Three years ago this month, I was sitting across from Charles Ivins, the elder brother of Army scientist Bruce Ivins — the man the FBI says was behind the 2001 anthrax attacks. Those attacks, which came in two waves just weeks after 9/11, were the first biological assault ever directed against this country. A handful of politicians and media outlets received envelopes in the mail that contained deadly anthrax spores. In all, five people died from breathing in the anthrax, and 17 others were sickened. When I met with Charles Ivins, it was just hours after a memorial service for his brother, who had committed suicide before the FBI could charge him in the case.

To hear Ivins’s family and colleagues tell it, the socially awkward, geeky scientist took his own life because the FBI had been hounding him for months. A strain of the anthrax known as RMR-1029 and linked to the 2001 anthrax mailings had been connected directly to Bruce.

Charles was sure that his brother couldn’t possibly have been responsible. “Bruce was compassionate and considerate,” he assured me. “He didn’t do it.” But as we talked, it became clear that he’d never seen the evidence the FBI had amassed against his brother. I had been reporting on the case for NPR, and I handed him what I had from the FBI and watched him read.

Ivins and his wife, Nita, wordlessly passed pages of the affidavits to each other. They pointed to passages, exchanged glances and seemed to fill in the redactions with names and dates and faces they knew. Eventually, darkness crept into Charles’s expression. “I’m stunned now, I am just totally stunned,” he said. While he was careful not to say he thought his brother was the anthrax killer, the evidence had clearly planted seeds of doubt.

David Willman’s new book, “The Mirage Man,” is meant to spark the same reaction in those who didn’t know Bruce Ivins and can’t fill in the blanks. Willman, a Pulitzer Prize-winning Los Angeles Times reporter, provides the first behind-the-scenes account of one of the most far-reaching investigations in the history of the FBI, and shows how circumstantial evidence and innuendo added to the expensive and ultimately unsatisfying conclusion of the case.

‘The Mirage Man: Bruce Ivins, the Anthrax Attacks, and America's Rush to War’ by David Willman. Bantam. 448 pp. $27 (Bantam)

The FBI identified the anthrax strain and traced it back to a U.S. Army lab in Maryland fairly early in the investigation. Working on tips, they zeroed in on a somewhat bombastic physician-researcher named Stephen Hatfill who had worked in the same lab as Ivins.

Under normal circumstances, Hatfill might well have been excluded as a suspect because he worked with viruses such as Ebola, not bacteria such as anthrax. He also hadn’t had access to the lab for more than two years before the attacks. Still, with the FBI under intense pressure to crack this case in the wake of its 9/11 failures, leaders at the bureau couldn’t shake their conviction that Hatfill was involved.

Willman writes that the FBI felt it had an unassailable source: a team of bloodhounds from Southern California. They had been trained specifically to sniff out RMR-1029. One four-footed sleuth named TinkerBelle had “alerted” on Hatfill, his apartment and his girlfriend numerous times.

But TinkerBelle had been in the middle of a controversial case in California in which she identified a man who authorities thought was a serial rapist. Turns out the dog got it wrong. The FBI must have known about the high-profile California case, but they still kept Hatfill in their sights largely on account of TinkerBelle’s nose. Willman says the bloodhound evidence had detractors even inside the FBI, but the case became such a priority that agents were reluctant to consider any other suspects. In fact, FBI Director Robert Mueller took over the case personally, demanding weekly briefings, which ended up, Willman writes, coaxing agents to tell their boss what they thought he wanted to hear.

Early tips that pointed to Ivins were ignored. A former colleague of his named Nancy Haigwood called the FBI after it sent out a letter to 30,000 members of the American Society for Microbiology, the largest professional organization for scientists working with bacteria and viruses. The letter could have described Ivins. “It is very likely that one or more of you know this individual,” it read. “This person is experienced working in a laboratory. Based on his or her selection of the Ames strain [of anthrax] one would expect that this individual has or had legitimate access to select biological agents at some time.”

After working with Ivins at the University of North Carolina at Chapel Hill, Haigwood had had a number of encounters with him over the years — nearly all of them negative. She suspected him of stealing her lab notes after she rejected one of his advances. He fabricated a letter to a local newspaper and signed her name. Particularly worrying was an unsolicited e-mail she received on Sept.21, 2001, just three days after the first round of anthrax letters would have been posted. In it, Ivins predicted a big terrorist event that would put the lab at Fort Detrick — and anthrax experts like himself — front and center. Haigwood called the FBI hotline. The bureau filed away her tip and didn’t come back to it until more than four years later.

Perhaps the most compelling argument Willman makes in assessing Ivins’s role in the attacks grows out of never-before-revealed details of the scientist’s fragile mental state. Ivins told friends he thought he had inherited his mother’s violent schizophrenia. In e-mails, he wrote about being two Bruces instead of one. The book also details for the first time a case in which Ivins told a psychiatrist that he had once planned to poison a colleague he felt had spurned him.

“Ivins was settling into a recognizable pattern,” Willman writes. “One-on-one he was the smiling, devout colleague who exuded empathy. Behind people’s backs he was prone to bizarre, secretive acts of vengeance for the most obscure of slights.”

Researchers, scientists and people who knew Bruce Ivins continue to disagree about the anthrax case. They, like Ivins’s brother Charles, say Bruce was odd — but not a killer. And, because Ivins took his own life before a jury could weigh the evidence, the anthrax mailings will probably always have their grassy-knoll theorists.

The Justice Department, for its part, can’t seem to catch a break when it comes to the anthrax killings. Just last month, DOJ filed a motion of summary judgment in a civil case brought by a victim of the anthrax attacks against the lab at Fort Detrick. In the filing, the Justice Department said that Bruce Ivins didn’t have the equipment — a special drying machine called a lyophilizer — needed to treat the anthrax in his lab. It seemed to cast fresh doubt on the FBI’s case against him. Naturally, the media pounced.

A day later, the Justice Department had to issue a correction. It said Ivins had ordered a lyophilizer on his own — and stationed it in the lab. The Justice Department said he had labeled it “property of Bruce Ivins.” The press release ended rather defensively: “We are confident that we would have proven his guilt beyond a reasonable doubt at a criminal trial,” the Department of Justice said, “and maintain in the civil suit that the evidence of his guilt meets the lesser civil standard that it is ‘more likely than not’ that Dr. Ivins mailed the anthrax attack letters.””

Dina Temple-Raston is the counterterrorism correspondent for NPR and the author of numerous books, including “The Jihad Next Door: The Lackawanna Six and Rough Justice in an Age of Terror.”