from an article by Debora MacKenzie, who has been knowledgeably reporting on anthrax and bioterrorism at the New Scientist for more than a decade:
. . . Next the team developed highly sensitive tests to screen all 1072 samples for four of the mutations. Eight samples had all four. One came from a flask labelled RMR-1029 that Ivins was responsible for at USAMRIID. The other seven came from cultures taken from that flask, only one of which was not located at USAMRIID. So while these findings show the attack spores came from one of these cultures, the FBI has gone further in concluding the attack came directly from the RMR-1029 flask.
Another question is how the attacker turned the water-based slurry of spores in the flask to the fine, dry powder in the letters. . .