He's run these critters through THOUSANDS of generations ~ for the most part there have been no mutations whatsoever in almost all of his samples.
Here we have the FBI claiming that forensic science is so good they are able to separately identify anthrax cultivars (in different jars/flasks) due to random mutations ~ and yet random mutations at that rate simply do not occur in the more highly controlled, longer duration experiment with the e-coli.
Let me suggest that the FBI's claim regarding rates of random mutation in bacteria are highly suspect.
Most of the "mutations" that the FBI are looking for are probably so minor that they wouldn't show up unless a complete DNA sequence was performed (something like swapping a few base pairs). I'm not validating their science, but it is in the realm of the feasible. On the other hand, the FBI would, of course, have the burden of proof on showing that the "mutation" only shows up in the samples from the attacks as well as in the flask sample, and shows up no where else (since the sample in the flask checked out to Ivin was inevitably produced from a parent culture elsewhere, that entire parent sample would have to be checked to verify that the mutation didn't occur earlier. Since that task is beyond our current ability to sequence, I'm starting to think that this too is just circumstantial "evidence". Random samples (a few thousand sequenced random samples, all of which would have to show consistent results) of the parent source could, however, provide a decent statistical check. But it could never be proof.
In fact, if the FBI was so inclined, they could obtain samples and sequence all three (progenitor, Ivin's sample, and the attack samples) and by determining the ratio of mutated to non-mutated, they could find how "distant" one sample was from the other. This would allow them to know if the earlier "crude" samples were merely refined to produce the later samples, or if another batch was "brewed up" instead. If they were all from one sample, then the ratios of swapped base pair mutations (assuming there are more than one, which there probably is) would be very close to identical. This would be very difficult to do without either a lot of automation or a lot of hard work. If the FBI actually had this kind of evidence, it would be easy for them to produce it (consisting of a LOT of data for the several thousands of sequences that would be required for any level of accuracy and code snippets of the comparison algorithms used).
They could also isolate the separate strains, and determine if their growth rate is identical or not. If they aren't, then they could even determine how many generations they are separated (i.e., how many "batches" the culprits made). A really clever terrorist could theoretically overcome this by brewing up a batch from a single, carefully selected anthrax bacterium, but I can't really see someone of the terrorist mindset going to that much trouble (since it would extend their growth time by a considerable amount while actually increasing the "uniqueness" of their product. It would be like making sure your fingerprint was on the delivery mechanism and clearly recoverable).