UPDATE: This Washington Post article shows that the Leahy and Daschle letters contained more material than is assumed in this post as it originally appeared. The post improperly assumes that a 0.013 g sample of spores removed for analysis was the entire amount of anthrax material, while the Washington Post article states:

The Daschle and Leahy letters each contained 1.5 grams of anthrax powder or less

As a result, the 0.013 g sample would represent about 1% of the anthrax material in the letter, and this accounts for the 100 fold difference in my calculations.

I apologize for this error and the erroneous conclusions I stated as a result of the error.

FBI photo of the flask labeled RM R-1029

Friday afternoon, the FBI released a number of documents in closing the case dubbed "Amerithrax". Substantial flaws still remain in the FBI’s explanation of the technical analysis on which they concluded that Bruce Ivins was the sole perpetrator of the anthrax attacks of 2001.

The documents can be found here. Only the 93 page investigative summary and the supporting exhibits are new, and all can be accessed from the Amerithrax page as pdf files.

On Friday, I briefly discussed the new information derived from the summary relating to the genetic analyses carried out to pinpoint Bruce Ivins’ infamous flask labeled RMR-1029 as the sole source of the material used in the attacks. Significant holes still remain in the FBI’s explanation of the genetic analyses. It is possible that many of these shortcomings could be addressed if the actual reports from the scientists who carried out the work were released. It is my hope that these reports now will be cleared for publishing in peer-reviewed journals so that the evidence can be scrutinized fully.

In looking further at the investigative summary, I have found what appears to be an error in the analysis of how much material from RMR-1029 would have been required to produce the spores used in the attack letters. The result of this error is an overestimate, by a factor of 100, of how much material from RMR-1029 would have been needed to be used for each letter. Partially because of this overestimate, the FBI excluded as suspects other researchers who received samples from RMR-1029, claiming that they lacked the expertise both to produce such a large volume of material and to then prepare it as attack material. With the smaller estimate, most of the basis for excluding these individuals goes away, as simple procedures could be used to dry such a small amount of material.

Exhibit F released on Friday is a pdf file of a page from Bruce Ivins’ notebook dated March 17, 1998. This page has the results of Ivins’ analysis of the material labeled RMR-1029. At this point in time, Ivins had approximately one liter of material distributed between two one liter flasks. The photo above is the one flask that remained after the two had been combined due to using a significant amount of material in Ivins’ research developing anthrax vaccines.

Here is the crucial part of the notebook entry:
Ivins notebook page

In doing his microscopic analysis, Ivins states clearly that he is working with a 100-fold (or, 1:100) dilution of material from the RMR-1029 flask. He also states that this dilution is at an approximate concentration of 3 X 108 spores per mL. From the information present on this page of the notebook, it is clear that the concentration of spores in RMR-1029 is approximately 3 X 1010 per mL.

From Exhibit G, this is Ivins’ analysis of the material he recovered from one of the attack envelopes on October 17, 2001 when the FBI asked him to analyze it. From the date and purity, this is most likely the Leahy envelope:

envelope analysis

Here, we see that Ivins recovered 0.013 grams of powder from the envelope. He suspended this powder in water and then plated it out to determine the concentration of bacteria. He then computed a concentration of 2.1 X 1012 colony forming units per gram of powder. For spores that are perfectly viable, one spore corresponds to one colony forming unit. That means that 0.013 g of the powder contains 2.7 X 1010 spores.

On page 29 of the investigative summary, footnote 15 states this:

A leading anthrax researcher who assisted the investigation expressed his expert opinion that 100 ml would have been required to create sufficient material to be used in one letter, for a total of 500 ml for the five letters. Nonetheless, we cannot say with certainty how much material was used in the letters.

One hundred mL of RMR-1029 would be 3 X 1012 spores, 100-fold more than Ivins recovered from the envelope he analyzed. The only way the opinion of the anthrax researcher makes sense is if they mistakenly took Ivins’ 3 X 108 notation in the notebook as the concentration of spores in RMR-1029, when Ivins clearly states that is the concentration of the diluted material he analyzed.

The lower concentration makes no sense as the spore concentration of RMR-1029 for several reasons. First, the description of how many large cultures were produced at Dugway and small cultures in Ivins’ lab to produce RMR-1029 would suggest that the purification process resulted in the loss of most of the spores produced, if the lower concentration of RMR-1029 is correct. In other words, the lower concentration for RMR-1029 would mean that the final concentration of RMR-1029 was approximately at or below the concentration of spores one achieves in a standard bacterial culture, even though over a hundred liters of culture were used to produce the one liter purified material in the RMR-1029 flask.

An alternate explanation for the discrepancy would be if Ivins collected only one percent of the material in the envelope for his analysis, but that would mean that there was so much material in the envelope that it would appear overly stuffed. For example, take a look at this photo which shows a 10 milligram capsule of material opened and emptied. Ivins recovered 13 milligrams of powder from the envelope. If there were 100 times that amount, it would be 1.3 grams of powder, or the equivalent of 130 capsules like the one in the photo. That would be a huge amount of powder to stuff into an envelope along with the letter.

The bottom line, then, is that only one mL, not 100 mL of RMR-1029 would be required to produce the material in one envelope. On page 30 of the investigative summary, we have this:

Drying the spores likely would have attracted attention unless the perpetrator accessed the equipment at night. Drying anthrax spores requires either a sophisticated drying machine called a lyophilizer, a speed-vac, or a great deal of time and space to let the spores air-dry – that is, to allow the water to evaporate – in the lab. Because drying anthrax is expressly forbidden by various treaties, overt use of any of these methods, if noticed, would have raised considerable alarm and scrutiny.

I can see that statement being accurate for someone drying 100 mL of RMR-1029 five times (or 500 mL once), but most of the concerns about equipment and space go away if only 5 mL needs to be dried to produce the attack material without a need to grow and purify a large new culture using the RMR-1029 material as inoculum. Rather than a lyophilizer, simple vacuum filtration or air drying could be used on such a small amount of material, and the procedure could be carried out without attracting much attention.

From the bottom of page 33 through the bottom of page 38, the investigative summary describes how the hundreds of people with potential access to material from the RMR-1029 flask were excluded as suspects. However, in making these exclusions, the summary relies heavily on the assumption that 100 mL of RMR-1029 itself (or presumably, the equivalent amount from a subculture followed by purification) would have been required to produce the material for one attack envelope and that these personnel could be excluded because they did not have the requisite expertise or access to the requisite equipment. On the other hand, most of those exclusions would seem to be invalid if a suspect merely had to remove only 5 mL of a larger sample and dry it to produce all of the material used in the attacks.

It appears to me that the FBI has excluded hundreds of potential suspects on the basis of a math error.

[For some of my previous posts on Amerithrax, click on this archive of my old blog and scroll down.]