"We had new information, confidential information, about benefits of this research, and we also had confidential information about the risks involved," said Paul Keim of Northern Arizona University, who is the acting chairman of the panel. "And the balance began to change."

Explaining its decision, announced last Friday, the National Science Advisory Board for Biosecurity (NSABB) said in a statement that "the data described in the revised manuscripts do not appear to provide information that would immediately enable misuse of the research in ways that would endanger public health or national security."

The board was unanimous in recommending that the study conducted at the University of Wisconsin, Madison, be published in full.

But it split 12-6 in favor of publishing a study from Erasmus Medical Center in the Netherlands. It did not explain the lingering concerns about that research.

The journals have said they will publish the papers this year.

FRAGILE GLOBAL COALITION

The most significant change in the calculation of the risks and benefits of publishing the papers reflected the delicate global politics of flu research, microbiologist Arturo Casadevall of Albert Einstein College of Medicine in New York City told Reuters.

A briefing to the board by the World Health Organization (WHO) raised fears that censoring the papers would threaten the "fragile" international collaboration that the WHO had assembled to combat avian flu, said Casadevall, a member of the biosecurity board.

The WHO had worked for years to persuade Indonesia and other countries to share samples of avian flu, or H5N1, with the international scientific community.

Previously, Indonesia had declined to do so under a principle its government called "viral sovereignty," by which it meant that microbes found in Indonesia belonged to the state and did not have to be shared with outsiders.

Indonesia viewed the withholding of the two papers as equivalent to its withholding samples of virus. That raised concerns that if the papers were not published, Indonesia and other countries coaxed into cooperating with the WHO would cease doing so.

"I and many others on the board were worried that if we had a flu outbreak our only hope would be international collaboration," said Casadevall.

Withholding the papers posed a risk to that collaboration, a risk the biosecurity board viewed as more dangerous than the possibility that terrorists would use the information to create an H5N1 pandemic.

BIOTERRORISM FEARS

Last December the NSABB recommended the two papers not be published in full by the journals where they were under consideration, Nature and Science.

The panel feared that details of the studies, which induced mutations in the H5N1 avian flu virus that made it transmissible among mammals by air rather than by close physical contact, could be used for bioterrorism.

Critics of the recommendation raised fears that important science was being censored.

The ensuing debate raised questions about whether the research should have been done at all, as well as whether current national and international rules on biosafety and biosecurity are sufficient to protect the public from dangerous microbes.

The biosecurity panel spent two days last week considering the papers. Both papers describe how scientists altered several genes of natural, or wild-type, H5N1 in a way that allowed it to spread from the airways of infected ferrets to other ferrets caged nearby.

The decision by the NSABB to sign off on the publications will have little practical effect. Experts convened by the WHO in February recommended that the papers be published. Keim was among the experts at that meeting.

So far, the natural form of H5N1 has infected tens of millions of ducks, geese, chickens, and other birds. But the only people to be infected - 598, of whom 353 have died - were those who came into close contact with the flocks.

The board was persuaded by an additional benefit of publishing the research: by informing countries where H5N1 is endemic, it would allow scientists there to be on the lookout for the mutations that make the virus more transmissible.

"Last fall we were told that there would be no benefit to surveillance because no surveillance was being done in these countries," said board member Casadevall.

"But that changed. We had presentations that if countries had this information -- these are the mutations that could get us into trouble -- it would catalyze surveillance."

MORE TRANSMISSIBLE, LESS LETHAL

Ron Fouchier, who led the Erasmus experiments, said the NSABB decision was "very much to our pleasure." He and Keim stressed that nothing would be censored in the paper.

Instead, the paper to be published by Science will include "clear and explicit" information about the lethality of the mutated virus, which is less than the NSABB originally believed. In other words, although the genetic mutations made H5N1 more transmissible among mammals, they apparently also made it less deadly.

Fouchier and Keim agreed that the 11th-hour recommendation last December against releasing the papers was not how biosecurity concerns should be addressed.

Instead, under the new U.S. policy, such studies will be reviewed in a "cradle to grave" way, they explained, suggesting that potentially dangerous research will receive a more intensive review before it is so far along as to be publishable.

Asked whether the NSABB had misunderstood the original papers, Keim said it had spent more than 200 hours reviewing them and faced enormous pressure from all sides - from the journals, from the researchers and from the National Institutes of Health, which funded both studies - to act quickly.

"I think this is not the process that should be used for reviewing these types of papers in the future," he said.

(Reporting By Sharon Begley and Kate Kelland; Editing by Xavier Briand)

• A Permanent Economic Emergency