One was a Rockville man who disappeared two days before the sniper attacks began. He had recently purchased a .223-caliber rifle, and he drove a white van with a ladder on top. A recovering drug addict, he had recently accumulated large debts.
Another was a former Marine from Baltimore with a white Chevrolet Astro van whose enthusiasm for firearms was shared by his girlfriend. Investigating the couple after a domestic shooting, police found several guns and other tantalizing clues: a manual for snipers and a misspelled note declaring "Gihad in America."
But after hours of excited news media reports about a possible sniper suspect in custody, first on Oct. 5 and then again Monday, police in both cases abruptly ended the building suspense: Neither man was the Washington-area sniper responsible for killing nine people and wounding two.
The seemingly promising leads were unusual, though, only because they became public. While the search for the sniper has drawn far more attention and resources than most homicide investigations, the case is unfolding behind the scenes in many ways just like any other.
Investigators have dangled rewards, interviewed witnesses and - though sparingly - gone public with concrete tips such as yesterday's release of two composite sketches of white minivans that could be linked to the shootings. They look for motives, knock on doors and try to knock down, or substantiate, the many theories about the mysterious sniper.
"Every dead end you get is one less person you have to investigate," said Brian H. Levin, a criminal justice professor at California State University, San Bernardino. "The process of elimination is every bit as important as the process of fulfillment."
Beyond the two men who briefly faced a torrent of public scrutiny, investigators have looked at dozens of other potential suspects who at first looked intriguing but quickly were ruled out. Montgomery County Police Chief Charles Moose told reporters that as investigators question individuals, "many, many of these people have been eliminated as suspects."
Criminologists say that is the most basic work of any criminal investigation. But it is made difficult in cases that are extremely high profile and complex, and the unsolved sniper shootings fit both descriptions.
Investigators in the sniper case are using an FBI computer program called Rapid Start to collect and catalog more 2,000 credible leads, culled from many thousands of tips. The high-tech tool, used in such complex investigations as the bombings at Oklahoma City's federal building in 1995 and at the World Trade Center in 1993, lets investigators quickly cross-reference information to find connections that otherwise might go unnoticed.
Officials have taken the familiar step of asking anyone with information about the white van described at Monday's shooting scene to come forward. They also are able to use another computer tool to generate all of the license plate numbers that would match the partial description offered by witnesses.
Investigators who have worked previous serial killer cases say success depends heavily on information management: categorizing and prioritizing the mountain of minutia pouring in every day from do-gooders, cranks and cops in the field.
'We were inundated'
Joseph J. Coffey, who led the investigation of the Son of Sam serial killer case, quickly learned that lesson. "We were inundated," said Coffey, now 64 and retired. "We had girlfriends giving up boyfriends, husbands giving up wives, mothers giving up sons."
In the days before computers, Coffey says investigators relied on filing cabinets, shoe boxes and luck to catch David Berkowitz, who was convicted of killing six New Yorkers and wounding seven others between July 1976 and August 1977. A neighbor and a co-worker had called in tips about Berkowitz, but it was a routine parking ticket that finally placed him near one of the crime scenes.
When a sniper struck in July 1994 on Long Island, N.Y., killing one person and wounding two others over 13 days, the Suffolk County Police Department ultimately combed through more than 200,000 tips.
Assistant Chief John McElhone, who helped coordinate the manhunt, said investigators built a computer database to handle the flood of leads generated by the some 300 officers who worked the case, knocking on 2,650 doors and questioning 4,100 people.
They fed the database with the name of every owner of every weapon confiscated in an arrest; the names of speeders, parking violators and parolees; and any patient recently released from a mental hospital. Then they added in every phone tip.
"You hope that each one is going to be the one," McElhone said. "People are dying and you want to stop it."
Inevitably, police wasted time on bad leads, he said. In the end, though, police linked a gun used by the killer to an auto-body deliveryman, Peter Sylvester, whose name had been entered into the computer a few months earlier after a routine arrest on a parole violation.
The current sniper investigation is under such scrutiny that FBI Director Robert S. Mueller III has included updates on its progress in his daily briefings to President Bush, the White House has said. To track the sniper's movements, investigators have taken the rare step of requesting from the Defense Department airborne surveillance equipment used recently to hunt for terrorists in Afghanistan. The high-profile case has produced an avalanche of information that investigators have to sort through.
"Having too much evidence is a common problem in any serial murder investigation," said Jack Levin, director of the Brudnick Center on Violence at Northeastern University. "How do you organize the hundreds of tips you get from the public? And by offering a large reward, you increase the chances of every pathological liar coming out of the woodwork."
A lot of tips are about people who may illegally own guns, for example. "There obviously are a lot of people in this country who own firearms," said Michael R. Bouchard, special agent in charge of the U.S. Bureau of Alcohol, Tobacco and Firearms in Baltimore. But that doesn't make someone a suspect in the deadly shootings.
Too few crucial details
Even with so much help, police in the sniper case appear to face the problem of too few crucial details. They have not released a composite sketch or any physical description of the shooter. Investigators apparently don't have such crime-scene staples as hair, fibers or DNA, and are relying heavily on witnesses accounts of vehicles leaving the scenes.
"Eyewitness observation, which is central to this investigation, is notoriously unreliable," Jack Levin said. "But it's the only way to go here."
Sometimes no amount of labor, high-tech tools or public interest guarantees results.
More than a year after the extensive search for the person behind last year's deadly anthrax mailings, that case remains unsolved. New York detectives faced the same grim reality in 1983, when a sniper killed one person and wounded six others in and around Penn Station.
Like the recent shootings near Washington, the Penn Station killings seemed entirely random. The only physical evidence were the .25-caliber rounds found at the shooting scenes.
Despite conducting more than 43,000 background checks and testing more than 1,500 handguns in search of the pistol used by the sniper, police never found the killer.
"As dopey as some of these people appear to be, they're so dopey they don't leave anything behind," said Joseph J. Maginnis, a retired New York police inspector who headed the Penn Station task force.