Staff file photo by Sam Yu In this September 2003 file photo, Bruce Ivins is seen at the American Red Cross Emergency Shelter in the Frederick Community College gym.

FBI scientists released that information Monday in a briefing at FBI headquarters, where researchers who assisted in the investigation discussed the scientific process they used to track the anthrax used in the 2001 mailings back to Fort Detrick and Ivins.

Two weeks ago, the Justice Department named Ivins as its sole suspect in the mailings, which killed five people and left 17 others hospitalized. Ivins' attorney has maintained Ivin's innocence. The noted scientist committed suicide on July 29, just before media reports revealed investigators were preparing to indict him for the mailings.

In documents released Aug. 6, the Justice Department stated that Ivins had failed to cooperate with investigators as they gathered samples of anthrax from labs for comparison with that used in the mailings, and cited that as one factor behind its naming him as a suspect. Investigators have said they have a strong circumstantial case against Ivins, but have conceded they have no hard evidence linking him to the crime.

Ivins submitted two sets of samples to the FBI. The first, sent in February 2002, did not meet the standards the FBI had set for a library of samples it was building, said Vahid Majidi, assistant director of the FBI's Weapons of Mass Destruction Directorate. The FBI destroyed that sample and Ivins submitted another set in April 2002.

Ivins' first sample would have been scientifically valid but difficult to use as evidence in court because it was gathered in a different manner than the other samples in the library, said an FBI scientist who refused to identify himself at the briefing.

Over the next few years, researchers at the National Biodefense Forensic Analysis Center, The Institute for Genomic Research and elsewhere worked to sequence the genome of the anthrax used in the attacks, and identified four mutations in that strain. They developed tests for those mutations, which they could then apply to samples from the FBI repository in an attempt to identify the source of the anthrax from the mailings.

The profile they created matched RMR-1029, a batch of anthrax under Ivins' control, but Ivins' second samples did not match that profile.

Although the scientists could not conclusively link RMR-1029 to the mailings until early 2007, by 2004, tests pointed to the U.S. Army Medical Research Institute of Infectious Diseases at Fort Detrick as the source of the spores.

In 2006, investigators went to Dr. Paul Keim, a Northern Arizona University biology professor who had been helping them identify the anthrax, and Keim gave them duplicates of all the anthrax samples, the unnamed FBI scientist said. Keim had kept his copy of Ivins' February 2002 sample, and testing identified it as RMR-1029.

Chris Hassell, head of the FBI laboratory, said he couldn't explain why Ivins would submit one sample of anthrax matching RMR-1029, then later submit a sample that did not match.

Out of 1,070 samples gathered, eight matched the powder used in the mailings, Majidi said. All of them were from RMR-1029 or its descendant.

More than 100 people at two labs -- USAMRIID and a lab officials refused to identify -- had access to RMR-1029, Majidi said. Officials did not explain how they eliminated those other people from their investigation.

The FBI used accepted science to track the anthrax back to USAMRIID, but applied it in a novel way, Majidi said. Researchers on the project have already published 15 papers based on their work, and more will come, though it could be as much as two years before everything is published in peer-reviewed journals.

More than 60 researchers worked on the project. It was subjected to critical analysis within that group, but has not been reviewed in the wider scientific community, Majidi said.

"Were we perfect?" Majidi asked. "Absolutely not ... we had missteps."

But the community learned lessons it can take forward, he said.

Dr. James Burans, associate lab director of the Department of Homeland Security's National Biodefense Forensic Analysis Center, said the anthrax powder used in the mailings could have been produced by one person in three to seven days.

Lab workers generally avoid making or working with powder because it is more dangerous than liquid preparations, he said.

Officials declined to give specifics of how the anthrax powder used in the mailings would be prepared, but Majidi said researchers at Fort Detrick had the necessary equipment.

"It would have been easy to make these samples at (USAMRIID)," he said.