Back in 2009, I noted that a report that USAMRID had lost track of its vials of anthrax sort of undermined the entire FBI case against Bruce Ivins.

One key to the FBI case against Ivins, after all, is that he had complete control over the sole flask that contained the strain of anthrax used in the attack. But now we come to find out that, more than six months after his death, they still don’t have a sound inventory of what they have where?

Well, as this important long Wired article on the FBI’s growing doubts about their case reveals, 2009 was not the first time USAMRID realized they didn’t have an adequate inventory of their anthrax. Discovering they had missed some samples is actually how they discovered the Ivins strain they claimed had been the source of the attack anthrax.

In December 2003, while conducting an inventory of one of USAMRIIDs biocontainment suites, investigators discovered 22 undocumented Ames anthrax samples. They began to fear that the repository they had spent nearly two years assembling might have gaping holes in it. So for the first time, the FBI decided to scour USAMRIID for any vials they had missed.

The institute staff fumed at the search—ongoing experiments would be disrupted, they shouted. Heine, Ivins’ coworker, decided to exact a bit of revenge on his FBI handler. While the agent was collecting samples in his lab—dressed in full protective gear—Heine handed her a vial and told her it was a deadly plague strain. The vial started shaking in the agent’s gloved hand. Heine cracked up. “They were entirely dependent on me to identify everything in every box,” he says. “I could’ve held up a critical piece of evidence, said it was something else, and put it aside. There’s no way they would’ve known.”

During the search, investigators took Ivins’ primary RMR-1029 store—not just a sample of the stuff, all of it. They skimmed a small amount into a vial, labeled it with an identification number, and sent it to Pat Worsham down the hall for analysis.

Now, it appears that investigators decided to focus on Ivins because 1) he had withheld the RMR-1029 in the past, and 2) he had concerning tendencies.

(And, probably, 3) their case against Hatfill was falling apart.)

But what Shachtman doesn’t explain is what happened to the other 22 vials they had missed … at USAMRID. Plus the ones (such as, at Dugway, which would be a more likely laboratory to have produced this anthrax) not declared elsewhere?

In other words, no matter how good the science was analyzing the specimens of anthrax they got, there’s abundant evidence that they didn’t do a comprehensive inventory in the early days of the investigation (at which point, legally, it was probably too late to apply this kind of analysis), and they can’t guarantee that the labs have an accurate inventory of their anthrax, much less that that anthrax all stayed in the official labs.

As one source for the story says,

“It would’ve been very easy to take the anthrax out, to steal some,” a former USAMRIID officer says. “Anybody could do that.”

So when they did analysis like this:

But of the 1,059 viable samples in the FBI’s Ames anthrax repository, eight regularly produced all of the mutants. One of those eight was Ivins’ RMR-1029 flask. The other seven were its subsamples. This ruled out Hatfill, who did not have access to RMR-1029 during his time at USAMRIID. (Later, the Justice Department agreed to pay Hatfill a $5.8 million settlement and issued an official letter exonerating him. Condè9 Nast also agreed to an undisclosed settlement. The New York Times case was dismissed.) And while dozens of other scientists did have access to the RMR-1029 subsamples, they were being slowly crossed off the list. As each alibi and exculpatory story checked out, the investigators gravitated closer to Ivins.

They weren’t necessarily starting from a valid initial list of suspects.

It’s a problem Wired’s article–and the scientists who did the analysis–admit.

But the National Research Council found that the FBI’s collection can’t be fully trusted: Too many of the samples were intermingled or descended from other labs’ anthracis to provide a truly representative cross-section of Ames anthrax. This may also be a reason why nearly one in 10 samples in the repository tested positive for at least one mutant.Paul Keim, who helped assemble the FBI’s Ames collection, still wonders how much to trust an anthrax repository that relied on scientists (and potential murder suspects) submitting their own samples. “We don’t know if people did it correctly, and there’s no real way to control for that,” Keim says.

Even if everyone was aboveboard, it’s unclear whether the FBI accounted for every last anthrax sample. Each time Ivins gave his colleague Hank Heine a batch of spores for an experiment, for example, Heine would save a milliliter or two, in case the experiment went wrong. “It’s just good scientific practice,” Heine says. “I had numerous samples of RMR-1029.” It’s hard to imagine he was the only scientist with such a collection. Because the subsamples were so small and largely undocumented, it took the FBI nearly three years to stock its repository—plenty of time for a researcher to dispose of an incriminating batch.

Which is why I think that–for all the value in this article–Wired is too creduluous.

But despite all these flaws, the circumstantial evidence remains compelling. It could just be a coincidence that the killer spores were ultimately traced back to a single parent flask and that this flask just happened to be overseen by a depressed scientist with occasional violent fantasies. It could just be a coincidence that this same scientist screwed up his anthrax submission to the FBI—even though he helped develop the submission protocols. It could just be a coincidence that his after-hours work spiked right before the mailings. But put all of those coincidences together and something stronger than happenstance emerges. For the Justice Department, it’s enough to prove Ivins was the anthrax mailer.

Put it this way. There’s one question–who made the anthrax. And the NAS has said only that it is likely tied to Ivins’ flask, but that it is at least one generation removed from that, and that they don’t know that the suspicious days would have provided nearly enough time to make the anthrax (not to mention the fact that my impression is that the FBI only showed that Ivins was spending a lot of time at the lab in their proposed production nights, not that his time there had spiked over time).

But that is entirely independent of the question of who stuck it in an envelope and mailed it to some explicitly political targets and some well-chosen media types (plus Judy Miller and her fake stash).

There is an entirely plausible scenario in which Ivins realized, because he was cooperating so closely with the FBI and because he was telling them to do the right thing, that his anthrax was a likely strain (though, as the NAS points out, that’s only one of the strains used in the attack–it doesn’t account for the journalists’ attacks). But Ivins’ behavior–particularly for a weird socially maladjusted science type–is as easily explained by a panic because he had no explanation for what happened. Or, alternately, he could have been covering for people who ordered him to give them a sample.

There are a whole lot of possibilities. But one thing’s clear. The FBI used faulty investigative work to equate biological evidence developed under evidentiarily inadequate conditions with guilt for the crime itself. And that’s really not what we’re paying the FBI to do.