If the assumption is that DNA delivers truth, then the Virginia Department of Forensic Science, otherwise referred to as the state crime lab, can do no wrong. It's the Mount Olympus of DNA. When DNA analysis in criminal investigations was in its infancy 15 years ago, the Virginia crime lab got in on the bottom floor and constructed its foundation on DNA.
It was the first state crime lab in the country to perform DNA testing on-site and the first to develop a DNA databank, which now contains genetic profiles of thousands of felons. Its examiners teach DNA analysis at the Virginia Institute of Forensic Science and Medicine. Its director, Paul Ferrara, helped develop the quality and proficiency standards against which all DNA testing is judged. As if all of this were not enough, former crime lab employee Patricia Cornwell has churned out a number of best-selling novels in which heroic forensic experts at a fictionalized version of the lab fight evil with science.
In the nonfictionalized world, however, recent revelations threaten both the lab's public image and long-held assumptions about DNA, or deoxyribonucleic acid. There is mounting evidence that suggests DNA analysis is far from foolproof. And recent gaffes by crime labs across the country are raising new doubts about the science of gathering and analyzing DNA and its potential abuse in a political system that puts a premium on swift criminal prosecution.
Virginia's crime lab is a case in point. In the spring, an audit of the crime lab's work on a single death penalty case found, among other things, that the lab botched and misinterpreted DNA tests twice, on separate occasions, seven years apart. When it was asked to review the results for error, the audit report says, the lab detected none.
Responding, Ferrara said the Department of Forensic Science is "gratified that the audit report does not suggest any evidence of systemic deficiency."
In light of its findings, the auditors, members of the American Society of Crime Laboratory Directors/Laboratory Accreditation Board (ASCLD/LAB), advised Gov. Mark Warner that the lab's results in other cases should also be reviewed. Warner agreed.
In early June, DNA experts from Texas, Indiana, North Carolina, Pennsylvania and Maryland were summoned to Richmond. Their project, now under way, involves scrutiny of crime lab documents produced in 161 cases. Among these are 26 cases in which the lab's DNA test results assisted the state in procuring a death sentence.
DNA analysis can, and often does, deliver the truth, experts say. But it's also highly sensitive to human error. In a related problem, the procedures used to conduct it are filled with opportunities for such error.
A growing number of scandals involving crime labs is less than reassuring. In Houston, the wrongful conviction of a 16-year-old boy based on an erroneous DNA analysis led to a massive review of DNA cases and the eventual shutdown of the Houston police department's crime lab. Misinterpreted and botched tests in Ohio, Washington, Montana and Oklahoma have also come to light.
"DNA can be black and white," says Betty Layne Desportes, a Richmond attorney who chairs the jurisprudence section of the American Academy of Sciences. "It really can be definitive and clear. But it's not that way in every case. The results in any forensic test are always dependent on the reliability and methodology of the examiner."
Every step of DNA analysis requires examiners to make judgments. And every judgment carries with it the possibility of an outcome-altering mistake. The potential for such mistakes is highest, Desportes says, when biological evidence contains DNA from more than one source or when the quantity available for testing is small.
The potential for outright misinterpretation, meanwhile, is highest in state crime labs like Virginia's, where examiners typically work closely with prosecutors and police, says DNA expert William C. Thompson, a professor of criminology, law and society at the University of California-Irvine.
Such a relationship increases the chances that examiner bias will affect test results. When this happens, Thompson says, test results that are exculpatory or ambiguous can end up being misinterpreted as incriminating because the examiner has been influenced by information supplied by prosecutors and police.
Thompson has identified a related and especially insidious phenomenon known as "inferential bootstrapping." This occurs when an examiner, knowing that another piece of evidence in the case points to guilt, resolves ambiguities in his or her own analysis in a way that also points to guilt.
Problems with forensic evidence are nothing new. But experts contend that problems with DNA evidence may be worse because of its reputation for absolute reliability. In Virginia, whose courts are notorious for upholding convictions, that reputation can be compounded.
The highly publicized case of Earl Washington Jr., the subject of the spring audit, is an instructive lesson on the danger of relying on that reputation.
Based largely on his own confession, the mildly retarded Washington was convicted and sentenced to death for the 1982 rape and murder of a Culpeper woman.
Subsequently, lawyers working on his appeals noticed that Washington's blood type did not match the blood on a blanket from the crime scene that had been submitted as evidence in the case. They also noticed that Washington, in his confession during which no lawyer was present described the victim as black when she was, in fact, white.
In 1993, as evidence of Washington's innocence mounted and his date with the electric chair loomed, the Virginia crime lab performed a DNA analysis of evidence from a vaginal slide in the case as well as from the blanket.
The analysis excluded Washington as the source of a semen stain on the blanket. But the results from the vaginal swab were less clear. Although the evidence contained a genetic trait that did not belong to Washington, the victim or the victim's husband, the lab concluded that Washington could not be eliminated as a potential suspect.
As a result of this finding, Washington's sentence was commuted to life.
Seven years later, in the summer of 2000, the Virginia crime lab performed another DNA analysis of the evidence in the case using newer technology.
This time, the tests revealed a genetic profile on the blanket that did not belong to Washington. The lab's DNA database linked the profile to Kenneth M. Tinsley, then serving a life sentence in the Virginia Department of Corrections for rape.
The lab was unable to obtain any results from the previously tested vaginal slide. But tests on a second, duplicate slide excluded both Washington and Tinsley and turned up two additional genetic profiles the lab could not identify.
The lab conveyed these results to Gov. James S. Gilmore, who was considering Washington's request for a pardon. At the time, the lab's reputation was intact. Across the state, prosecutors and police relied, as they still do, on its results. The analysis excluding Washington had been performed on DNA extracted from sperm found in the victim's body and smeared onto a slide the gold standard of DNA exonerations. For Gilmore, a former prosecutor and attorney general, this was not enough.
Instead of immediately pardoning Washington, Gilmore ordered more tests on the blanket. When those tests again excluded Washington and implicated Tinsley, Gilmore agreed to a pardon.
Yet he refused to concede that Washington might be innocent and went out of his way to imply that Washington was not innocent. The "absence of DNA evidence does not necessarily mean that an individual was absent from the crime scene just that he has not left any DNA markers," said Gilmore's statement announcing the pardon.
In the statement, Gilmore did not apologize to Washington; he apologized to the victim's family. And instead of ordering Washington's release from prison, Gilmore resurrected details from Washington's past assault conviction and put the matter into the hands of his parole board. Washington, despite his exoneration, spent another seven months in prison.
In this way, Gilmore virtually ensured that Washington's legal team, made up of lawyers from the New York-based Innocence Project, would never rest with respect to Virginia. In the wake of Washington's release, they filed a civil lawsuit in 2002 in an attempt to clear his name alleging, among other things, wrongful prosecution and imprisonment.
So began the series of events that led, three years later, to the discovery that the work of the Virginia crime lab was not, as previously supposed, infallible.
In the spring of 2004, at the request of Washington's attorneys, a California DNA expert named Edward Blake performed a DNA analysis on biological evidence from Washington's case. Blake's lab, Forensic Science Associates in Richmond, Calif., has performed innumerable tests for prosecutors, police and defendants alike. On at least two occasions, officials in Virginia have hired him to analyze DNA in death penalty cases. Evidence from one of those cases, that of Roger Keith Coleman, remains in a walk-in freezer in his lab.
The evidence Blake tested for Washington's lawyers was, arguably, the same evidence tested by the Virginia crime lab. It was contained on a vaginal slide made by the state medical examiner during the 1982 autopsy of the victim in the case. Two sets were prepared; the same swab was used to smear all of them. One set went to the Virginia crime lab for analysis. The other stayed at the medical examiner's office until, 22 years later, it was discovered during a search requested by one of Washington's lawyers.
The evidence was the same. The test results were not.
In fact, every aspect of the analysis differed profoundly when performed by Blake. On the slide where the Virginia lab found only a few sperm cells, Blake found hundreds. And the DNA he extracted practically screamed Tinsley's name.
Asked to summarize these differences in findings, Blake did not equivocate. "It is the difference between night and day," he says. "It is the difference between competent and incompetent."
When news of these findings reached Virginia crime lab director Ferrara, he dismissed them with disdain. He told the Chicago Tribune that Blake was a "hired gun." The Virginia crime lab stood by its work. Any concerns raised by Blake were baseless and motivated by envy. "When you are on the top of the heap," Ferrara said in the Chicago story, "you are going to have someone trying to knock you down." He continued, "As far as we're concerned, there is no error."
Throughout the summer of 2004, Ferrara stood his ground. When Washington's lawyers called for an independent review of the lab's work, he dismissed the idea as preposterous. "We don't need anybody to pass judgments on our results," he told Virginian-Pilot editorial writer Margaret Edds that summer. "You're talking about one of the best DNA labs in the country."
In an effort to clarify the matter, Edds, the author of a recently published book on the Washington case, sought out three DNA experts and sent each a copy of Blake's lab report and a copy of the Virginia crime lab's report. What did they think?
The results of this informal poll, like the analysis conducted by Blake, did not flatter the Virginia crime lab.
Dr. Robert Shaler, director of forensic biology in the New York City medical examiner's office, called the Virginia analysis a misguided attempt to "make a legitimate interpretation from bad evidence."
Criminology professor Thompson was also among those consulted. "What a mess," he said of the lab's analysis. "What an enormous botched job."
Warner ordered the audit in September.
Eight years ago, in a case that has never been reviewed or audited, the Virginia crime lab made another mistake. But for a series of improbable events, it would not have been revealed. But for the discovery of the errors in the Washington case, it might have been forgotten.
The story begins in the spring of 1997 when two lawyers drove from Fredericksburg to Richmond for meetings with examiners at the Virginia crime lab. Their client, a 44-year-old itinerant house painter named Karl Michael Rousch, was headed for trial in a month for the abduction, rape and murder of a 16-year-old girl. This crime, one of the worst, most-publicized cases in Spotsylvania County history, had dominated the local news since the girl's disappearance the previous September.
Rousch, jailed since the previous November, insisted he was innocent. Tests performed on evidence at the crime lab insisted he was not.
The most damning of this evidence was explained in intricate detail by the lab's fiber examiner during the lawyers' meeting with her in April 1997. Three different kinds of fibers found in Rousch's 1984 Dodge Caravan perfectly matched fibers taken from a blanket in which the body of the dead girl, Sofia Silva, had been wrapped. The lab had also performed a DNA analysis of biological evidence taken from Silva's body. Those results, while less conclusive than the fiber match, did not exactly buttress Rousch's claim of innocence.
One of the lawyers, Ben Woodbridge, recalls the lab examiner's explanation that day: "If there was only one contributor [to the DNA sample], that contributor could not have been Karl Rousch. But if there was more than one contributor, then Rousch cannot be ruled out."
To a DNA crime analyst, this language denotes a result known as "conditional inclusion." To the criminal defense lawyers that day, it meant another kind of result: conviction. On the drive back to Fredericksburg, Rousch's prospects seemed grim.
"We thought it was all over," Woodbridge recalls. "The fiber evidence was just devastating. I mean, [the fiber examiner] was going to get on the stand and say it was a complete match. And now the DNA guy was basically suggesting that Rousch and another person had raped the girl before killing her."
Because Rousch could not afford to pay a lawyer, Woodbridge and Phillip Sasser had been appointed to his case by the judge. This meant that, in order to get a second opinion on the forensics tests, they had to ask the judge for money to hire an expert. But the judge said no. Then they asked for more time to prepare. The judge delayed the trial until September.
As it turned out, that delay would be indefinite. In a series of unlikely events, both tragic and extraordinary, the state's case against Rousch turned to vapor.
It began with the disappearance, on May 1, 1997, of two Spotsylvania County sisters: Kristin Lisk, 15, and her 12-year-old sister, Kati. The girls had come home from school and vanished, just like Silva. Investigators with the Spotsylvania County Sheriff's Department regarded this as mere coincidence. They knew Rousch had killed Silva because the Virginia crime lab had proved it. Rousch was in jail. Hence, no connection.
The FBI, brought in to help with the case, was unhampered by such certainty. Soon after the girls' bodies were found, the agency decided to retest the evidence in the Silva case. The results were unambiguous and dramatic: The three fibers from Rousch's Dodge Caravan did not match fibers from the blanket. In fact, none of the fibers matched. The Virginia crime lab, in addition to being flat wrong, was not even close. The FBI was so adamant on this point, Woodbridge says, that it eventually became clear to everyone involved that the original fiber analysis was "unconnected to science or fact."
In the drama caused by this revelation, the lab's DNA analysis lost all significance. The charges against Rousch were dropped. Not long afterward, the FBI announced the establishment of an undisclosed "forensic link" between the Lisk and Silva murders. The case remained unsolved until June 2002, when police attempted to capture a rape suspect named Richard M. Evonitz in the parking lot of an International House of Pancakes in Florida. Surrounded by police, Evonitz shot himself in the head. In a subsequent search of his car and his South Carolina apartment, police found newspaper articles and notes about the Silva and Lisk cases. A DNA analysis by the FBI crime lab confirmed that the evidence in the cases matched Evonitz.
The Virginia crime lab's failure to exclude Rousch as a possible suspect in the Silva case was, according to Thompson, accurate. But in explaining the result to Rousch's lawyers, the lab did not explain that the exclusion was statistically meaningless.
By Thompson's calculations, the test results indicate that 36 percent of Caucasian-Americans and 24 percent of African-Americans also cannot be excluded as suspects. "That," he says, "is like saying the suspect has brown hair. So what?"
The lab's fiber analysis has never been explained. How, at what was then one of the most reputable labs in the country, was an examiner able to pass these results off as science?
In 2001, three years after the revelation, a reporter for the Fredericksburg Free Lance-Star posed a version of this question to Ferrara. He seemed to have no answer. "I still, for the life of me, don't know what the hell she did," Ferrara said.
Perhaps, he speculated, the examiner's ties to law enforcement her previous career as a state trooper, for example, or her husband's job as a sheriff's deputy had clouded her objectivity. In any event, he said, the incident had ended the woman's 12-year career at the lab. New policies and procedures were now in place to ensure accuracy and integrity. The lab had learned a hard lesson. But it had emerged improved.
"Ferrara said no lab is ever immune from problems," the Free Lance-Star reporter wrote later. "But if suspicions exist, they need to be addressed."
Virginia's next execution, scheduled for July 11, will be that of Robin Lovitt. Sentenced to death for the 1998 murder of a night manager in an Arlington pool hall, Lovitt has insisted since his arrest that he is not the killer. His conviction was based, in part, on a DNA analysis performed by the Virginia crime lab in 1999.
The analysis, performed on two spots of blood on the pair of scissors thought to be the murder weapon, excluded Lovitt as the possible contributor of one of the spots, but not the other. According to Lovitt's current lawyers, the results also implicate one out of every three white men and one of out of every three black men. Another test, on blood from a jacket, was inconclusive.
In Lovitt's case, no new DNA analysis is possible. An Arlington County court clerk threw out the evidence in May 2001. S
Letters to the editor may be sent to: firstname.lastname@example.org