DNA testing is far less subjective than eyewitness descriptions and police lineups. The results of other forensic methods like handwriting analysis and the matching of fingerprints, bite marks, bullets, and hair are usually expressed in a "match/no match" format that doesn't address how common such a match might be. According to a Department of Justice study, non-DNA forensic tests aided the prosecution in roughly two-thirds of 28 wrongful convictions that were later overturned based on DNA evidence. Mistakes in eyewitness identification figured into almost all the cases.


Surplus extracts of DNA from tissue samples are kept in a freezer at the Bode lab in case a second round of testing is needed to clarify results.
Reports on DNA matches, by contrast, include scientifically rigorous probabilities of the likelihood of finding the same DNA profile in a random, unrelated individual. The chances are typically far less than 1 in 10 billion for a full DNA profile from a single individual. It is that degree of improbability that forms the basis for the common perception that DNA testing is foolproof. These probabilities do not incorporate the possibility of laboratory error, however. Sometimes evidence contains a small amount of DNA, degraded DNA, or a mixture of DNA that's difficult to analyze.

Errors can occur in these cases because test results are ambiguous, and analysts often know in advance what the police expect them to find. "It's kind of like you know what you're looking for, and if it looks like it's there, you kind of convince yourself it's there," says Robert Shaler, former director of forensic biology at the Office of Chief Medical Examiner of New York City and current director of the forensic science program at Pennsylvania State University. Shaler points to a case in which a forensic scientist analyzing a difficult DNA sample reported that the chances of a random match similar to the one made to the defendant were 1 in 500. By Shaler's calculation, the chances were more like 1 in 6. "I think there was a huge amount of pressure on the lab to come up with an interpretation that was favorable to the prosecution," he says.


Scientific rigor withers in such an environment. "Some of the CSI mentality—the notion of being very involved with the police—that mentality is at war with doing really rigorous work," Thompson says. "The [laboratory workers] don't want to be blind; they want to be in there with the detectives figuring out who the bad guy is. The major problem in the field of forensic science is the difference between the scientific imperative and the law enforcement imperative. What's required to do good science and what's required to do good law enforcement come into conflict."

Paul Ferrara counters that his scientists at the Virginia Department of Forensic Science need to know what evidence to focus on in order to do their jobs efficiently when faced with a mountain of evidence. But he admits to the link with law enforcement. "Forensic science laboratories work for law enforcement agencies to solve crimes," he says. "You can't separate the two."

According to Shaler, "the secret to this is education—the people in charge of DNA labs need another round of education that these problems exist, and they have to learn how to deal with them."

Despite the issues complicating forensic DNA testing, advocates like Ferrara are already calling for an expanded role for DNA in crime fighting by law enforcement agencies. "I think taking samples upon arrest and using DNA the same way we use fingerprints is the future," he says. "I say, why do we treat DNA differently than we do fingerprints? And my critics will say, because DNA contains"—Ferrara leans forward conspiratorially—"all of our secrets. All of our personal information." He walks to his desk and pats a plastic DNA model. "It does," he adds. "Potentially. But not here in Virginia."

Ferrara opens a computer file containing a mock-up of his Virginia driver's license. At the bottom is an extra line that includes the shorthand names of the 13 "CODIS loci," the locations on a person's genes that are read to make a DNA profile. At each of these spots, scientists analyze "short tandem repeats," which are portions of DNA where a small sequence of the DNA base pairs—A, T, C, and G—repeat. These genetic stutters are denoted by a pair of numbers because they're inherited from both parents. No two individuals should have the same 13 pairs.

Ferrara taps the computer screen with his pen. "When I'm giving a talk, I say: 'Here, there's my DNA profile. Copy it down, do what you want with it.'" His point is that the numbers contain no information about him—not his predisposition for genetic illnesses, his intelligence, or even his eye color. Ferrara flashes his Social Security number on the screen. "This is more dangerous," he says. "This you can try to steal my identity with. You can't fake somebody's DNA."

You can't fake DNA—at least not yet—but Ferrara knows of a case where someone tried, unsuccessfully, to plant it. Thompson knows of several such cases. Criminals are getting more sophisticated about tampering with evidence. A criminal might, for example, pick up a used condom from lover's lane and use it to spread the DNA of an innocent man at a rape scene. "If he's in the database, there's a hit that could falsely incriminate him," Thompson says.

Ferrara acknowledges that investigators must weigh cold hits along with other facts in a crime. "DNA databanks are investigative tools," he says. "A match made doesn't mean a person committed a crime."


After DNA is extracted from tissue at the Bode lab, the genetic material is processed to develop a profile.
More often in cold cases the problem is no match at all. That has led crime fighters to begin expanding the ways they use DNA. In and near Charlottesville, identical DNA profiles turned up at several scenes of sexual assault. The police had had the DNA since 1997, but checking CODIS produced no suspects. Based on the little eyewitness information they could collect, Charlottesville police in 2003 started approaching dark-skinned men with a request for what some call get-lucky swabs—swipings from the inner cheeks of nonsuspects to collect cells for DNA analysis. The swabs were voluntary, but men would be subject to suspicion if they refused.

The idea of DNA sweeps or "dragnets" originated in Europe, and police departments have employed the technique more than a dozen times in the United States, typically stirring controversy. In Charlottesville, African Americans protested the policy and accused the police of racial profiling. At least one man sued a police detective for harassment. Critics questioned why the police were so sure the rapist was black. "And that got us thinking about what more DNA can tell us," says Charlottesville police captain Chip Harding.

Harding had the crime scene evidence sent to DNAPrint Genomics, a company in Sarasota, Florida, that offers a test for what its inventors at one time called "racial inference" or "race proportions" and now refer to as "BioGeographical Ancestry makeup." The test grew out of the company's efforts to tailor medical therapies to segments of the population. In their research they amassed a number of DNA markers that they believed could establish ancestry, which they developed into a test first marketed to genealogy buffs. Now they say it can also shed some light on appearance.

In September 2004, DNAPrint sent results to Harding that said the Virginia serial rapist's DNA was 85 percent sub-Saharan African, 12 percent European, and 3 percent Native American. Harding received a CD containing several photographs of people with similar proportional mixes of DNA. Thus far, the Charlottesville crimes remain unsolved. DNAPrint's senior scientist Matthew Thomas rejects the charge that the company's test, known as DNAWitness, is itself a form of racial profiling. "DNAWitness will hold up to scientific scrutiny whereas personal feelings and biases won't," he says.

The company has handled about 100 criminal and victim identification cases. "The technology can aid pretty much any case where you don't have a suspect but you have a biological sample," Thomas says. DNAPrint has received a $50,000 subcontract to provide raw data to a National Institute of Justice grantee.

The rapid expansion in the categories of people who must submit DNA for testing is challenging the capacity of the nation's DNA laboratories. Already there is a backlog of samples from the most serious crimes. A 2003 study found that it takes, on average, about six months to process DNA evidence from no-suspect rape cases in the United States. British police, by contrast, get answers in only two to five days. Meanwhile, in a test project at the Bode Technology Group in Springfield, Virginia, the FBI is exploring the use of robots to speed up testing.


All photographs by Tony Law
A row of heads on a robotic arm process DNA samples taken from the evidence storeroom at the Bode Technology Group's laboratory in Springfield, Virginia.

Tania Simoncelli of the American Civil Liberties Union worries that the larger the database, the greater the possibility of spurious matches, which could lead to more errors. "You're going to get a higher percentage of hits that have nothing to do with the crime."

By contrast, Sarah Hovis, the former University of Virginia student who was raped, wishes that CODIS would expand even further. "I think [with] every child born, DNA should be taken," she says. When the lab came up with a cold hit in the Hovis case, her rapist was already in jail for an unrelated driving offense. Now he is serving a 90-year sentence for rape, forcible sodomy, and four other felony counts. "The fact that I got to face him in the courtroom was so therapeutic, and I feel in the end I beat him," Hovis says. "I'm also acutely aware of the fact that we had nothing apart from his DNA."