About a third of an ounce of botulism toxin poured by bioterrorists into a milk truck en route from a dairy farm to a processing plant could cause hundreds of thousands of deaths and billions of dollars in economic losses, according to a scientific analysis that was published yesterday despite efforts by federal officials to keep the details secret.

The analysis by researchers at Stanford University, posted yesterday on the Web site of the Proceedings of the National Academy of Sciences, seeks to quantify security weaknesses in the nation's milk-supply chain and makes recommendations for closing those gaps.

Although some suggested changes are underway, federal officials felt the material had enough potential for misuse to warrant a last-minute effort to halt publication. That effort, which delayed the report's release by a month but ultimately did not keep it from becoming public, proved to be as contentious as the publication itself. It has assured the report's place in the scientific canon as one of the first test cases of how to balance scientific freedom and national security in the post-Sept. 11 era.

Study leader Lawrence M. Wein, whose previous research had forecast the likely effects of terrorist attacks involving anthrax and smallpox, said he was surprised by the government's push to block publication, which involved a flurry of phone calls and meetings with officers of the National Academies. The organization advises the federal government on matters of science and publishes the journal.

Last fall, Wein said, he briefed high-ranking officials of the departments of Homeland Security and Health and Human Services, along with dairy industry representatives, on his work.

"It was clear the dairy people were nervous about this paper coming out," Wein said. But when federal officials did not follow up, he said, he assumed they had concluded -- as had every reviewer at the National Academies -- that the information in the article was publicly available and easily obtained through a Google search.

Bill Hall, a spokesman for HHS, said yesterday that his department still opposes publication but was not in a position to block release of the data, which are not classified.

"We don't see eye to eye on this," Hall said. "If this ends up being the wrong decision down the road, the consequences could be quite severe and HHS will have to deal with it, not the National Academies."

The analysis by Wein and graduate student Yifan Liu considered what might happen if terrorists poured into a milk tanker truck a couple of gallons of concentrated sludge containing as much as 10 grams of botulinum toxin, a potent bacterial nerve poison now popular in low doses as a wrinkle eraser.

Because milk from many sources is combined in huge tanks holding hundreds of thousands of gallons, the toxin would get widely distributed in low, but potentially lethal, concentrations and within days be consumed by about 568,000 people, the report concludes.

The researchers acknowledge that their numbers are very rough. But depending on how thoroughly the milk was pasteurized (which partially inactivates toxins) and how promptly the outbreak was detected and supplies recalled, about 400,000 people would be likely to fall ill, they conclude.

Symptoms of botulism food poisoning arise within hours and progress from cramps, nausea and vision problems to paralysis and death by asphyxiation. Although only 6 percent of victims would generally be expected to die, the death rate could easily hit 60 percent, they conclude, because there would not be nearly enough mechanical ventilators or doses of antitoxin to treat so many victims.

Children could be hit first and hardest, because milk goes directly from processing plants to schools, avoiding the grocery-distribution system.

"They'd be the canaries," Wein said.

The report concludes that the most efficient ways to reduce such risks are to insist that latches on tanker trucks have locks; improve pasteurization processes; and develop tests that can detect contamination before milk is delivered to outlets -- changes, the team concludes, that are likely to cost just a few pennies per gallon.

Publication was scheduled for the week of May 30, but was abruptly postponed days before that date when HHS officials contacted the National Academies with concerns that the paper might inadvertently aid terrorists, according to an accompanying editorial written by Bruce Alberts, president of the Academies.

Those concerns were discussed in detail on June 7, after which the Academies decided to publish. By then, a preprint of the article had been widely distributed to journalists as part of the journal's standard procedures, and the New York Times had published a summary by Wein in an opinion piece.

Barry R. Bloom, dean of the Harvard School of Public Health, who oversaw an independent review of the paper earlier this spring, said he is convinced that the report did more good than harm by quantifying the risks posed at each point in the milk-delivery system -- a difficult job that now allows the industry and regulators to concentrate security efforts where they are most needed.

"This paper didn't just slip in with no one thinking about it," Bloom said. "But science depends on openness and the free exchange of ideas. And being aware of threats gives us a better chance of protecting against them than not being aware of them and having only the terrorists aware of them."

A national security directive signed by President Ronald Reagan and still in force demands that fundamental scientific information remain openly accessible unless it is formally classified.

Chris Galen, a spokesman for the National Milk Producers Federation in Arlington, criticized the Academies' decision, saying the information "could inform someone with malicious designs on food safety, even just as a prank."

The need for improved pasteurization "is something that has already been addressed" by the industry, he said, as has the need to keep locks on truck latches.

He acknowledged, however, that those improvements, encouraged by the Food and Drug Administration in recent years, are not mandatory. And although he said the newer standards are being "widely followed," he conceded he had no data to indicate what proportion of dairies and milk processors are adhering to the tougher recommendations.