Where did it come from? More than three years into the pandemic and untold millions of people dead, that question about the Covid-19 coronavirus remains controversial and fraught, with facts sparkling amid a tangle of analyses and hypotheticals like Christmas lights strung on a dark, thorny tree. One school of thought holds that the virus, known to science as SARS-CoV-2, spilled into humans from a nonhuman animal, probably in the Huanan Seafood Wholesale Market, a messy emporium in Wuhan, China, brimming with fish, meats and wildlife on sale as food. Another school argues that the virus was laboratory-engineered to infect humans and cause them harm — a bioweapon — and was possibly devised in a “shadow project” sponsored by the People’s Liberation Army of China. A third school, more moderate than the second but also implicating laboratory work, suggests that the virus got into its first human victim by way of an accident at the Wuhan Institute of Virology (W.I.V.), a research complex on the eastern side of the city, maybe after well-meaning but reckless genetic manipulation that made it more dangerous to people.
If you feel confused by these possibilities, undecided, suspicious of overconfident assertions — or just tired of the whole subject of the pandemic and whatever little bug has caused it — be assured that you aren’t the only one.
Some contrarians say that it doesn’t matter, the source of the virus. What matters, they say, is how we cope with the catastrophe it has brought, the illness and death it continues to cause. Those contrarians are wrong. It does matter. Research priorities, pandemic preparedness around the world, health policies and public opinion toward science itself will be lastingly affected by the answer to the origin question — if we ever get a definitive answer.
But much of the evidence that might provide that answer has either been lost or is still unavailable — lost because of failures to gather relevant material promptly; unavailable because of intransigence and concealment, particularly on the part of Chinese officialdom at several levels.
Take the natural-spillover hypothesis, for instance, and assume that the virus passed to humans from a wild animal — maybe a raccoon dog (a foxlike canine) or a Malayan porcupine — somewhere in the Huanan market. To test that hypothesis, you would want samples of blood, feces or mucus taken from the raccoon dogs, porcupines and other wildlife that languished, caged and doomed, in the market. You would screen those samples for signs of the virus. If you found the virus itself, or at least sizable bits of its genome, you would then make a comparative analysis of genomes, including some from the earliest human cases, to deduce whether people got the virus from the wildlife or vice versa.
But you can’t do that, because whatever raccoon dogs or porcupines or other wild animals were on sale in the market during December 2019 had vanished by Jan. 1, 2020. On that date, the market was closed by order of Chinese authorities, with no (reported) effort to sample the most suspect forms of wildlife.
Or take the lab-engineered-bioweapon hypothesis, as recently offered in an article in The Sunday Times of London. The two Times reporters cited unidentified “U.S. investigators” who “scrutinized top-secret intercepted communications” and concluded that the Chinese military was supporting a covert project to develop a weaponized coronavirus. The article also posited a related vaccine effort, to protect China’s populace once the killer virus was unleashed on the world. It’s a riveting narrative. The virus engineering occurred, according to this account, at the Wuhan Institute of Virology. The reporters didn’t name their intelligence sources or supply evidence to make their allegations concrete, but if they did, it would be explosive news.
Or take the lab-leak scenario, some versions of which point accusingly at a nonprofit organization in New York, EcoHealth Alliance, and its collaborative relationship with Dr. Zhengli Shi, a senior researcher at the W.I.V. Shi and her team study coronaviruses, especially those carried by bats, extracting fragments of viral RNA (the molecule in which coronavirus genomes are written) and occasionally live virus, from samples of guano and other bodily material, and assembling whole genome sequences, like jigsaw puzzles, from the fragments. They perform experiments, sometimes combining an element of one virus with the backbone of another, to learn how that element might function in the wild; and they publish scientific papers, warning which bat viruses could pose a threat to humans. What if a researcher or technician under Shi’s leadership, handling a virus very much like SARS-CoV-2, became infected by accident and then spread the infection to others? That question became, from the early months of the pandemic, a suspicion and then a hypothesis and then an accusation.
Even now, the trade in claims and counterclaims remains brisk. Last month, in a Substack newsletter called Public, three authors asserted — citing unnamed “U.S. government officials” — that one of the first people infected with SARS-CoV-2 was a scientist named Ben Hu, from Shi’s lab. That assertion was significant, and important if true, but no proof or identified sourcing so far supports it. Ten days later, the Office of the Director of National Intelligence released (as required by a law passed three months earlier) a declassified report outlining whatever was known to the U.S. Intelligence Community about potential links between the Wuhan Institute of Virology and the origins of the pandemic. The report concluded, among other things, that W.I.V. personnel had collaborated at times on coronavirus work with scientists associated with the People’s Liberation Army, but that (so far as available evidence showed) such work involved “no known viruses that could plausibly be a progenitor of SARS-CoV-2.”
And then, on July 11, the House Select Subcommittee on the Coronavirus Pandemic, led by Representative Brad Wenstrup, an Ohio Republican, convened a hearing at which he and colleagues interrogated two scientists, Kristian Andersen and Robert Garry, about their authorship of an influential 2020 paper that appeared in the journal Nature Medicine. That paper was titled “The Proximal Origin of SARS-CoV-2.” The tenor of the hearing was foretold by its own announced title: “Investigating the Proximal Origin of a Cover-Up,” and the proceedings that day consisted of accusation and defense, without shedding any new light, let alone yielding certitude about the origin of the virus.
Certitude is an elusive goal and a high presumption, even for science, even for a director of national intelligence, even for the chairman of a select congressional subcommittee. Philosophers have recognized that, and so have novelists and poets. “I was of three minds,” wrote Wallace Stevens, “Like a tree/In which there are three blackbirds.” In the poem, Stevens found 13 different ways of looking at a blackbird. There are at least that many ways of viewing the origin of SARS-CoV-2, and to do justice to the question, you’ll need, like him, to hold several possibilities in your mind at a time.
How you regard a blackbird or an origin hypothesis may be influenced by where you’re coming from. That’s an old truth, but I was reminded of it during a conversation with Jesse Bloom, an evolutionary biologist at the Fred Hutchinson Cancer Center in Seattle, and one of the best-qualified among those who argue that the lab-leak hypothesis deserves robust investigation. Bloom studies the evolution of viruses, for two reasons: It happens fast, and therefore illuminates evolution in general, and it has large implications for public health.
When I spoke with him back in February 2021, a year into the pandemic, and asked about the origin question, Bloom said, “I think what you have is a lot of people strongly defaulting to their prior beliefs.” Scientists who study zoonotic diseases (those that spill over from nonhuman animals into people) might be inclined to assume a natural origin. Scientists who have long argued against the risks of “gain of function” research (experimental work exploring the evolutionary capacities of potentially dangerous pathogens) might readily assume a lab leak. National security experts with strong views of the oppressive, secretive Chinese government might lean toward scenarios involving Chinese malfeasance and cover-up.
More recently, Bloom told me that his own “prior” inclination would be toward a natural spillover. “But you certainly wouldn’t think it’s, like, 99.99 percent the most likely explanation,” he said, adding, “There could be other possibilities.”
That gave me pause to consider my own priors. For the past 40 years, I’ve written nonfiction about the natural world and the sciences that study it, especially ecology and evolutionary biology. During the first half of that, my attention went mainly to large, visible creatures like bears, crocodiles and bumblebees and to wild places like the Amazon jungle and the Sonoran Desert. I came to the subject of emerging viruses in 1999, during a National Geographic assignment, when I walked for 10 days through Ebola-virus habitat in a Central African forest. Later I spent five years writing a book about zoonotic diseases and the agents that cause them, including the SARS virus, the earlier killer coronavirus now often called SARS-CoV-1, which emerged in 2002 and spread in human travelers from Hong Kong to Singapore, Toronto and elsewhere, alarming experts deeply. Scientists traced SARS-CoV-1 to palm civets, a type of catlike wild carnivore sold as food in some South China markets and restaurants. But the civets proved to be intermediate hosts, and its natural host was later identified as horseshoe bats.
The story of SARS is only one chapter in the saga of dangerous new viruses emerging from animals. The grim tale of how H.I.V. got into humans and caused the AIDS pandemic is another — a tale known partly by inference and partly by molecular evidence, and traceable back to a single blood-mingling event between a person and a chimpanzee, probably hunter and hunted, in the southeastern corner of Cameroon around the start of the 20th century. Human contact with nonhuman animals accounts for our influenzas as well, which usually emerge from wild aquatic birds. Hendra virus, in Australia, comes to humans from bats, generally through an intermediate host: horses. Machupo virus, in Bolivia, abides in rodents when not infecting people. Hantaan virus, discovered in Korea, and its relative Sin Nombre virus, in the American Southwest, also spill over from rodents. Nipah virus, in Bangladesh and some surrounding countries, comes from bats. It’s excreted in bat feces, saliva and urine, and when certain fruit bats visit date palm trees that are being tapped for their sugary sap — a custom in Bangladesh — the virus contaminates the sap, which is sold fresh on the street to local customers, some of whom die. These cases and many others like them are among my own priors, and no doubt they do incline me toward the idea of natural spillover. It happens often, sometimes with dire consequences.
Research accidents have occurred, too, in the history of dangerous new viruses, and longtime concerns over such accidents constitute the priors of some who favor the lab-leak hypothesis for Covid. Such accidents might number in the hundreds or the thousands, depending on where you put the threshold of significance and how you define “accident.” There was an event that (probably) reintroduced a 1950s strain of influenza in 1977, causing that year’s flu pandemic, which killed many thousands of people, and a 2004 needle-stick injury of a careful scientist, Kelly Warfield, while she was doing Ebola research (but she proved uninfected by Ebola). Also in 2004, just a year after the global SARS scare, two workers at a virology lab in Beijing were independently infected with that virus, which spread to nine people in total, one of whom died. This followed two other single-case lab-accident infections with SARS virus the previous year, one in Singapore, one in Taiwan.
When the first known cases of an “atypical pneumonia” began turning up at Wuhan hospitals in late 2019, and then exploded into a coronavirus outbreak in early 2020, the location itself seemed to fit, in different ways, the priors that might incline one toward either a natural-origin explanation or a lab-leak explanation. The potential lab-leak connection was easiest to note: The city contained a research facility, the Wuhan Institute of Virology, with a well-known laboratory devoted to coronavirus research. On the other hand, Wuhan was also a major nexus for the significant national trade in wild animals for food, fur and traditional medicines (estimated at more than $70 billion annually), where such creatures, and the viruses they carry, were sold at many crowded markets — one of which, Huanan, lay at or near the center of the spatial pattern of earliest known cases.
So, starting from simply those circumstances, was a lab accident more “likely” than a natural spillover? And under either of those scenarios, how much did Chinese-government pressure and obscurantism constrain the availability of evidence for assessing one or the other? Because there exists no definitive account — yet — of the particular events that delivered SARS-CoV-2 into the human population, even experts are forced to frame their views as probabilities, based on data and circumstance, influenced variously by prior beliefs as to how the world works.
In assessing the probabilities for yourself, you might want to step back from the noise, anger, vitriol and politicization that have clouded the controversy and focus on the evidence we do have. To that end, it may help to note some events in the order they occurred.
On Jan. 11, 2020, in Shanghai, just 11 days after first reports of the outbreak in Wuhan circulated globally, a team of scientists led by Yong-Zhen Zhang of Fudan University released a draft genome sequence of the novel virus through a website called Virological.org. The genome was provided by Edward C. Holmes, a British Australian evolutionary biologist based in Sydney and a colleague of Zhang’s on the genome-assembly project. Holmes is famous among virologists for his work on the evolution of RNA viruses (including coronaviruses), his pristinely bald head and his mordant candor. Everyone in the field knows him as Eddie. The posting went up at 1:05 a.m. Scotland time, at which point the curator of the site there in Edinburgh, a professor of molecular evolution named Andrew Rambaut, was alert and ready to speed things along. He and Holmes composed a brief introductory note to the genome: “Please feel free to download, share, use and analyze this data,” it said. They knew that “data” is plural, but they were in a hurry.
Immediately, Holmes and a small group of colleagues set to analyzing the genome for clues about the virus’s evolutionary history. They drew on a background of known coronaviruses and their own understanding of how such viruses take shape in the wild (as reflected in Holmes’s 2009 book, “The Evolution and Emergence of RNA Viruses”). They knew that coronavirus evolution can occur rapidly, driven by frequent mutation (single-letter changes in a roughly 30,000-letter genome), by recombination (one virus swapping genome sections with another virus, when both simultaneously replicate in a single cell) and by Darwinian natural selection’s acting on those random changes. Holmes traded thoughts with Rambaut in Edinburgh, a friend of three decades, and with two other colleagues: Kristian Andersen at Scripps Research in La Jolla, Calif.; and Robert Garry at the Tulane University School of Medicine in New Orleans. Ian Lipkin, of Columbia University’s Mailman School of Public Health, joined the huddle later. These five would form a sort of long-distance study group, aimed toward publishing a paper on SARS-CoV-2’s genome and its likely origin.
Holmes, Andersen and their colleagues recognized the virus’s similarity to bat viruses but, with more study, saw a pair of “notable features” that gave them pause. Those features, two short blips of genome, constituted a very small percentage of the whole, but with potentially high significance for the virus’s ability to grab and infect human cells. They were technical-sounding elements, familiar to virologists, that are now part of the Covid-origin vernacular: a furin cleavage site (FCS), as well as an unexpected receptor-binding domain (RBD). All viruses have RBDs, which help them attach to cells; an FCS is a feature that helps certain viruses get inside. The original SARS virus, which terrified scientists worldwide but caused only about 800 deaths, didn’t resemble the new coronavirus in either respect. How had SARS-CoV-2 come to take this form?
Andersen and Holmes were genuinely concerned, at first, that it might have been engineered. Were those two features deliberate add-ons, inserted into some coronavirus backbone by genetic manipulation, intentionally making the virus more transmissible and pathogenic among humans? It had to be considered. Holmes called Jeremy Farrar, a disease expert who was then director of the Wellcome Trust, a foundation in London that supports health research. Farrar saw the point and quickly arranged a conference call among an international group of scientists to discuss the genome’s puzzling aspects and the possible scenarios of its origin. The group included Robert Garry at Tulane and a dozen other people, most of them distinguished European or British scientists with relevant expertise, like Rambaut in Edinburgh, Marion Koopmans in the Netherlands and Christian Drosten in Germany. Also on the call were Anthony Fauci, then head of the National Institute of Allergy and Infectious Diseases, and Francis Collins, then director of the National Institutes of Health and therefore Fauci’s boss. This is the famous Feb. 1 call on which — if you believe some critical voices — Fauci and Collins persuaded the others to suppress any notion that the virus might have been engineered.
“The narrative going around was that Fauci told us, Change our mind, yada, yada, yada, yada. We were paid off,” Holmes said to me. “It’s complete [expletive].”
Andersen concurs. “There is no universe in which this would even be possible,” he told me. Recently, based on selections of their private email and Slack traffic made public, Andersen and his colleagues have been accused of concealment and dissembling: Their messages, critics contend, prove that even as they were deeply concerned in private about the engineered-virus or lab-release possibilities, they were striving to keep both out of public discussion. But as the researchers describe it, these apparent contradictions were simply a reflection of their fast-evolving views. After initial concern that the receptor-binding domain in SARS-CoV-2 might be a sign of engineering, for instance, they learned soon after the Feb. 1 conference call of a very similar RBD in a coronavirus that infected pangolins. It was detected from a public database by a bioinformatician in Houston, Matt Wong, and posted on the Virological website, where it eventually came to the group’s attention. It showed that such an RBD had evolved in the wild and might well have gotten into SARS-CoV-2 by recombination, the natural gene-swapping process. Andersen and the others also recognized that furin cleavage sites occur naturally in other coronaviruses, like the MERS virus, though not (as so far detected) in any other member of the subgenus to which SARS-CoV-2 belongs.
Such new data led to a new conclusion, in what Andersen called, on Twitter, “a clear example of the scientific process.” Sixteen days after the conference call, they posted a preprint (a draft, not yet peer-reviewed) of their paper, and four weeks later it appeared in the journal Nature Medicine — this was the one titled “The Proximal Origin of SARS-CoV-2.” Andersen and his co-authors stated their conclusion at the top: “Our analyses clearly show that SARS-CoV-2 is not a laboratory construct or a purposefully manipulated virus.” That still left the possibility of a natural virus, evolved in an animal host and passed into humans by zoonotic transfer — or perhaps a natural virus accidentally leaked? Near the paper’s end they stated something more nuanced: that while intentional engineering of the virus could be ruled out, “it is currently impossible to prove or disprove the other theories of its origin described here.” That said, they added, “we do not believe that any type of laboratory-based scenario is plausible.”
One other coronavirus quickly came to light as the closest known match to SARS-CoV-2. This wasn’t actually a virus “in the flesh” — in physical presence. It was a genome sequence, assembled from RNA fragments extracted from a fecal swab sample of a bat, captured in a mine several years earlier in 2013. The mine was in Yunnan Province, 1,200 miles southwest of Wuhan. The genome was 96.2 percent identical to the SARS-CoV-2 genome as sampled from people during the early days of the pandemic. That degree of similarity — or a 3.8 percent difference — suggests a common virus ancestor some years ago and independent evolution in the years since. So this represented a cousin to SARS-CoV-2, not its progenitor.
The work of sampling the bat and assembling the sequence (first just a portion, then, with better technology, nearly the whole thing) had been led by Zhengli Shi, at the Wuhan Institute of Virology. Shi and her team labeled the sequence RaTG13, coding the facts that it came from an individual of Rhinolophus affinis (Ra), the intermediate horseshoe bat, captured in that mine in Tongguan (TG), a town in the Mojiang district of Yunnan, in 2013. RaTG13 has attained renown, not just because it constituted strong evidence of SARS-CoV-2’s ancestry in bat viruses but also because the Mojiang mine figures in some of the more lurid scenarios for a lab-leak origin.
Part of what makes the very name Mojiang seem lurid is that in 2012, three workers at the mine died of unidentified respiratory infections after days of underground labor there. What got into their lungs and killed them? Was it a fungus? Was it a virus? Some lab-leak proponents suggest that those deaths, described in two obscure medical theses written in Mandarin, represent the earliest known fatalities from a virus — possibly RaTG13 — that either already was, or in Shi’s lab became, SARS-CoV-2 or its immediate progenitor (that is, something far more similar than a cousin). The inference is that Shi’s team, a year after the mine workers died, may have taken the virus back to Wuhan. But the Mojiang deaths were also reported in 2014 in the journal Emerging Infectious Diseases by scientists who found an entirely different virus, also potentially dangerous because it had similarities to Nipah and Hendra viruses, and was carried in the Mojiang mine by rats, not bats. One takeaway: Sample the rats and bats and other fauna in a mine, and you might well find a variety of viruses you wouldn’t want in your lungs.
Another problem with the RaTG13 scenario: Its genome differs from that of SARS-CoV-2 at more than 1,100 scattered positions throughout its genome. To engineer SARS-CoV-2 into existence by starting with RaTG13 would have been unreasonable and impractical, according to Holmes and other experts in coronavirus genomics. Furthermore, it’s important to remember that RaTG13 was a genome sequence, not a live virus: It was information, not a biological entity. Coaxing a virus that lies dormant in bat guano to grow in a cell culture is difficult, and usually the effort fails. Zhengli Shi told Jon Cohen, a senior correspondent for the journal Science, in her answer to a set of emailed questions, that she never grew RaTG13 in her lab. She told me the same thing during a two-hour conversation by Zoom: “No, no. We couldn’t culture any of the sample from this cave at Mojiang.”
Shi was in Shanghai for a conference on the night of Dec. 30, 2019, as she explained it to me, when word reached her about a mysterious respiratory illness spreading dangerously among people back in Wuhan. Preliminary lab results suggested a coronavirus — not SARS virus, but something similar — might be the cause. She was asked to help identify the thing. She put her lab team to work on that immediately and took a train back to Wuhan the next day. Within hours, her lab had received a partial sequence from another lab. Her first instinct was to compare it with sequences of viruses they had worked on themselves, “and we found it’s different,” she told me. “So, the afternoon of Dec. 31, I already know it’s nothing related to what we have done in our laboratory.”
Some critics, she was well aware, had suggested that her urgency in checking her own records was an implicit admission of error or guilt. “It’s normal!” was her response.
Jon Cohen mentioned the possibility of a lab leak in a report published in Science on Jan. 31, 2020, noting that not all the earliest confirmed cases had some direct link to the Huanan market. Fourteen of the first 41, according to one study, did not. Might those people have picked up their infections somewhere else, and maybe not from an animal at all? After describing a couple of vivid but unsupported allegations, including the idea that SARS-CoV-2 resembled a snake virus (and snakes were sold at Wuhan wet markets), Cohen added, “The Wuhan Institute of Virology, which is the premier lab in China that studies bat and human coronaviruses, has also come under fire.” Concerns had been voiced, he wrote, about the security of the W.I.V.’s biosafety procedures and facilities.
Evidence regarding the origin of the virus, apart from what could be read from the genome itself, remained scarce during those early months. In place of evidence, there was the weight of scientific authority on one side and the volume of outcry on the other. On Feb. 19, 2020, an open letter appeared online in The Lancet, a British journal, signed by 27 scientists, some of them eminent senior figures in virology and public health, others researchers in the full heat of distinguished careers. It was a statement of solidarity with Chinese scientists and health professionals, who were then on the front line in efforts to understand and control the virus. The letter was organized by Peter Daszak, a British American disease ecologist, president of EcoHealth Alliance and a collaborator with Zhengli Shi. Besides voicing support for Chinese colleagues, it said: “We stand together to strongly condemn conspiracy theories suggesting that Covid-19 does not have a natural origin.” That expression of confidence, so soon, would prove to be counterproductive, and the phrase “conspiracy theories” landed like bacon grease thrown into a campfire, causing skeptics to flare and sizzle.
The lab-leak idea, meanwhile, took hold in some political circles, partly because it dovetailed with attitudes toward the Chinese government, its repressive policies and its penchant for secrecy. In late January 2020, even before Cohen’s Jan. 31 article, The Washington Times ran an article suggesting links between the W.I.V. and a covert bioweapons program of the Chinese military. The article (later walked back with an editor’s note) was based largely on assertions by a former Israeli military-intelligence officer. Several weeks afterward, Senator Tom Cotton of Arkansas voiced a similar suspicion about the Wuhan lab on Fox News. “We don’t have evidence that this disease originated there,” Cotton said, “but because of China’s duplicity and dishonesty from the beginning, we need to at least ask the question.” Soon enough, Donald Trump’s mind began to change. The president spoke supportively about China throughout the early weeks of the pandemic and on Feb. 7 said of President Xi Jinping, “I think he’s handled it really well.” Then the winds shifted, and four months later Trump was inciting his rally crowds by calling Covid-19 “the kung flu.”
The attractions of the lab-leak idea weren’t entirely partisan. Jamie Metzl is an author and political commentator who worked in the Clinton administration and, at one point, as a Senate committee staff member working closely with Senator Joe Biden. Metzl has a blindingly luminescent and liberal-tinged résumé that includes a Ph.D. from Oxford, a J.D. from Harvard Law School, a senior fellowship at the Atlantic Council and 13 ironman triathlons. A former member of the W.H.O. expert advisory committee on human genome editing, Metzl called early on for an investigation into the origins of the pandemic, including, in his words, “the distinct possibility this crisis may stem from a research-related incident in Wuhan.”
Having spoken up about this in the early months of 2020, Metzl encountered resistance that seems to have startled and aggrieved him. “When I was seeing this different story,” he told me, “and I started speaking publicly about it, friends of mine would say two things.” The first was, “You’re a progressive, liberal Democrat” — but — “you’re delivering a message that’s helpful to Trump.” Implication: Metzl should get back on the right side of scrimmage. The second sort of comment, he says, was: “Who the F are you? You have all these senior scientists and Nobel laureates and others who are saying it comes from nature? Who the F are you to say that, based on your analysis and your deductive reasoning, you have additional questions?”
The proselytizing by Metzl and others who saw a “different story” from natural spillover — plus the swing of Trump’s message, plus the prevailing cultural disposition to distrust experts, plus no doubt other factors — had an effect on public opinion and media attention, if not on scientific consensus. According to the Pew Research Center, polling Americans in March 2020, 43 percent believed that the virus emerged naturally, against less than 30 percent who thought it came from a lab, developed either by accident or intentionally. By September 2020, another polling organization found the natural versus lab options embraced almost equally. By June 2021, a Politico-Harvard poll put the lab-origin idea ahead by a two-to-one margin: 52 percent of Americans versus 28 percent.
Metzl himself has maintained the somewhat agnostic position that accidental release is a possibility but not the only possibility. In his eventual March 2023 testimony to Congress, he urged “fully examining all relevant origins hypotheses, obviously including a lab origin, but also a market origin, which some experts I respect believe to be more probable.” Among those experts, he cited Michael Worobey, an evolutionary virologist at the University of Arizona.
Worobey is a Canadian-born, Oxford-educated scientist who speaks mildly and sometimes entertains provocative theories. One such theory was O.P.V., the “oral polio vaccine” hypothesis for the origin of the H.I.V./AIDS pandemic. I first interviewed Worobey a dozen years ago to hear about that. The O.P.V. hypothesis asserted that the virus (H.I.V.-1, Group M) was put into humans, inadvertently, during reckless trials of an oral polio vaccine on unsuspecting African “volunteers,” including hundreds of thousands of children. The vaccine had been developed in chimpanzee cell cultures — so the hypothesis claimed — and contaminated with a chimpanzee virus that became H.I.V.-1-M. In early 2000, Worobey left his doctoral studies in Oxford, flew to a war zone in the Democratic Republic of the Congo and spent weeks collecting chimpanzee dung in the forest to test that hypothesis.
His senior partner on this wildcat expedition was William Hamilton, a famous Oxford biologist who considered the O.P.V. hypothesis plausible. Worobey and Hamilton collected their chimpanzee samples, with help from local forest guides, and then scrambled out of Kisangani, Worobey with his arm in a sling from a badly infected forest wound, Hamilton desperately ill with malaria. They reached England, and Hamilton died soon afterward from complications. The samples got lost in baggage handling, then found, then tested negative for the chimpanzee virus, except for one sample that proved inconclusive.
Such are the labors and frustrations of science. Worobey, along with other scientists, drawing on other evidence, eventually showed that the oral-vaccine hypothesis was false. Open-mindedness toward a provocative hypothesis, and a commitment to confirm it or refute it as the evidence may dictate, are among his priors.
With SARS-CoV-2, 20 years later, Worobey likewise felt inclined to give the provocative, heterodox hypothesis all due consideration. Concerned by what he saw as premature dismissal of the lab-leak possibility, he signed a public letter in spring 2021, with 17 other scientists, arguing that “greater clarity about the origins of this pandemic is necessary and feasible to achieve. We must take hypotheses about both natural and laboratory spillovers seriously until we have sufficient data.” One of the letter’s other co-signers, in fact the first as listed, was Jesse Bloom. Worobey had helped initiate the letter, with emails to Bloom on March 21 of that year, including the suggestion, “I have been thinking about something like a Perspective in Science or an Op-Ed in the NY Times.”
The letter was initially drafted by Bloom and two others: Alina Chan, a molecular biologist who was an author of a preprint in 2020 arguing that SARS-CoV-2 was already well adapted to infecting humans at the start, raising questions about its provenance; and David Relman of Stanford, a distinguished microbiologist with a long-term concern about biosecurity issues and some gain-of-function research. Others of the group contributed input, and the letter ran in Science on May 14, 2021, under the imperative title “Investigate the Origins of Covid-19.” But from that point, with passing months and more research, Worobey would diverge from the most vocal of his co-signers regarding what constitutes “sufficient data.”
Strong tides of opinion were moving by spring 2021. An international team of scientists, recruited by the World Health Organization to its joint W.H.O.-China study of the origins of SARS-CoV-2, had returned from a month in Wuhan and issued its Phase 1 report, finding a laboratory leak “extremely unlikely.” That finding took criticism from Worobey, Bloom and their co-authors of the letter to Science, published weeks afterward. Even the director-general of the W.H.O. himself, Dr. Tedros Adhanom Ghebreyesus, hoped for further investigation. At a news conference marking the report’s publication, Tedros said, “As far as W.H.O. is concerned, all hypotheses remain on the table,” noting the need for continued research. Notwithstanding Tedros’s hopes, and mainly because of Chinese resistance, there has been no official Phase 2 follow-up study per se. Instead, the W.H.O. created a Scientific Advisory Group for the Origins of Novel Pathogens (SAGO), a body of disease scientists who will continue to study the origin of SARS-CoV-2 as well as other dangerous new bugs.
Maria Van Kerkhove, the technical lead for Covid at the W.H.O., has been vocal about the barriers to progress. “There’s very little information that can be accessed with regard to lab leak, with regard to breach of biosafety or biosecurity, and that’s the problem,” she told me recently, saying that she had discussed the issue directly with Chinese officials. “That’s what’s frustrating,” she added. “With that lack of information, you’re left with these gaping holes.”
Popular articles espousing the lab-leak idea also began to blossom forth around this time, in magazines and newspapers and on web platforms. In January 2021, New York magazine carried a Covid-origin article by Nicholson Baker, who had lately published a book on American bioweapons research in the early 1950s and his frustrations with the Freedom of Information Act. Baker now raised the “What if?” question about coronavirus research. In May 2021, Nicholas Wade (who once worked for The New York Times) published a long article in The Bulletin of the Atomic Scientists describing the collaborations between Zhengli Shi’s lab and EcoHealth Alliance, for research on bat coronaviruses as potential threats to human health — research that, Wade suggested, could have led to the escape of a virus intentionally made more dangerous to humans. Soon afterward, another science writer with former connections to The Times, Donald G. McNeil Jr., found himself moved by Wade’s article to do further probing and questioning and posted a more judicious essay, concluding, “All we have so far is speculation, and all the explanations are unsatisfactory.” In early June, Vanity Fair followed with a feature by the journalist Katherine Eban, suggesting that research at the W.I.V. — or, alternatively, field collection of bat samples, and the accidental infection of a fieldworker — might have put the virus, engineered or not, into people.
Then came Jon Stewart. On June 14, 2021, the comedian appeared on Stephen Colbert’s show and announced, with sublime confidence and transcendent shallowness, his grounds for certainty that the virus first detected in Wuhan had come from a Wuhan lab. “If you look at the name!” he shouted. “Look at the name!” Stewart got the name of the institution wrong, in fact — he called W.I.V. the “Wuhan Novel Respiratory Coronavirus Lab” — though he got the name of the city right. How much that mattered to Colbert’s millions of viewers is unknown.
Throughout 2020 and 2021, scientists with deep expertise in relevant fields, especially molecular evolutionary virology, veterinary virology and molecular phylogenetics (the drawing of family trees by comparison of genomes), were busy too. Their efforts added data and analyses to the natural-origin side of the balance.
One study, by two Chinese researchers and three Westerners, showed that the wet markets in Wuhan — not just the notorious Huanan but also three others — contained numerous shops selling wild animals for food from May 2017 to November 2019. The offerings included raccoon dogs, masked palm civets and Malayan porcupines. Many of the animals showed what seemed to be wounds from gunshots or traps, suggesting harvest from the wild (as distinct from farm-raised wildlife), but lacked the requisite documentation to make their sale legal under China’s Wildlife Protection Law.
That’s important because it gave local authorities incentive, as the pandemic spread, to close the market (as they did on Jan. 1, 2020) and conceal whatever illegalities had been ignored by enforcement officers there. For all the assumptions made about China’s motivation to cover up a lab leak, it’s worth remembering that they would have had similar motivations — including that $70 billion national industry — to cover up a devastatingly consequential market leak.
Another study, published by Science in July 2022, with Michael Worobey as first author, joined by Eddie Holmes and Marion Koopmans and many others, considered the spatial pattern of more than 150 of the earliest Covid-19 cases from December 2019. Worobey and colleagues found that not only were Huanan market customers and workers (and people in contact with those customers or workers) among those cases living close to the market, but so were patients with no known epidemiological link to the market. Therefore, that market was “the early epicenter” of the pandemic, as the paper’s title declared.
A distinct but related study that appeared around the same time, with Worobey and other co-authors but in this case with Jonathan Pekar as first author, looked at the shape of the SARS-CoV-2 family tree. It was unexpected. As drawn from comparison of genomes sequenced from human samples, taken at the beginning of the pandemic, it consisted of two thick limbs branching from a trunk, then each limb exploding into many tiny stems, without intermediate branches.
The two major limbs were lineages, labeled Lineage A and Lineage B, from which all the virus’s later diversity arose. Lineage B was the more prolific and successful, accounting for most of the world’s Covid-19 cases, including all early cases directly linked to the market. Lineage A had been found in the market, too, by the Chinese team that swabbed after the place was summarily closed. That smudge of A turned up on a pair of discarded gloves. Lineage A was also detected in two Covid patients living near the market. Pekar and his colleagues did a high-tech analysis of the tree pattern — those two big limbs, then that explosion of twiggy stems from each — and concluded that the virus probably entered humans multiple times. The outbreak of human infections, they judged, most likely had (at least) two separate beginnings.
Why did this finding matter? Two spillovers into people, from a market stall holding infected raccoon dogs, was more parsimonious than two separately infected laboratory workers, carrying their infections independently to the same market. Partly that’s because of geography: The Wuhan Institute of Virology, as Jon Stewart tried to say, is indeed in the city of Wuhan, but it sits on the other side of the Yangtze River, more than seven miles (as the crow flies) from the Huanan market.
Beginning early this year, the popularity of the lab-accident idea, which grew steadily from 2020 through 2022, received several additional boosts. On Feb. 26, The Wall Street Journal reported that the U.S. Department of Energy, one of the organizations assigned earlier by President Biden to study the origin question, offered a new judgment. Previously undecided, the D.O.E.’s intelligence people now concluded, though with “low confidence,” that the pandemic most likely began from a lab leak. The Journal’s reporters had this from “a classified intelligence report,” unavailable for scrutiny by the public but delivered to the White House and “key members of Congress,” who were unidentified.
The next day, CNN’s website posted a follow-up story stating that three sources, also unidentified, had told CNN that the D.O.E. based its shift partly on information about research done at the Wuhan Center for Disease Control and Prevention, another disease-related facility in the city, more than seven miles from the Wuhan Institute of Virology. This caught my attention, because I knew that the Wuhan C.D.C., having recently relocated, was now just a few hundred yards from the Huanan market. It occurred to me, disquietingly, that a virus leak from the center might fit the spatial clustering of early cases around the market, as analyzed by Worobey and his colleagues, in a way that a putative leak from the W.I.V. did not.
But nothing more on that provocative assertion has come from CNN, or from any other news outlet, in the months since. I was reminded that the Wuhan C.D.C. moved to its new location, near the market, only on Dec. 2, 2019; that date, plus the time presumably needed to bring laboratory work back online, might not jibe with a viral outbreak that most likely began in late November. In any case, two different sources with good access to the Chinese research community told me that the Wuhan C.D.C. (as distinct from the national C.D.C. in Beijing) had no coronavirus research program before the pandemic. One of those sources, Jane Qiu, a China-born independent journalist, added that the Wuhan group’s mandate was mainly technical tasks such as disease surveillance, rather than research. (Qiu couldn’t name her own sources due to their potential jeopardy in China.)
Still more recently, in mid-June, came the Substack article I alluded to earlier, claiming that Ben Hu and two others from Zhengli Shi’s lab were “the first people infected by the virus” and therefore the starting point of the pandemic. Posted by Michael Shellenberger and two co-authors, this article cited unidentified sources “within the U.S. government.” Hu was a first author on a 2017 paper describing the Shi group’s discovery of multiple coronaviruses related to SARS-CoV-2, in bats from a cave (not the Mojiang mine) in southern China, and experimental work on three of those viruses that some critics considered risky. Hu and the other two scientists, according to Shellenberger and his co-authors, had contracted “Covid-19-like illnesses” in November 2019, suggesting that they were the conduits of a lab leak. Hu himself promptly denied the allegation in an email to Jon Cohen at Science: “I did not get sick in autumn 2019, and did not have Covid-19-like symptoms at that time.” Furthermore, Hu told Cohen, he and both colleagues had tested negative for signs of recent Covid infection (antibodies) in March 2020.
The surge of opinion toward the lab-leak idea was interrupted in March, when Florence Débarre, a scientist working for France’s National Center for Scientific Research, discovered another body of interesting evidence, long missing but now found. This was genomic data — from the swab sampling of door surfaces and equipment and other items, including that pair of discarded gloves — gathered at the Huanan market in early 2020 but withheld since that time. The data were released, perhaps by mistake, and Débarre was alert enough to spot them and recognize what they were. A team of researchers, including Worobey, detected a pattern in the data: strong proximity between samples containing raccoon dog DNA and others containing SARS-CoV-2 fragments (and some samples that contained both), from stalls in the southwest corner of the market where wild animals had been sold as food. Malayan porcupine DNA and Amur hedgehog DNA were also found near the virus, but raccoon dogs were of special interest because of their proven susceptibility to SARS-CoV-2.
These findings didn’t establish that raccoon dogs had carried the virus into the market. But they added plausibility and detail to that scenario.
Notwithstanding the Débarre group’s revelations, the lab-leak idea has remained strongly preferred by public opinion, and not just in the United States. According to one poll, as of April 2023, 62 percent of Italian respondents, 56 percent in France and 50 percent in the United Kingdom found the lab-leak idea most compelling, with sizable segments of undecided (and flummoxed) people, leaving only modest minorities embracing natural spillover. Earlier polling showed lab-related scenarios even more strongly favored in still other countries, ranging from 73 percent in Kenya and 64 percent in Hungary to 58 percent in Brazil.
Various factors may account for this public drift to the lab-leak hypothesis. In my view, a preponderance of empirical evidence is not one of them. I agree it’s important to remain open-minded toward a lab-leak possibility, but most of the arguments made in support of that possibility boil down to conjecture from circumstance and unsupported accusations.
To speak of a “lab-leak hypothesis” in the singular is, of course, misleading. There are multiple lab-leak hypotheses, just as there are multiple ways a natural spillover could have occurred. A more encompassing and emollient phrase is “research-related incident,” preferred by Jamie Metzl and some other critics. That covers several possibilities, including the chance that misbegotten gain-of-function research, at the W.I.V. or the Wuhan C.D.C. or who knows where, yielded a dangerous new hybrid virus that escaped through a malfunctioning autoclave or an infected technician or grad student. (In support of this scenario, proponents point to a grant proposal known as DEFUSE — made by EcoHealth Alliance to a U.S. defense research agency in 2018, though never funded — for experiments that some critics construe as potentially dangerous gain-of-function research.) Another “research-related” possibility: the nightmare that some Chinese biowarfare program created a murderous virus intentionally but let it escape to the world by some catastrophic goof. Still another: the notion that a scientific fieldworker became infected while taking samples from bats in, say, the Mojiang mine, where Zhengli Shi’s team found RaTG13.
They’re all vivid but not all logical, and it seems to me they don’t reinforce one another. If a wild coronavirus from the Mojiang mine was capable of infecting and transmitting among humans, for instance, then it didn’t need a furin cleavage site to be inserted during reckless or malevolent lab work. And if it infected a scientific fieldworker in 2013, and that fieldworker returned to Wuhan, where did the virus linger for six years before exploding through the city’s population in 2019? And if the virus was engineered in Shi’s lab, using sophisticated gene-editing methods, or was transformed into this dangerous pathogen by passaging a less dangerous virus through cell cultures or living mice (which seems far-fetched), and then subsequently escaped, then the Mojiang mine, with all its sinister narrative appeal, is irrelevant. In other words, the various research-related-incident hypotheses may each be plausible (some more than others), but they compete against one another. You can’t pile them all on the scale and judge the likelihood of an unnatural origin by their combined weight.
Lab-leak partisans have focused intently on Shi and her lab, but it’s important to bear in mind that Shi has made her career by publishing research and issuing warnings about potentially dangerous coronaviruses found in the wild, not by keeping them secret. If she had such a formidably dangerous virus in her lab in 2018 or 2019 — a virus similar to the original SARS virus but with a receptor-binding domain and a furin cleavage site well shaped for human infection, features that could make it even more dangerous — she presumably would have announced that important discovery from the pages of a leading journal, to her professional gain as well as the benefit of the world. She didn’t.
And there’s a small body of lost evidence, recently recovered, that seems to support this logic. In 2018, a scientist named Jie Cui led a study of SARS-related coronaviruses in bats. His purpose was to illuminate the evolution of the original SARS virus by placing it on a family tree of its relatives. Cui had been a postdoctoral fellow in Eddie Holmes’s lab, going from there to the Wuhan Institute of Virology for a couple of years, then to a position in Shanghai. Cui and a group of colleagues, including both Holmes and Zhengli Shi, analyzed partial genome sequences of 60 coronaviruses detected in bat samples collected from 2011 to 2016. They wrote a paper and submitted it to a leading virology journal. It was rejected. They tried another. Rejected. The journals’ reviewers wanted complete genome sequences, but the team had only partials. So in October 2018, they gave up on that paper. They pulled it from the submission process. They forgot about it. In the meantime, they had submitted their partial but telling genome data to an international database, GenBank, with a routine stipulation that it would be embargoed, in this case for four years. The embargo allowed them to retain exclusive access to the data for that period, in the event they wanted to revive the project.
Four years passed, and then, in October 2022, the embargo expired. The data, mothballed since just before the pandemic and now publicly available, were revealing for what they did not include: a progenitor of the pandemic virus. Here were 60 coronaviruses that Zhengli Shi and others had considered intriguing in 2018. But nothing that matched SARS-CoV-2.
“Where’s the virus?” said Eddie Holmes, recounting this to me recently. “The virus is absolutely not there.”
Two other arguments on the lab-leak side deserve attention. Each can be phrased as a question. Why did SARS-CoV-2, from the start, seem to be very well adapted to humans? And why, if its natural host was some kind of bat, has that host not yet been found, after three and a half years?
The first of those questions ignores the fact that SARS-CoV-2 has shown itself, from the beginning, to be quite capable of infecting other mammals (cats and dogs), and eventually a wide range of them (tigers, gorillas, mink, white-tailed deer and others), not just humans. The second question betrays a lack of familiarity with the history of emerging viruses. When a novel virus appears suddenly in humans, causing disease and alarm, the search for its natural host is always an urgent task. But such ecological work is difficult to do amid the public-health emergency of an outbreak, and once the outbreak (or epidemic, or pandemic) is controlled, the sense of urgency and the available research money tend to disappear.
Finding the host animal is sometimes easy, by luck, and sometimes hard. Identifying horseshoe bats, with high confidence, as the likely hosts of the original SARS virus took 15 years. Tracing the Marburg virus to its reservoir host in Egyptian fruit bats took 41 years (or 42, if you count the time to publication). And the natural host of Ebola virus, despite what you may think you’ve heard, is still unidentified, 47 years after its emergence at a remote mission hospital in what was then Zaire. The suggested linkage between Ebola virus and some form of bat is still suppositional, not settled scientific fact — and we have enough suppositions entangled with this subject already.
So, what’s tilting the scales of popular opinion toward lab leak? The answer to that is not embedded deeply in the arcane data I’ve been skimming through here. What’s tilting the scales, it seems to me, is cynicism and narrative appeal.
I asked about this in conversation with David Relman, the biosecurity expert who was also an author of the “Investigate” letter with Jesse Bloom. To some extent, Relman agreed. “When you sow the seeds of distrust, or suggest that you haven’t been transparent with what you knew,” he told me, “you’re setting yourself up for a persistent, insidious, continuing distrust.” That inclines people to assume that “there was something deliberate, or deliberately concealed.”
The seeds of distrust have been growing in America’s civic garden, and the world’s, for a long time. More than 60 percent of Americans, according to polling within the past several years, still decline to believe that Lee Harvey Oswald, acting alone, killed John F. Kennedy. Is that because people have read the Warren Commission report, found it unpersuasive and minutely scrutinized the “magic bullet” theory? No, it’s because they have learned to be distrustful, and because a conspiracy theory of any big event is more dramatic and satisfying than a small, stupid explanation, like the notion that a feckless loser could kill a president by hitting two out of three shots with a $13 rifle.
Most of us don’t reach our opinions by fastidious calibration of empirical evidence. We default to our priors, as Jesse Bloom noted, or we embrace stories that have simple plots, good and bad characters and melodramatic trajectories, and that seem commensurate in scope to the event in question. The process of scientific discovery is a complicated story involving data collection, hypothesis testing, hypothesis falsification, hypothesis revision, further testing and brilliant but fallible humans doing all that work. Scientific malfeasance driven by hubris and leading to runaway trouble, on the other hand, is a much simpler story that goes back at least to Mary Shelley’s 1818 novel, “Frankenstein.”
Carl Bergstrom is an evolutionary biologist and an author of commentaries on scientific misinformation. He ponders, among other things, how students of science are taught — or at least should be taught — about not just what science says but what science is. I asked Bergstrom about the human affinity for dark theories of big events.
There was something about that in Thomas Hardy, he told me. “It’s in ‘Tess of the d’Urbervilles,’ where Tess is doomed by hapless chance. It really sucks! To live in a world where we are at the mercy of hapless chance.”
I had never read “Tess of the d’Urbervilles,” to my embarrassment, so I stuck with SARS-CoV-2. “This is not a contest now, in the public domain, between bodies of evidence,” I proposed. “This is a contest between stories.”
“Yeah!” Bergstrom said. “That’s right.”
David Quammen is an author and a journalist based in Montana. His 18 books include “Spillover” (2012) and “Breathless” (2022): the first, predicting a pandemic that would be caused by a newly emerged virus; the second, tracing the scientific effort to discover the origin and track the evolution of Covid.