Last year Yuk Yung, an expert in planetary atmospheres at the California Institute of Technology, was amazed to hear the Bush administration's vision of swapping the colossal system of refineries, pipelines, industrial furnaces, and vehicles devoted to fossil fuels for a pollution-free hydrogen economy. The goal: no more climate-warming carbon dioxide, no more smog, only benign water vapor wafting from tailpipes and smokestacks. But boy, he thought to himself, that will take a lot of hydrogen.

With colleagues he soon calculated that it would take around 600 million tons of hydrogen yearly to generate the energy that natural gas and oil now provide the world. And Yung is sure some of that hydrogen would leak. In a series of reports--most recently at last week's American Geophysical Union meeting in San Francisco--he and others are warning that hydrogen that escapes could seriously harm the environment.

The lure of a hydrogen economy is clear. Energy Secretary Spencer Abraham calls it a route to "a world where our pollution problems are solved and where our need for abundant and affordable energy is secure." Surrounded by auto industry chiefs, he unveiled its first major component, a program called FreedomCAR, in early 2002 at a Detroit auto show. The overall plan even got a mention in President Bush's State of the Union address early this year--with at least $1.7 billion to be spent on research over the next few years.

Critics call the program a distraction from more immediate steps to clean the air, like boosting fuel economy. They also stress that no one knows how to liberate hydrogen from water or fossil fuels cleanly and cheaply.

But even critics didn't worry much about the hydrogen itself until last June, when Yung and like-minded colleagues published an article in the journal Science listing potential consequences if much hydrogen gets loose. If the tanks, pipelines, and other components of a hydrogen economy leaked 10 or 20 percent of the gas, as they feel is plausible, the air's minuscule hydrogen content could rise as much as fourfold. While the free hydrogen wouldn't be toxic and would be too dilute to burn, Hindenburg-style, they say it could:

Affect weather patterns by cooling the stratosphere, a dry layer of the atmosphere 10 to 50 miles up.

Worsen the ozone holes over the north and south poles. The cooling of the stratosphere could intensify the vortexes of wind where ozone is trapped and broken down.

Warm the climate. Methane, a greenhouse gas, could increase as the hydrogen reacted with airborne chemicals that normally degrade methane.

Change the mix of bacteria and other microbes in the soil, with unknown effects on the environment.

Many have said Yung exaggerates the leakage. "Ten percent is high, 20 percent ridiculous," says Michael Prather of the University of California-Irvine. But Prather does agree that some hydrogen would escape. "People who say this is a clean gas are smoking something," he says. "It could turn out OK, but right now they are calling for an experiment on the entire planet."

In a study presented at the AGU meeting, atmospheric chemist Manvendra Dubey of Los Alamos National Laboratory concluded that a hydrogen economy might increase the gas's concentration in the air by about one third, rather than quadruple it as Yung suggests. But even that, he agrees, could bring unexpected "wild cards." He compares the prospect of a hydrogen economy to the adoption of chlorofluorocarbon refrigerants like freon to replace hazardous ammonia decades ago. "CFCs were green 50 years ago," he says. "Then we found the ozone hole."

As for Yung, he says, "I am a fan of green hydrogen." But before the world starts manufacturing huge quantities of the gas, he says, it had better be sure about what happens when it spills.