WASHINGTON — In March 2007, federal agents convened an elite group of outside experts to evaluate the science that had traced the anthrax in the letters to a single flask at an Army lab in Maryland.
Laboratory work had built the heart of the case prosecutors had against Bruce Ivins, an Army researcher who controlled the flask. Investigators had invented a new form of genetic fingerprinting for the case, testing anthrax collected from U.S. and foreign labs for mutations detected in the attack powder.
Out of more than 1,000 samples, only eight had tested positive for four mutations found in the deadly germs sent to Congress and the news media.
Even so, the outside scientists, known as the Red Team, urged the FBI to do more basic research into how and when the mutations arose to make sure the tests were "sound" and the results unchallengeable.
Jenifer Smith, a senior manager at the FBI's laboratory, shared the team's concerns. Smith recalled that she was worried the FBI didn't have a full understanding of the mutations and might see a trial judge throw out the key evidence.
"The admissibility hearing would have been very difficult," Smith recalled in an interview. "They had some good science but they also had some holes that would have been very difficult to fill."
The FBI rebuffed the Red Team's suggestion, describing it as "an academic question with little probative value to the investigation."
Ivins committed suicide in July of 2008 as prosecutors were preparing to charge him with capital murder in the cases of the five people killed by the anthrax mailings. Prosecutors announced that Ivins was the sole perpetrator and the parent material for the letters had come from his flask.
Three years later, that assertion remains an open question. A separate panel, from the National Academy of Sciences, found that prosecutors had overstated the certainty of their finding. Committee members said newly available testing methods could prove the FBI's case much more definitively or lead to other potential suspects. But federal investigators, who closed the case more than a year ago, have expressed no interest in further scientific study of the evidence.
A re-examination of the anthrax case by "Frontline," McClatchy and ProPublica has raised new questions about some of the evidence against Ivins. The reporting uncovered previously undisclosed tensions between researchers who were trying to create a new form of forensic science and criminal investigators whose boss was under intense pressure from the president of the United States to crack a case that had few leads and hundreds of plausible suspects.
Paul Keim, an anthrax expert at Northern Arizona University who assisted in the FBI investigation, said he had qualms about whether the bureau's groundbreaking laboratory method would have survived a rigorous legal review.
"I don't think that it was ready for the courtroom at the time Bruce committed suicide," Keim said.
If Ivins hadn't killed himself, he said, the FBI would have launched a "hard push" for additional data that showed the methods were reliable. Such research, he said, also could have shown it wasn't valid.
Keim, a member of the Red Team who attended the March 2007 meeting in Quantico, Va., said he didn't find out that its call for further research had been rejected until a year later, after Ivins had committed suicide and prosecutors were hastily organizing a news conference to describe the science.
Keim and other scientists involved in the case said the strictures of a criminal investigation prevented them from talking to one another or sharing information as they would on a typical research effort.
"The investigation was being driven by a small group of bureau scientists and investigators," Keim said. "Broader engagement with an expert panel sworn to secrecy would have been good. Having the best scientific consultants embedded would have been good."
Jenifer Smith was the section chief of the Intelligence and Analysis Section in the Weapons of Mass Destruction Directorate of the FBI until 2009, and she observed the process from inside. During the anthrax case, she said, the FBI lab departed from its traditional procedures and allowed top investigators to influence how the science was conducted.
"They deviated from traditional lab practice in this particular case," Smith said. "There were some political things going on behind the scenes, and it was embarrassing not to have this solved. Yes, it was a long, drawn-out investigation. But that's when you don't deviate from your practices."
Rachel Lieber, the lead prosecutor, said law enforcement officials did try to make sure the science was rigorously vetted. But Lieber said there were limits and that the science was only a piece of a much larger mosaic of evidence against Ivins.
"You look at the lines of a trial and where do we spend our resources," Lieber said. "Are we doing a science project or are we looking for proof at trial? These are two very different standards."
Questions about the definitiveness of the scientific findings began to arise soon after prosecutors said Ivins was the anthrax mailer. U.S. Attorney Jeffrey Taylor declared without equivocation in August 2008 that the FBI had proved that Ivins' flask, RMR-1029, "was the parent flask for the spores" used in the mailings — "effectively the murder weapon."
At a briefing a few weeks later, the FBI's lab director, Christian Hassell, was asked to give a "level of confidence" for the findings. "It's very high," he replied. "This whole exercise shows that they were traced back to a single flask."
Another official at the briefing, who spoke on condition of anonymity as a ground rule, claimed that the Red Team of outside experts had vetted and approved the work and the FBI had heeded its calls for further research.
"We invited a cadre of scientists to conduct a Red Team review of the science that we performed, and we took their suggestions," the official said. "We made additional experiments and the data available to the Red Team at their suggestion. And so all of the science that went behind this was well-reviewed."
Critics on Capitol Hill weren't mollified. Sen. Patrick Leahy, a Vermont Democrat and one of the intended recipients of an anthrax letter, said he didn't believe that Ivins had acted alone. So FBI director Robert Mueller announced in September 2008 that a panel of the National Academy of Sciences would conduct an independent review of the scientific findings.
The FBI didn't wait for the outcome of those deliberations. In February 2010, with the panel still taking testimony, prosecutors announced that the case was closed, Ivins was the sole perpetrator and RMR-1029 was "conclusively identified as the parent material to the anthrax powder used in the mailings."
Nearly a year to the day later, the outside committee looked at that same evidence differently, saying scientific data "provided leads" as to where the spores had come from, but "alone did not rule out other sources."
The tests that identified the flask as the source of the parent spores for the attack showed an association but didn't "definitively demonstrate such a relationship," the panel said.
When the case of the anthrax mailings broke in late 2001, genetic sequencing was in its early days. It was only a year earlier that two teams of American scientists had announced that they'd mapped the human genome. Work on anthrax was under way but not complete.
A fortuitous lab miscue may have provided the biggest break. Early in the investigation, a colleague of Ivins' at the Army research facility in Fort Detrick, Md., left spores from one of the letters growing in a dish longer than she'd planned. When she next looked, the researcher, Terry Abshire, noticed something potentially significant: The cultures had small numbers of visually distinguishable colonies, or morphs, which are caused by genetic mutations.
One of the leading scientists in the field describes morphs this way: Imagine a jar of M&M's that are almost all chocolate — the anthrax spores — but sprinkled with a few blues and reds — the morphs. Scoop out a big enough portion of spores, grow the anthrax on plates and a morph or two probably will appear.
Morphs that look the same can be caused by different mutations. To create a true genetic fingerprint of the attack powder, scientists had to find and sequence the mutations that spawned the morphs.
Once they had that fingerprint, investigators could look for a match with the stocks of anthrax in bio-defense labs. The Ames strain used in the letter attacks had been cultured from a dead cow in Texas in 1981 and the Army frequently used it for animal testing of vaccines.
The attack powder had multiple morphs. Over several years, contractors developed tests for four morphs that produced what the FBI said were reliable results.
It was no easy task. Just a year before the anthrax attacks, it would have been inconceivable, Keim said. By late 2001, it was merely revolutionary.
When the anthrax samples from U.S. and foreign bio-weapons labs were screened, only 10 out of more than 1,000 tested positive for three or more of the morphs. All came from Ivins' flask, RMR-1029.
In its report earlier this year, the National Academy of Sciences panel raised numerous questions about this finding, some of which had been posed by the Red Team back in 2007.
The panel also noted that the FBI didn't have a complete understanding of how temperature or other growth conditions might affect how many morphs appeared. Perhaps a particular morph would become visible only under certain growth conditions. It was this question that the Red Team had suggested for additional research in 2007.
The FBI's own records show that the tests didn't always deliver reliable results. In trying to prove that a sample Ivins provided from his flask in April 2002 was deceptive because it contained none of the morphs from the attack powder, investigators sampled the flask 30 times. All came back with at least one morph and 16 came back with all four. Six of them showed only two or fewer, even though they were grown directly from the Ivins culture.
The National Academy of Sciences panel's report raised the possibility that some of the morphs could arise through the process of "parallel evolution," in which identical mutations occur in separately growing colonies of bacteria.
Richard Lenski, a Michigan State University professor and specialist in this field, said the FBI faced a significant challenge. In what he termed a "dream world" with unlimited resources, researchers could separately grow hundreds of colonies of the anthrax in both Ivins' flask and the attack powder and compare the morphs that evolve.
Such an experiment, of course, would be an expensive long shot, because the person who made the powder did so in secret with methods that remain unknown.
Claire Fraser-Liggett, a pioneering genetics researcher, was part of the effort to track the morphs. In August 2008, she sat onstage alongside senior FBI science officials as the findings were presented. By then, she said, the technology had leapfrogged far beyond the techniques used in 2001 and 2002 to compare morphs.
Beginning in about 2006, she said, "next generation" sequencing came on line. What cost $250,000 and took one to two months of work in 2001 now could be done for about $150 in a week. With price no longer an issue, the DNA in a colony of anthrax could be sampled over and over and over, assuring accuracy. And you wouldn't be limited to four morphs; you could test as many as you felt were useful.
Fraser-Liggett doesn't fault the FBI for not switching to an untried technique in 2006. But she said the enhanced precision now available could bring new evidence to light, confirming investigators' original conclusion or pointing in other directions.
For now, the case remains a matter of dispute, with prosecutors and law enforcement officials insisting that the combination of science and circumstantial evidence would have been more than sufficient to win a conviction.
One area of contention revolves around whether the killer tried to add a chemical to make the spores float more easily, so they'd have a better chance of being inhaled.
McClatchy first reported last spring that the FBI had failed to explain the presence of unusual levels of silicon and tin in two of the letters, since those elements arent part of the process of growing spores. Scientists pressing for answers to those questions published a paper in a scientific journal this month. The FBI says the silicon was present through a natural process, not from any special treatment by Ivins.
David A. Relman, the vice chairman of the National Academy study committee and a professor at the Stanford University School of Medicine, said the scientific picture remained incomplete.
Relman said, for example, that the high level of silicon measured in the letter sent to the New York Post remained a "big discrepancy," one for which the panel received no explanation. None of the spores in Ivins' now-infamous flask contained any silicon.
Relman said the panel questioned FBI officials about whether the high silicon measurement had arisen from an anomaly in the testing.
"We asked: Is it nonrepresentative sampling?" he said.
"And they said, 'No, we don't think that's the answer.'"
"There is no answer," Relman said, "That's why we said it's not resolved."
Lieber, the prosecutor, said she would have moved to exclude the high silicon reading at trial since it came from a single measurement.
According to the FBI, the anthrax from RMR-1029 was used as starter germs for the attack powder. Investigators suspect that there were at least two separate production runs. The powder sent to the media was relatively coarse; the anthrax mailed to Congress had much smaller particles and floated like a gas. The germs in the flask were a mix of anthrax grown by Ivins and the Army base in Dugway, Utah. During the growth phase, the anthrax killer introduced several impurities, including silicon, trace elements of tin and, in two of the letters, an unusual strain of another bacterium, Bacillus subtilis.
Had Ivins not committed suicide, his trial very likely would have included vigorous sparring over the scientific evidence. Experts probably would have offered conflicting testimony about the reliability and certainty of the genetic tests.
Relman said a trial would have been the only way in which the prosecution's evidence "could have been weighed and challenged by experts."
It's worth considering whether an independent panel should evaluate the full case against Ivins, looking at the science, the evidence investigators gathered and all other relevant material, Relman said.
"We have to decide how important this case is to us as a society, and I'm not presupposing it is," he said.
Fraser-Liggett and Lenski said it would be valuable to continue testing the anthrax samples in the case as new, more sensitive technologies came on line.
"Speaking as an individual citizen," Lenski wrote in an email, "I think it would benefit the public. Even if it didn't resolve the Amerithrax case with respect to criminal culpability, it would be a valuable test run of what science could contribute if a similar terrorist event were to occur."
(McClatchy collaborated with the investigative newsroom ProPublica and PBS' "Frontline" to produce this series. Gordon works for McClatchy. Engelberg and Matsumoto work for ProPublica and Wiser and Jim Gilmore, who contributed reporting for this story, are with "Frontline.")