ROCKVILLE -- Back in college, Gary Ellis worked summers as an umpire at Little League baseball games, explaining the rules to puzzled kids and furious parents. Today, he is chief of the federal office charged with protecting people who serve as the subjects of scientific experiments.

Umpiring has a lot in common with his current job, he's found. "The similarities extend to being cursed at and disliked by most, while believing one is doing an important thing," he says.

But Ellis is in the big leagues now. This month, he shut down 2,000 studies at Duke University Medical Center for several days, saying the center's board that oversees research failed to guarantee the safety of its subjects. The move, which drew national attention, followed similar suspensions of research hospitals in Los Angeles and Chicago in the past six months. For those institutions, it was like being thrown out of the game.

Today, academic officials across the country are anxiously re-examining their policies. Scientists experimenting on people are checking their consent forms. Researchers who might have cut corners, some say, may now think twice.

"The Duke story has had a very beneficial effect outside of Duke," says David J. Rothman, professor of social medicine at Columbia College of Physicians and Surgeons.

In the name of science, U.S. researchers have sometimes engaged in shameful acts. Decades ago, they injected children with plutonium and hepatitis, and put cancer cells in elderly men. In the infamous Tuskegee study, physicians watched and did next to nothing as 399 African-American men were infected with syphilis for 40 years. More than 120 died of the disease.

More recently, researchers have tested new drugs and procedures on unconscious trauma victims, then sought their "consent" after the fact. And they have deliberately withheld anti-psychotic medications from schizophrenic patients to see how quickly they became sick again.

Staff stretched thin

Ellis is responsible for preventing the outrages of 25 years ago from happening again. He and his tiny staff -- four investigators, two of them half-time -- police studies at 4,000 institutions supported by the National Institutes of Health, which spends more than $15 billion a year on medical research.

"He's given a pittance," says Rothman. "Four people to govern all of human experimentation in this country? The notion is absurd."

With his limited budget, Ellis can't do in-depth investigations. Instead, his staff is restricted to reviewing forms and booklets, protocols and procedures, to make sure they conform to federal regulations. In other words, they enforce the rules.

Scientists, it seems, can have a hard time following those rules. Each year, a score of institutions ask if Ellis will waive the requirement that they review ongoing human experiments once a year. His answer is always no.

"The regulations say at least annually," he says. "This is neither hard to interpret, nor hard to comply with, except for the calendrically impaired."

Like any good umpire, Ellis is despised by partisans on both sides. Some scientists accuse Ellis and his Office of Protection from Research Risks of nit-picking, wasting time and money. "They're rigid and inflexible on matters that, in my view, have almost nothing to do with the protection of human subjects," says one researcher, who says he fears retribution if his name were used.

Some patient-rights advocates, on the other hand, grumble that Ellis is too willing to cajole and conciliate -- and too reluctant to impose sanctions. They point out that Ellis and his staff have negotiated for years with some negligent institutions.

"They did not enforce those rules in prior years," says Adil E. Shamoo of the University of Maryland's School of Medicine, editor of a journal about ethics in research. Among administrators, he says, that record "has generated a lack of concern and maybe even cavalier attitudes."

Ellis will say only that he has to balance the need for medical research with protection of research subjects, and the best way to do that is to make sure that everyone abides by the letter of the law.

The boyish biologist-turned-bureaucrat has always sweated the details. Growing up in Cleveland, he read the dictionary cover to cover. He was a very obedient child. Told by his mother not to eat between meals, he still won't nibble on doughnuts that co-workers bring to the office. Asked to pose for a photo recently with chin in hand, Ellis, 44, refused: his mother had told him never to put his hand on his face because it was unattractive.

Mona Lisa tie

After earning a doctorate in biology, he launched his research career at the University of Texas at Austin. There he met his wife, Mona, now a family physician. He honors her some days by wearing a tie with pictures of the Mona Lisa. (When he meets with animal researchers, he chooses among his collection of ties decorated with the classes of vertebrates: mammals, reptiles, amphibians, birds and fishes. When umpiring is on his mind, of course, he's got one with baseballs.)

Ellis plans everything carefully. On trips with his wife and 2 1/2-year-old son, he even schedules the bathroom stops. In interviews, he dictates his answers, correcting his thoughts mid-sentence and commenting on his own cleverness ("That's a good statement!" he exclaims.) Typically, he recites his office's regulations off the top of his head, occasionally whipping out the rule book to confirm he got it exactly right.

Ellis abandoned his lab career in 1983 after he developed an allergy to rats -- a common occurrence among scientists who study them. After working for Congress as a science analyst, he moved to the National Academy of Sciences.

Mustard gas tests

There, he supervised an investigation of the Navy's so-called "man-break tests." During World War II, thousands of U.S. servicemen were sealed into chambers and exposed to mustard gas, some until they were "broken," or overcome by the toxic fumes. The men were not told that they would be deliberately exposed to chemical weapons, and were sworn to secrecy.

Some test volunteers developed chronic skin diseases. But they kept silent for the next 40 years, until 1991, when news accounts of the tests surfaced. Twenty-two veterans agreed to come forward at an NAS hearing. "The testimony had a deep effect on me," Ellis says. "It was a horrible story. And true."

Ellis moved to the Office of Protection from Research Risks in 1993, determined to make the office a more forceful voice for test subjects. He quickly ordered a halt to experiments of emergency treatments on people who were unconscious.

A few months after that, a New Mexico newspaper revealed that scientists with World War II's Manhattan Project had injected people, without their knowledge, with plutonium and other radioactive materials. The news triggered a flood of unrelated allegations to Ellis' office.

Typically, the office only conducts on-site investigations at institutions with pending complaints. In the fall, he decided to audit a research center with no alleged violations, as a test of general compliance.

Meeting in his office, he asked his staff about the top 15 recipients of federal research dollars. (The Johns Hopkins University's medical school is at the top of the list.) At 14 of those places, Ellis' staff had regular contact with researchers in the past several years.

Singling out Duke

The lone exception was Duke University Medical Center, with $138 million a year in federal research funds. Ellis wondered: Why the silence? So in December, he and his investigators arrived in North Carolina and spent two days poring over records.

"The deficiencies at Duke were broad and deep," Ellis says. Duke's research review board, they found, had approved projects without knowing enough about them, failed to keep track of people enrolled in studies and didn't monitor serious side effects of treatments.

Ellis' office called for changes, but the medical center moved slowly. "Their nonresponsiveness was remarkable," Ellis says. Finally, on May 10, the government ordered research halted in all but those cases where it would have harmed patients.

A spokesman for the Pharmaceutical Research and Manufacturers Association of America says Ellis' office could have gotten the same results without disrupting research. Dr. Edward Holmes, dean of the medical school, declined to be interviewed. But he said in a statement that he was "grateful" to Ellis' office. "We were very pleased with the collegial and constructive nature of our meeting with them last week, and that our plan was responsive to their concerns."

Schizophrenia research

Acting on complaints, Ellis' office investigated a Catonsville psychiatric research center run by the University of Maryland School of Medicine last year. Researchers there have used drugs to induce the symptoms of schizophrenia, or have taken patients off medications to study relapse. Ellis' staff directed the center to explain the consequences of a relapse in more explicit terms. They ordered the hospital to make it clearer to volunteers that there was little chance the studies would benefit them.

"There are a lot of little rules and regulations about dotting the 'i's and crossing the 't's," says Dean Donald E. Wilson. "But I don't think they [regulators] are unnecessarily aggressive, because the stakes are very high."

Ellis says that researchers often overstate the benefits of medical experiments and understate the risks. His biggest concern, though, is the research over which the federal government has no oversight. Many studies are exempt from federal regulation, including some done at colleges that do not receive federal research grants or at private doctor's offices.

Ellis was shocked a few years ago by one of these maverick experiments. Two plastic surgeons decided to test differing techniques by operating jointly on 21 patients at a Manhattan hospital. Each doctor used a different technique on half of each patient's face. None of the patients was told of the experiment nor warned that they might end up with a mismatched face.

Even when making close calls, Ellis doesn't get rattled. On a wall in his office, he's taped a photo he snapped of alligators ominously circling a muddy pond, waiting for supper at a place called Gatorland. Whenever he starts to feel besieged, he says, he glances up at the picture.

"Things could be a lot worse," he says. "I could be at Gatorland, just prior to feeding time."