The latest studies were not without uncertainty; the biggest one, for example, did not have information on whether smokers continued to smoke throughout the 16-year period.

But for the science panel, the evidence was clear enough.

Its conclusion, in 2005, was that the daily limit for fine particles should be lowered sharply from 65 micrograms per cubic meter of air to 35 micrograms or below. That is, pollution readings in a given county would not be allowed to spike above 35 for more than a few days a year.

Average yearly limits, the science panel felt, should be lowered from 15 micrograms per cubic meter to 14 or 13.

For every decline of a single microgram in average levels, deaths would fall roughly 1 percent, the evidence indicated. That could translate into preventing thousands of premature deaths.

But it would require significant pollution controls: from putting costly scrubbers on coal-burning power plants to mandatory cutbacks on engine idling of diesel trucks.

In recent years, Philadelphia's fine-particle measurements have averaged about 15 micrograms per cubic meter of air - on the cusp of meeting the standards. New York tends to be slightly higher, and Los Angeles is the nation's worst, at times above 20.

Fine particles have been a special concern in Camden's Waterfront South neighborhood, where the problem is aggravated by a cement plant, other industry, and the steady rumbling of about 700 diesel trucks per day.

Henderson, the science panel chair, had been studying air pollution since 1967, publishing 240 articles. She had no doubt the rules needed to be tougher.

"Setting it at 15 didn't give any margin of safety," she said.

'Silly' logic

All seven of the panel's charter members agreed, as did most of the additional 15 scientists who were brought in for their expertise.

The only two to dissent were both former chairmen of the panel: George Wolff, a scientist at General Motors Corp., and Roger O. McClellan, a toxicologist with extensive experience both at an industry-funded group and at an independent research institute.

McClellan said he did not think it was the role of scientists to recommend specific numbers, as the law gives that power to the administrator.

"I think we start getting on pretty thin ice when scientists start saying what is going to be an adequate margin of safety," he said.

In September 2006, Johnson announced his decision. He cut the daily limit from 65 micrograms to 35 - a dramatic reduction that fell within the range recommended by the scientists and his own staff.

The annual average standard, on the other hand, Johnson left unchanged at 15 - despite recent studies suggesting that this everyday background pollution killed even more people than occasional spikes.

His explanation, published in the Federal Register, was curious:

The big studies found evidence of premature deaths well below the existing 15-microgram pollution limit. But on average, the pollution level for the dozens of cities studied was about 18 micrograms. Since the existing standard was 15 micrograms, well below 18, Johnson decided it was already tough enough.

The evidence of harm, he wrote, was strongest at the long-term average - around 18 - because that's where most of the data were.