It's easy to get the impression from a cluster of recent episodes that shoddy ethical practices are plentiful and growing in medical research. But no one knows for sure, because the science establishment runs like a bank without auditors.

As a result, only the gross, unconcealable delinquencies come to public attention. And a bunch of them have lately. On May 10, the federal ethics cops shut down some 2,000 medical research projects at Duke University for four days because of noncompliance with regulations for experiments on humans. The Department of Veterans Affairs ordered a similar shutdown in March at the VA hospital in Los Angeles. Both shutdowns were virtually unprecedented, though the types of lapses that spawned them have been complained of for years.

Meanwhile, the alleged theft of an invaluable gene discovery is the basis of a nasty patent infringement suit that the University of California at San Francisco has brought against Genentech, a leading biotechnology firm. And it's no rarity to find academic and scientific publications with headlines such as "Professor Gets Jail Time for Lying to Win Grants" (Chronicle of Higher Education, Feb. 5); "Two Former Grad Students Sue Over Alleged Misuse of Ideas" (Science, April 23); and "Editors Call for Misconduct Watchdog" (Nature, June 12, 1998).

Science in America is self-governing. The chieftains of science prefer it that way, and diligently have resisted external surveillance as well as internal checks. Politics, baffled by science, but recognizing its importance, defers to them.

The system for ensuring propriety in the conduct of some $16 billion a year of government-financed medical research is a ramshackle pretense -- and that's just fine with the leaders of the establishment. In the Department of Health and Human Services, the Office of Research Integrity -- which is supposed to police scientific fraud, fabrication and plagiarism -- has been headed by an acting director since April 1996. The staff there is capable and industrious, but the absence of a full-fledged director for more than three years signals low caste in the federal bureaucracy. A proposal for a government-wide definition of scientific misconduct disappeared into the White House Office of Science and Technology Policy three years ago and has never been heard of again.

The shutdown of research at Duke was ordered by an understaffed, underfinanced federal unit with vast responsibilities, the Office of Protection from Research Risks (OPRR), deep in the National Institutes of Health. For ensuring compliance with federal research rules at nearly 4,000 universities and hospitals, OPRR has two full-time and two part-time investigators and an office staff of 30.

In nine years, OPRR investigators have visited 38 research institutions to check on compliance with rules to protect human experimental subjects. Basically, medical research employs an honor system to ensure that volunteers in experiments are honestly informed of potential risks and benefits. They are informed by scientists eager to get on with their experiments.

A government research manager familiar with the OPRR likens it to the internal affairs units that monitor police behavior: They provide credibility, he said, without seriously interfering with wayward practices. "If OPRR did its job too well," he added, "it would not be good public relations for NIH."

In each university and hospital, the protection for people enrolled in experiments is entrusted to local institutional review boards. Studies conducted by the General Accounting Office and the inspector general of the Department of Health and Human Services found that many of the boards are sham operations, providing rubber-stamp approval for experiments -- sometimes in hurried sessions that speed through piles of applications. In some cases, scientists seeking approval of their own experiments have participated as members of the review boards. With rare exception, no money or staff assistance is provided for the review boards, a sure sign that they don't count for much in the priorities of science.

The Pollyanna mind-set of the science establishment was encapsulated last year in a defense of the failed institutional review board system by Jordan Cohen, president of the Association of American Medical Colleges, which lobbies hard to keep regulatory strings off the money that Washington provides for medical research and education:

"Guided by principles of beneficence, justice and respect for persons," he stated, "[institutional review boards] weigh the risks posed by research against the benefits that the research may offer to the patient and society. They work collaboratively with investigators, the vast majority of whom are motivated by altruism."

He's probably right about the good intentions of the vast majority of scientists. The worry is that safeguards for deterring the others are close to nonexistent.

Daniel S. Greenberg, a science journalist, is a visiting scholar at Johns Hopkins University.