Defense attorneys have doubted eyewitness testimony throughout the annals of crime, and often with good reason: People don't always accurately recall what they see, even when the stakes are huge.
Consider the playgoers who sat helplessly as Abraham Lincoln was shot at Ford's Theatre on April 14, 1865. Some swore the assassin they watched escape across the stage couldn't possibly have been a man they knew well - acclaimed actor John Wilkes Booth.
Despite eternal questions about the reliability of memory, criminal lineups remain a mainstay of American justice: Witnesses peer at a handful of potential suspects - sometimes in photographs, sometimes in person - and try to pick out the culprit.
But in a small but growing number of jurisdictions, the traditional lineup is undergoing a makeover. Armed with academic studies, defense lawyers and university researchers say the current system, which confronts witnesses with several potential suspects at once, is rigged against the innocent.
"Witnesses compare one person to another in the lineup, they decide who looks most like the perpetrator, and then they decide that must be the perpetrator," says Gary Wells, an Iowa State University psychology professor and a leading reform advocate. "That seems like a reasonable thing to do. The problem is if the real perpetrator is not in the lineup, there's still somebody who looks more like the perpetrator than the others. That somebody is at great risk."
Professor Wells and others support so-called "sequential" lineups, in which witnesses view each person one by one instead of with five others. In a sequential photo lineup, police officers place each photo in front of a witness, ask if the person committed the crime, then pick up the photo, not allowing the witness to see it again.
The witness "can't compare one to another," Wells says. "The theory is that the victim has to dig deeper to compare each person in the lineup to their memory, not to each other. You end up with a somewhat more conservative procedure."
There's a downside. Wells acknowledges that sequential lineups produce 15 percent fewer accurate identifications, according to some studies.
But the important point is that incorrect identifications dip by a third, Wells says.
The validity of lineups is hardly a trivial question, even in these days of high-tech sleuthing.
"Much has been made of DNA and trace evidence and fiber evidence, and the TV programs like 'CSI' have really built up the expectation of it being available in every case. But it's not available in the majority," says Paul Logli, state's attorney for Winnebago County in Illinois. Eyewitness testimony is vital, he adds, "and it's important that there be accuracy."
Sequential lineups are now routine in Boston and the entire state of New Jersey, and the state of Illinois is testing the system in three jurisdictions. Elsewhere, traditional lineups - typically consisting of photos, not real people lined up behind glass - remain in place.
Many prosecutors oppose mandating the reforms, which they say will give a free pass to criminals. If sequential lineups become routine, "it will be much more difficult for the [witness] to offer any identification," says Joshua Marquis, district attorney of Oregon's Clatsop County, best known as the home to the town of Astoria. "What we're trying to do is find the truth. We ought to make it easier, not make it more difficult."
While he agrees that lineups shouldn't be suggestive, Mr. Marquis says the research supporting reform is "thin"; indeed, some psychology experts question whether existing studies provide enough support for sequential lineups. Marquis is more willing to support another reform, known as "double blind," in which the police officer conducting a lineup doesn't know which one is the actual suspect. But even on that front, he doesn't accept the assumption that cops try to influence lineups.
"There's no percentage for them in doing that," he says. "We don't get bonuses for getting the wrong person convicted. It's the worst nightmare."
And what of wrongful convictions based on false identifications? Marquis isn't too worried about the prospect. "Has it happened? Yes. Is it a big problem? No."
Defense attorneys disagree, pointing to a number of cases like that of an Illinois man who was convicted of rape after the victim picked him out of a lineup even though she'd initially said the assailant had an earring and tattoo; he had neither. DNA evidence later cleared the man. In another case, a Wisconsin woman erroneously identified an innocent man as her rapist, sending him to prison for 18 years before DNA results cleared him in 2003. The victim is now an advocate for criminal lineup reform.
"We know, in general, that erroneous eyewitness identifications are the largest single cause of wrongful convictions," says Rob Warden, director of Northwestern University's Center on Wrongful Convictions.
Ultimately, "eyewitness identifications are so inaccurate that there's a question about whether they even ought to be admissible in court," Mr. Warden adds, pointing out that lie detector tests - generally considered to be 85 percent accurate - aren't admissible in most American courts.
No one seems to expect that skepticism will lead to the demise of criminal lineups and eyewitness testimony. But the double-blind approach is becoming more accepted, and some law-enforcement officials, like Illinois's Mr. Logli, president-elect of the National District Attorneys Association, are willing to accept tests of the sequential approach.
However, Logli acknowledges the ultimate challenge facing the legal system's approach to the criminal lineup: "I don't know how we're going to make it perfect."