The chickens were already getting sick when Yoshihiro Kawaoka arrived in the U.S. in August 1983. A few months before, in April, a bird flu virus had arisen in the poultry farms of eastern Pennsylvania, but veterinarians had deemed it to be “low pathogenic”—meaning it made chickens sick but did not kill many of them. As the virus swept through the poultry farms, however, a new strain developed. Chickens began to die in large numbers, and farmers started to fear for their livelihoods. The state called in the U.S. Department of Agriculture, which set up a temporary command and control center in a strip mall outside of Lancaster. To contain the epidemic, it culled 17 million birds from Pennsylvania down through Virginia.

Kawaoka was a young researcher from Japan who was starting work at St. Jude Children’s Research Hospital in Memphis. His boss, virologist Robert Webster, had a theory that human influenza viruses originate in bird populations—that they circulate harmlessly among ducks and geese and that, every once in a while, a strain will evolve the ability to live in the human upper respiratory tract. To combat human influenza, Webster asserted, you first had to understand bird flu. In November, when Webster heard that the outbreak had become serious, he dropped everything and headed to its epicenter.

Kawaoka stayed behind and watched the crisis unfold from behind the air lock of the Memphis hospital’s biocontainment laboratory. He took samples sent back to him from the field, extracted the virus and cultured it. He then infected chickens that he kept in cages along a wall and waited to see what happened. What he found disturbed him: each and every chicken died—a mortality rate of 100 percent. In autopsies, he found that the virus was a ruthless pathogen, attacking almost every organ—similar to what some strains of Ebola do to humans.

In the months after the crisis, Kawaoka puzzled over why the April strain of the virus was so mild and why the strain that it evolved into by November was so deadly. He decided to compare the two. The difference, he discovered, came down to relatively small changes in the virus. “What this tells you,” Kawaoka told me in an interview in 2010, “is that a highly pathogenic virus was generated from a single mutation. And it tells you there are many sources of highly pathogenic influenza viruses. It’s all out there in birds.”

The experience brought home to Ka­wa­oka the urgency for scientists to figure out how bird flu can cause trouble for humans—the better to detect it early or to prepare effective vaccines and treatments. In particular, he wanted to know if a lethal bird flu akin to the one that burned through poultry farms in 1983 could turn into a human disease. And if so, what sequence of genetic code would the virus have to acquire?  

Nearly three decades later Kawaoka got an answer. He took an avian flu—an H5N1 type—that lives in birds and combined it with the H1N1 pandemic virus of 2009. Then he tested his hybrid virus on ferrets—a common research stand-in for humans—and found that it spread easily by airborne droplets. With this result, the notion that an H5N1 influenza virus could become a human pathogen was no longer hypothetical. If he could do it in a lab, nature could do it, too.

Kawaoka submitted his paper to the journal Nature, which then sent it out to his peers for review, a standard practice. (Scientific American is part of Nature Publishing Group.) Virologist Ron Fouchier of Erasmus Medical Center in Rotterdam also independently concocted a potentially human transmissible H5N1 virus and tested it on ferrets; he submitted his paper to the journal Science. At some point, the White House got wind of the studies. By December 2011 biosecurity officials were pushing for a delay in publication and a moratorium on the research.

What had biosecurity experts worried is that one of these viruses could possibly do to people what the 1983 virus did to chickens. If that were the case, the research might serve as a blueprint for a bioweapon. Or perhaps the virus itself could escape from a lab via a worker who became infected accidentally. For the months after the submission of the papers, scientists argued publicly and often vociferously with one another about whether the new viruses were potentially lethal and what kinds of restraints, if any, should be applied to work on H5N1 influenza viruses. The practice of science, which thrives on the free flow of information and the propensity of scientists to follow their curiosity wherever it may lead, collided with the need to keep people safe from a pathogen that could arguably be considered a potential weapon of mass destruction—every bit as devastating, and troublesome to manage, as nuclear weapons.

The first recorded instance of a “fowl plague” on poultry farms occurred in the countryside of northern Italy in 1878. It was thought to be a particularly virulent form of cholera.  By 1901 scientists had pegged it to a virus of some kind. By 1955 they realized it was type A influenza, similar to strains that infect humans, which later led Webster and others to wonder if there was some relation between influenza in birds and human outbreaks.

Webster’s hunch about birds being a reservoir for precursors to human viruses is now conventional wisdom. Wild birds carry such viruses around in their gastrointestinal tract without becoming sick and transmit the virus through feces. If a wild bird infects a chicken on a poultry farm, the virus may get opportunities to interact with a range of additional viruses through close contact with pigs and other animals. This is indeed what has happened in the live animal markets and backyard farms of China and southern Asia. Influenza viruses are notorious for their ability to change, through a combination of mutation and “reassortment”—a borrowing of genes from other viruses. An open farm acts like a virus convention, where different strains swap genetic material like conventioneers swap business cards.

In the past few decades influenza specialists have focused their worry on the H5N1 strains circulating on Asian farms. Type A influenza viruses are categorized by their surface proteins hemagglutinin and neuraminidase—the “H” and “N” in the species designations. (The 1983 virus was an H5N2.) If a virus can be said to have a personality, the H5N1 virus seems restless and unpredictable. For instance, the virus was thought to be benign in wild birds, but in 2005 thousands of ducks, geese, gulls and cormorants were found dead in Qinghai Lake in Central China, apparently killed by H5N1. In the past decade H5N1 has killed civets in Vietnam and tigers in a Thai zoo.

It has killed people, too. During the outbreak among poultry in Asia in 1997, a three-year-old boy in Hong Kong became the first known human fatality. By year’s end the death toll was six. To stem the outbreak, authorities in China and neighboring countries oversaw the culling of millions of birds. Still, the virus came surging back in 2004 in Thailand, Vietnam, China and Indonesia.

All told, around 350 people have died from H5N1, most  from contact with birds. The absolute number is not high, but the virus, according to the World Health Organization, has a mortality rate of about 60 percent. In contrast, the 1918 influenza virus, which killed 20 million to 50 million people, had a mortality rate of about 2 percent. Since the Kawaoka and Fouchier papers surfaced last fall, the actual mortality rate of H5N1 has been the subject of intense debate. Some scientists—notably, Peter Palese, professor of infectious diseases and chair of microbiology at the Mount Sinai School of Medicine—argue that mild cases of H5N1 have gone underreported or do not register in tests, which has artificially driven up the mortality rate. Others argue that deaths from H5N1 have gone underreported, which may make the mortality rate appear lower than it actually is. Kawaoka and Fouchier have reported low mortality among ferrets for their lab-made viruses. Whatever the danger these particular viruses might or might not pose, the fact that H5N1 could potentially spread easily among humans is not good news.

In September 2001 anthrax that had been weaponized as a fine white powder made its way through the U.S. mail, killing five people and terrorizing a nation already skittish from the World Trade Center and Pentagon attacks on 9/11. Spending on biodefense soared. Since 2001 the U.S. government has plowed more than $60 billion into vaccine stockpiling, disease surveillance and basic research into potential bioweapons agents, including influenza. The National Institute of Allergy and Infectious Disease (NIAID), the major source of funds in the U.S., nearly tripled its budget on influenza research in fiscal year 2003—from $17 million to $50 million—and doubled it again to $100 million in 2004. In 2009 funding hit a peak of nearly $300 million, from which it has come down slightly. Kawaoka was the recipient of some of that largesse. Since 2006 he has received nearly $500,000 a year from NIAID for research on the “pandemic potential of H5N1 influenza viruses,” according to the National Institutes of Health Web site. Fouchier got his funding from Palese’s group at Mount Sinai, which subcontracted the work from a grant from NIAID. Fouchier’s lab made mutations to an H5N1 virus to enhance transmissibility and then passed the virus to ferrets until it spread via airborne droplets among them. The Centers of Disease Control and Prevention also had a group investigating transmissibility of H5N1, but it was not as successful as Kawaoka’s and Fouchier’s groups.

For years after 9/11, however, concerns over smallpox as a potential bioweapon eclipsed those of influenza. The variola virus that causes smallpox kills one in three people infected and persists for years between hosts. It was declared eradicated in 1979. Although officially only two samples are kept under lock and key in Atlanta and in Koltsovo, Russia, there have been persistent rumors of other, illicit samples. In response to heightened fears after the 9/11 attacks, the U.S. stockpiled about 300,000 doses of smallpox vaccine, which now sit in secret warehouses throughout the country.

Influenza made it onto the bioweapons agenda in 2005, but biosecurity officials gave it a pass. Scientists had succeeded in reconstructing the 1918 pandemic flu virus from tissue samples of human remains that had been frozen in Arctic ice. The National Science Advisory Board for Biosecurity (NSABB) conferred and decided that the benefits to science and public health outweighed the security risk. Current NSABB chair Paul Keim recently called that decision “a mistake.” The 2009 pandemic virus, an H1N1 type with low pathogenicity, made the issue moot by conferring at least partial immunity to the 1918 virus to much of the world’s population. Since H5N1 is novel to the human immune system, there is no natural resistance.

Some defense experts now consider Kawaoka’s and Fouchier’s lab-made H5N1 viruses to be potentially more dangerous than smallpox. Influenza viruses are more contagious than variola and move more quickly through human populations, which gives public health officials less time to marshal vaccines and treatments. “Influenza is the lion king of transmissibility,” says Michael Osterholm, director of the Center for Infectious Disease Research and Policy at the University of Minnesota and an outspoken member of the NSABB. A highly transmissible H5N1 virus with a human mortality rate even approaching the 60 percent observed so far among bird flu victims is a terrifying prospect. As Osterholm has pointed out, even at one-twentieth the pathogenicity, H5N1 would be deadlier than the 1918 pandemic virus. NSABB called for a withholding of details in the Kawaoka and Fouchier papers last December but gave the go-ahead for full publication in March.

There is general agreement in the biosecurity community that bird flu—or, to be specific, H5N1 viruses made in the laboratory to be transmissible among mammals—is a potential bioweapon, which, like smallpox, has to be managed. “The very fact that this virus exists creates a risk,” says Richard H. Ebright, a biodefense expert and chemical biologist at Rutgers University. “It creates the risk of accidental release, and it creates the risk that someone will turn it into a weapon.”

What has defense experts, as well as many scientists, miffed is that the research proceeded without any analysis of the benefits and the risks beforehand. The NSABB, purely an advisory board with no oversight responsibility, got involved only after prodding by the White House. In 2007 John Steinbruner and his colleagues at the Center for International and Security Studies at Maryland wrote a report recommending “some constraint on freedom of action at the level of fundamental research, where individual autonomy has traditionally been highly valued for the best of reasons.” The report was largely ignored. After the Fouchier and Kawaoka papers came to light, however, the U.S. government called on funding agencies to perform risk assessments on research involving the H5N1 and 1918 flu viruses.

Steinbruner and others recommend some kind of international oversight group with some power to impose mandatory constraints on potentially dangerous research and oversee it, much as the WHO does now with smallpox. “It wouldn’t be an airtight protection, but it would establish the norm that nobody can go off into a closet and do these experiments,” Steinbruner says. An H5N1 virus engineered to spread among mammals “is an agent of mass destruction that gets into the nuclear weapons league and even exceeds it,” he adds. “It is a very dangerous pathogen. It’s not a matter of [scientists] being personally careful. There’s got to be some institutional safety procedure.”

How restrictive should those procedures be? Nuclear weapons technology is subject to military classification, which means some research can only be conducted in secret. Unlike nuclear weapons, however, influenza is a matter of global public health. Classifying some aspects of H5N1 research would leave scientists and health officials in the dark about one of the world’s bigger public health threats. In contrast, many security experts argue in favor of restricting research on mammal-transmissible viruses to only the most secure labs—more secure than the labs Kawaoka and Fouchier did their work in. Such restrictions would put the research out of reach of many scientists.

Many investigators have been passionate in their defense of the kind of work Kawaoka and Fouchier have done on the grounds that the more we know about H5N1, the better we can protect ourselves from the natural threat. Science, the argument goes, advances best when research activities are unfettered. Pinning down precisely what genetic components are needed to confer traits such as lethality and transmissibility on H5N1 would allow health experts to be on the watch for dangerous new strains that emerge in the wild and prepare for them in advance. Once a novel human flu virus crops up and begins to spread, it is too late to stop the first wave of infection. Flu vaccine production typically takes six months to complete, sometimes more. For instance, by the time the H1N1 virus came to the attention of health officials in April 2009, it had spread widely in Mexico and the U.S. and was well on its way to becoming a pandemic.

Moreover, one of the genetic components Kawaoka identified as conferring transmissibility on H5N1 has been observed in natural viruses, which suggests that the roulette wheel is already in spin. “Because H5N1 mutations that confer transmissibility in mammals may emerge in nature, I believe that it would be irresponsible not to study the underlying mechanism,” Kawaoka wrote in an essay in Nature. (He declined to be interviewed for this article.) Fouchier has defended his work on the same grounds.

Having the genetic details of potentially deadly flu viruses is of little use, however, without the funding, networks and access to animals out in the field. During the H5N1 outbreaks, virologists began rigorous monitoring of the live animal markets in southern China, but those measures have not been applied consistently elsewhere in China or Southeast Asia. In the U.S., livestock farms often bar health officials from testing their pigs even though precursors of the 2009 H1N1 pandemic are thought to have kicked around U.S. pig farms for years before emerging in Mexico [see “Flu Factories,” by Helen Branswell; Scientific American, January 2011].

Surveillance may never be good enough to forestall human pandemics. “We’re better prepared now than we were prior to the H1N1 pandemic,” says Nancy Cox, director of the Influenza Division of the CDC, “but the world is not prepared for the emergence of highly transmissible, highly pathogenic influenza virus in humans. Honestly, I don’t think the world ever will be unless we have a universal vaccine that protects against all strains.” A universal vaccine is not in sight, which leaves us in the uncomfortable position of having too much knowledge and too little.

This article was published in print as "Waiting to Explode."