove the chemicals caused the terrible health effects. But in the real world--the world of human fear, employee safety, and public relations--IBM can't prove they didn't.

FEEDBACK nvarchaver@fortunemail.com