Why doesn't everybody get the flu at once? Why do some drug addicts die or go crazy, and others live normal lives? Why did 34 of the 224 veterans who contracted Legionnaire's disease at that Phily convention die from it, but 190 recovered, despite being exposed to the same tainted air?
You are trying to read way too much into the data. There are thousand variables. The amount of anthrax, whether it was cut with a base, how well the envelope was sealed up, how it was dispersed (e.g. was it dropped on the floor, like the Daschle anthrax), how many people were exposed to it and by what route, whether people recieved treatment promptly, if they were already on the alert to be suspicious of the anthrax, was their case was reported in the press at all if it was quickly treated, etc., etc. All this in a tiny, tiny sample of people.
Published reports indicate that some of the recipients received either more or better quality anthrax, although all the anthrax was of the Ames variety. Although it is possible that the terrorists were supplied with multiple grades of anthrax, my guess is that if the quality was different, it was more likely achieved by cutting the same original material with a base, much as drug dealers cut heroin or cocaine. The extreme case of this may be the "blanks" -- the samples with little or no anthrax at all, like the one at the NYT. The reportedly "soapy" texture of the AMI anthrax is, to my mind, a hint that that was the way it was done. But, the bottom line, it doesn't amount to a hill of beans how it was done, or even whether it was done at all. No policy consequences follow.
Would a soapy medium have resulted in inhalation anthrax, while a powdery medium caused only cutaneous infections? This seems backwards.
And there's another apparent difference, this time between the two populations of inhalation anthrax survivors (AMI vs. Northeast postal worker). The Northeast postal workers who survived are suffering continuing symptoms, while the AMI survivor feels fine. (There is only one AMI survivor, so this seems more questionable statistically than the first difference. It seems that this difference might be due to chance variation, which is why I did the p-value computation.)