Police crime labs are churning out tainted evidence — and nobody's doing anything about it.
Editors note: This story was originally published May 31, 2001.
They found Virginia Lowery lying in the garage of her Excelsior home, an electrical cord around her throat, an ice pick jammed through her skull — in one ear and out the other. For the next 11 years San Francisco homicide detectives made no progress on the case. Promising leads turned into dead ends. Theories collapsed. The cops assigned to the case retired. It looked like Lowery's 1987 slaying would never be solved.
Then in April 1998, by pure chance, police found Robert C. Nawi. Or rather, they found his fingertips.
When Nawi, a 57-year-old carpenter, got in a shouting match in a North Beach watering hole, he was picked up by the cops on misdemeanor charges and shuttled to county jail, where he was fingerprinted and booked. The computer spat out some interesting news: Nawi's digits, according to the database, resembled a fingerprint found at the scene of Lowery's slaying.
Soon thereafter, police evidence analyst Wendy Chong made a positive print match, and the new suspect found himself facing murder charges and life in a cage.
Nawi's fate, to be decided at trial next year, rests largely on police readings of his fingerprints, as well as some DNA gathered by the coroner. Which raises some questions: How, exactly, did the cops and their computers analyze the evidence? Did they get it right? Is anybody checking their work?
Making a match between the distinguishing ridges and whorls, often microscopic, of two fresh fingerprints is a relatively simple task for a print expert. However, cases like Nawi's aren't so clear-cut: the print collected in Lowery's garage is faint, smudged, and missing in patches.
Michael Burt, the resident forensicscience guru at the San Francisco Public Defender's Office, shows me an 8-by-10-inch enlargement of the print discovered at the murder scene; it's blurry, grainy, and only about 60 percent complete. To my layperson's eye, it bears little resemblance to the clear, fresh mark left by Nawi at his booking. "The one print is so washed out you can't see anything," says Burt, who is representing Nawi. "This is not science at all; it's subjective and shouldn't be allowed."
Burt, a 22-year veteran defense lawyer known around the Hall of Justice for his trademark cart full of documents, has plenty of cause to doubt the cops' evidence. Despite what you may have seen on Law and Order, fingerprint examiners can — and often do — get it wrong. Last year 141 of America's top forensic labs were tested to see if they could accurately match two fingerprints: 39 percent failed; 11 labs made false IDs. San Francisco analysts are rarely, if ever, graded for accuracy.
Jim Norris, head of the San Francisco Police Department's forensics division, argues that new computer imaging tools are making it possible to match even sketchy, partial prints. "When somebody shows a print that was originally collected at the crime scene, and it looks very difficult to deal with, what they're not looking at is the image that has been [digitally] enhanced," Norris explains. "It's a lot easier to deal with." Norris admits that the department has seldom tested its print examiners for accuracy, but he says their work is constantly checked by superiors.
According to Burt, in this particular instance analysts didn't turn to computers but simply enlarged the prints before making the call. The district attorney's DNA evidence against Nawi is equally flawed, he says. When coroner Boyd Stephens autopsied the corpse, he — per routine — snipped the woman's fingernails with a household nail clipper and stuck them in an envelope.