The possible connection between a pathogen and obesity is just one more example of a startling revisionist hypothesis that is, well, infecting biomedical research: the notion that germs cause, or at least contribute to, an increasingly long list of chronic diseases, many of them exceedingly common. With the help of new molecular techniques, scientists can now easily zero in on disease organisms in human patients and, using statistical methods, link them with particular maladies. Infection is suspected of being at least partly responsible not only for metabolic disorders like diabetes and mental disorders like schizophrenia, but also for the two ailments that will eventually kill most of us: cancer and heart disease.

"In my medical school I was taught that gastric ulcer is due to hurry, worry and curry," says Nikhil Dhurandhar, Ph.D., who was educated in Bombay and is now an assistant professor at Wayne State University in Detroit. He is talking about the disease that is everybody's favorite example of biomedicine's new infectious paradigm. For decades the medical party line was that ulcers were stress-induced and incurable, the archetypal ailment of the Age of Anxiety -- except that almost all ulcers are caused by a bacterium, Helicobacter pylori, and can be easily, permanently and inexpensively cured in less than a month with antibiotics. Researchers first described the relation between H. pylori infection and ulcers in 1983, but their findings were scoffed at for years and incorporated into the medical canon only recently.

Dhurandhar is on a similar mission, but meeting less resistance. Backed by a powerful mentor, obesity researcher Richard Atkinson of the University of Wisconsin, Dhurandhar is slowly amassing evidence that some proportion of obesity -- at this point his work suggests, scarily, that it may be a very large proportion -- is the consequence of infection by a virus.

Most of this research has been done in animals, but some small human studies, so far mostly unpublished, back it up. The latest published contribution appeared last month in the International Journal of Obesity and Related Metabolic Disorders. The paper reported on experiments in which Dhurandhar and his colleagues inoculated chickens and mice with a human virus they suspect of promoting fat gain. It did: The infected animals gained two-thirds more fat than uninfected control animals.

The virus, designated Ad-36, is one of 50 adenoviruses known to infect people. In an electron microscope, adenoviruses look a bit like the World War II mines used to block shipping channels -- spherical and studded with spikes that help them attach to host cells. Despite that ominous configuration, most appear to be benign. A few cause respiratory infections, pinkeye and diarrhea.

As for Ad-36, it doesn't seem to give diarrhea or colds to Dhurandhar's birds and rodents, at least. "They are quiet for a day or so, and then they bounce back," he says.

A handful of viruses are known to make animals fat, which is how Dhurandhar -- who operated obesity treatment centers in India before he came to the U.S. nearly a decade ago -- got interested in them. An unexpected characteristic of these infections is that when the animals get fat the levels of cholesterol and triglycerides in their blood plummet. Increased body fat is generally accompanied by increased cholesterol and triglycerides, which ferry fats around in the blood.

In a study published in 1997, Dhurandhar examined blood from patients in Bombay looking for antibodies against an adenovirus that was epidemic among Indian chickens -- and which killed them after making them fat. He found viral antibodies in 10 of his 52 human patients, evidence that they had been exposed to the virus at some time in their lives. The 10 tended to be fatter than the other patients, and also tended to have lower cholesterol and triglycerides.

Dhurandhar says he has strengthened the case in a U.S. study. In findings presented at conferences and in a paper Dhurandhar says has been submitted for publication, Ad-36 antibodies turned up in 100 of 313 obese subjects, but in just four of 92 lean controls. In short, nearly one in three fat Americans in Dhurandhar's study showed evidence of prior infection with the virus, compared with only one in 20 slim ones. Moreover, the infected subjects also had lower-than-usual cholesterol levels.