A Seattle Post-Intelligencer special report on how police here and around the nation fumble missing-person reports, originally published in 10 parts.
Wednesday, February 26, 2003
Part 9: After 21 years, the bones get a name
By LEWIS KAMB
SEATTLE POST-INTELLIGENCER REPORTER
DANVILLE -- Just before dark one early October day in 1991, two hunters found a skull just south of here, wedged beneath a log in the river's floodplain.
Investigators soon gathered along the winding Kettle River near the Canadian border, unearthing more pieces of a woman's torso.
An autopsy was performed, dental records charted and the information was entered into a national database to search for matches among reported missing persons.
When a match wasn't found, the bones were boxed and stored in a basement evidence room.
Time, distance and a deep flaw in the FBI's National Crime Information Center computer would conspire to hide the identity of a beautiful young woman whose dream was to make movies, but who vanished more than a decade before her remains were found. Another decade would pass before those remains were identified.
The bones might still be on that shelf had a young detective relied only on the vaunted nationwide computer database that doesn't work as well as the public -- and many police officers -- think it does.
Federal officials who operate the NCIC system say they can recall just one case in 20 years that the system made a match on dental records.
In its yearlong investigation of state and national systems for handling missing-person reports, the Seattle Post-Intelligencer found that nearly all states depend on the NCIC computer system when trying to identify bodies through dental records.
Many state identification officials said they were unaware that the computer they trust simply doesn't work.
In Ferry County, Detective J.R. Sharp, who is still working the long-cold case of the woman found in the Kettle River, said, "If the NCIC system worked, this thing would've been solved 10 years ago.
"What good is it to p ut the information in the system if it doesn't work?"
Sharp was a young volunteer deputy still wondering what to do with his life when the bones were found in 1991. Much later, after he had become a full-time cop and the only full-time detective on the eight-deputy force, he still wasn't aware of the bones, which had long been boxed and mostly forgotten. He learned of the case only because a Teletype from a national computer crossed his desk in 1998.
Investigators from another agency were trying to determine whether any unidentified bodies anywhere in the nation could be a woman missing from their area. They had fed information about her into the FBI computer that links police departments nationwide, and the NCIC computer had spit out a possible match with Ferry County's remains.
Gilbert W. Arias / P-I
Ferry County Sheriff's Detective J.R. Sharp stands near Kettle River and the spot where hunters found remains in 1991. After being entered into a national database, the bones were stored in an evidence room.
Sharp compared the two cases and quickly determined that the woman from the Kettle River was a routine "false hit." Then he got curious.
"Officers had worked the body recovery really well," he said. "But the case wasn't really followed up on after they couldn't find a quick match."
Investigators had assumed they belonged to a woman who had vanished a few years earlier from the Republic area. But when the dental records from the body didn't match those of the missing woman, the investigators "kind of lost interest," Sharp said.
Sharp began exploring ways to put a name on the anonymous bones.
"The driving factor was we had some human remains in our evidence room and a family out there," Sharp said. "It's our responsibility to that family to do all we can to make an identification."
The best way to do that, the detective thought, was to have an expert re-examine the bones, looking for anything missed in the initial autopsy. Sharp located a forensic anthropologist, but had to call in his boss when he learned the steep price.
Sheriff Pete Warner and Sharp petitioned the Ferry County Commission for money to solve the case.
"We didn't want to push the envelope too much," Warner recalled. "We're a department out here in the sticks, working on a shoestring budget. But they finally agreed to give us $3,000."
In March 1999, the forensic anthropologist examined the bones and tried to reconstruct the woman's face based on the contours of her skull. He conducted a more detailed autopsy, offering information the first post-mortem didn't provide.
The woman was likely 25 to 35 years old and white -- or a Caucasian-Asian mix. She was short and stocky, and likely left handed, he concluded, based on the few remains he had to work with.
A sketch artist then created a composite drawing of what the woman may have looked like in life.
Armed with this new information, Sharp asked the state's Homicide Investigation and Tracking System to send a new alert on the case to other police agencies around the Western United States and Canada. The bulletin drew about 100 calls.
"I was overwhelmed," Sharp said. "I'm only one person, and it consumed me for a couple of months."
With each call, the detective compared details of his case with others, ruling them out one by one. Sharp also sent out meticulous packets that included full dental charts and other information about the remains.
Still, there were no hits.
"I still had the desire to get the identification," Sharp said, "but I was running out of hope."
Then, about a month after he sent out the alert, California's Missing and Unidentified Persons Unit asked for the dental records.
"I assumed California was a long-shot," Sharp said. "It was so far away . . . I didn't see the connection."
But Sharp sent a packet to Sacramento. With 25,000 missing-person reports and 2,000 unidentified bodies at any given time, California leads the nation in sheer volume of such cases. To manage its caseload, the state in 1995 adopted the "Computer Aided Postmortem Identification" system -- a dental comparison program developed by the U.S. Army to help identify war dead.
But when Sharp sent the dental records to Sacramento in 1999, technicians there were still slogging through a huge backlog to update the new system. Sharp's package sat for more than 18 months before it was entered into the California database.
By then, the detective was out of leads.
Teeth offer best clues
The CAPMI system works because of a basic fact of death. When decomposed bodies or bones turn up, teeth are often the best way to bring the dead back their name. Even a single tooth can be linked to dental records of a missing person.
Enamel is the hardest substance in the body, said Gary Bell, forensic dental adviser for Washington's Missing and Unidentified Persons Unit.
"Teeth are the last thing to go, if they go at all."
The Washington state identification unit reports that dental records account for about 70 percent of all positive identifications in this state, and fingerprint matches yield most of the remaining 30 percent. The state has no database for DNA of missing persons, and a new national DNA system is still in its infancy. Police nationwide have been required to collect dental records to match missing and unidentified-dead cases since a national dental record database was started 20 years ago.
But from the very beginning, there were problems.
In 1982, Congress passed the Missing Children's Act mandating stricter police reporting for missing and runaway children.
The new law also requires the FBI's National Crime Information Center -- a sort of computer filing cabinet containing dozens of information files that links police agencies nationwide -- to be expanded to help connect the missing to the dead.
The law ordered that the system's Missing Persons File, established in 1975, include more descriptive case information. And for the first time, a federal registry of the unidentified dead was created to compare cases with those in the missing persons file.
For the first time, critical data to help identify the unnamed dead -- heights, weights and ages; scars, marks and tattoos; jewelry and clothing; and medical and dental records -- could be stored into one database that linked cases in every state.
About the same time the new computer file was established, a grisly string of slayings in the Pacific Northwest became one of the first cases to expose the system's failures.
Women in the Seattle area were disappearing. By the time their bodies were found, little more than teeth and bone remained.
In turn, a task force investigating the Green River killings was having trouble putting names to remains. With little else to go on, they focused on victims' teeth.
Aiding the investigation, forensic dentists Bruce Rothwell and Tom Morton turned to the new federal database for help. With dental records of several victims entered into NCIC's Unidentified Persons File, they hoped to find links to the reported missing.
But later, after the dentists identified some remains by manually comparing dental records, they learned a disconcerting truth:
"We were told that the individual victim's information was in the system, but it wasn't making matches," Morton said.
By 1989, Gary Bell was well aware of the problems with the FBI system. But he also knew of another search program that worked.
In less than a second, the Army's CAPMI system could sort up to 5,000 dental records, compare them to teeth of a missing serviceman, and spit out a short list of likely matches. The name at the top of the list was often correct.
Recognizing its potential, Bell helped implement CAPMI in Washington, painstakingly entering dental records for all missing- and unidentified-persons cases on record with the State Patrol.
Not long after the program was running, Bell noticed that CAPMI was making connections in Washington cases that had long languished among the tens of thousands of records stored in the NCIC files. He contacted FBI officials and was granted permission to test the national system by using records of four murder victims he recently had identified through CAPMI:
Diana Hopkins, 22, found in Kittitas County in 1990; Dawn Jennings, 37, found near North Bend in 1991; Brenda Gere, 12, found in Snohomish County in 1991; and Tammy Blair, 26, whose skull was found in Pierce County in 1991.
Bell put each case in NCIC's missing- and unidentified-persons' files just as it was entered when the cases were active. The FBI's computer was then asked to match the records to the dental charts made from the remains of the four women.
NCIC didn't make a match.
"It just doesn't work," Bell said. "No one I know has ever told me of a single time that it did."
In NCIC's sprawling steel and glass headquarters near Clarksburg, W.Va., analyst Susan Davis does recall one case in which the computer's dental system may have worked. Though vague about details, Davis said it involved a body found in the Mississippi River linked to a man reported missing in New York.
One case in 20 years -- tens of thousands of searches.
"The system is user-generated," said Davis, who has been involved with making improvements to the national system. "If something isn't working, the users (local police agencies) need to tell us that."
But FBI officials have long been aware of the system's problems. NCIC officials were alerted to the Green River examples and about Bell's test.
Bell had even published an article about his findings in one of the nation's leading forensics journals, and had personally notified FBI officials about them.
That was a dozen years ago. Still, the system remains unchanged.
Twenty years ago, when NCIC's dental program was established, the computer was programmed to sift through all information available. Everything from hair and eye color, stature, age, gender, scars, marks, tattoos, dental characteristics and other features were mixed into one search, with the computer told to pay more attention to physical descriptions than to dental information.
"Scars, marks and tattoos, we know they do generate positive hits," Bell said.
The problem is, a person's physical characteristics can change drastically from the time they go missing to the time a body is found.
The computer reports a match only if the points of similarity add up to a threshold score, but because of the weight placed on other factors, a perfect dental match alone won't meet the minimum score.
The results often mean that a computer search yields long lists of potential, but incorrect, matches -- say 50 blue-eyed, blonde-haired women about 25 years old. Often, the dental records for candidates clearly do not match the teeth of the victim. Local police typically must check each potential match by making time-consuming and usually futile phone calls and other inquiries.
Simply assigning more points to dental comparisons might not fix the problem. Nationwide, dentists chart and submit dental records to the national computer. Such entries are consistently riddled with mistakes.
That's because even experienced dentists without forensics training come up with far different interpretations under NCIC guidelines when charting the same teeth because the form is complicated and subjective.
Years after learning about the system's flaws, the FBI in 1996 appointed Bell and other experts to a task force to examine improvements.
Three years later, NCIC's policy advisory board approved the panel's recommendations to upgrade the system by creating two separate searches -- one of physical descriptors and another involving a more simplified dental record comparison.
But because missing and unidentified persons represent only a fraction of the data in NCIC, criminal information deemed more pressing -- everything from alerts about wanted criminals to the license numbers of stolen cars -- took precedence in a 2000 upgrade. The FBI now says it plans to add a stand-alone dental search program in December, although all states must update their own systems by then to use it.
"The advisory board had to make priorities," Davis said.
According to a P-I survey, 47 states depend entirely on the NCIC system when using dental records in trying to make "cold hit matches" by computer to identify bodies. Many state identification officials said they were unaware that the computer they trust simply doesn't work.
In the meantime, no one knows how many of the roughly 5,000 unidentified bodies and 100,000 missing persons the NCIC has on file are good matches the computer cannot spot -- just as it did not after a skull was found beneath a log in the Kettle River's floodplain, just before dark one early October day in 1991.
'Out of the blue'
"I'd kind of given up," Deputy Sharp said of the long-cold case. "Then, out of the blue, they called."
On Nov. 22, 2000, California's CAPMI system gave a name to Ferry County's Jane Doe.
When that state's identification unit called to break the news, Sharp was in Olympia for a State Patrol training program. The deputy who was filling in as the department's detective "came running into the lunchroom, yelling, 'We got a hit, we got a hit,' " Sheriff Warner recalled. "I said, 'What the hell are you talking about, a hit? Someone got hit? We've got ourselves a homicide?' "
Warner's first call was to Sharp.
"I didn't jump up and down and holler, but there were good feelings," Sharp recalled. "I felt satisfied that she was identified, but now I was looking more long-term. We had a homicide to investigate."
What Sharp didn't know then was that the bulk of his work -- getting the bones re-examined, the skull reconstructed, a composite sketch made -- did little to solve the puzzle. In fact, much of the new information from the second autopsy and composite sketch was dead wrong.
But simply by realerting other police agencies about the case, Sharp helped to draw the attention that would ultimately get an ID. The information needed to solve the case already was in the national database -- and had been for years.
The missing woman's dental information had been on file in California since 1981, said Greg Truax, supervisor of that state's Missing and Unidentified Persons Unit. The records were added to NCIC in the late 1980s, he said.
No one seems to know why the FBI system failed to make a match when Ferry County first entered dental records from the remains in 1991, though computer failure or human error are common.
"It's not a routine occurrence -- returning positive hits like that on cross-jurisdictional cases between different states," said Truax. "It's fortunate we have our CAPMI system."
Four months after California's CAPMI hit, a dentist in Washington confirmed the findings by comparing hard copies of the dental records.
In Republic, an evidence technician replaced an old label reading "Jane Doe" with one that said "Valerie McDonald."
Finally, more than two decades after the crime, a homicide investigation could begin.
The remains of Valerie McDonald were found by hunters in the Kettle River near Danville in Ferry County in 1991, about 11 years after she vanished.
A stunning beauty
Five-foot-6 and slender, Valerie McDonald was a stunning beauty with almond-shaped eyes and a model's bone structure. At 26, the strawberry blonde already was working as a movie extra in San Francisco, and dreaming of one day directing.
In June 1980, she moved into the Tower Apartments -- a converted hotel in North Beach, a funky neighborhood of trendy boutiques, bohemians, actors and artisans. She had come to live among others with similar dreams, some of them friends she'd met while a student at the San Francisco Art Institute.
Less than five months later, she was living in fear.
Michael Hennessey, John Abbott and Phillip Thompson had met while serving time in California prisons. When they took over management of the building where Valerie lived, several tenants were driven out.
"Val called and told me she was frightened," said her mother, Dee Dee Kouns. "I told her to get out of there."
On Nov. 9, 1980, Valerie and some friends were moving her belongings to a new apartment when Hennessey told her that he could get her a job in a movie. The script, he said, involved a detective played by Dustin Hoffman who was chasing a serial killer down the coast -- a killer with a taste for blondes. Hennessey claimed he was supplying cocaine to the director, who had asked him to find a pretty blonde for a bit part -- the role of a murder victim.
Valerie and her friends were leery of the story. Her best friend asked Hennessey if she could escort Valerie to the film shoot, but he said no, the set was closed. To persuade Valerie, Hennessey paid her $200.
But the story about the movie was a lie -- there was no film being shot. When Valerie had not returned home by the next morning, her friend called police.
"They said to me that she's an adult and there's really nothing they could do," Valerie's friend said recently. "At first they told me I had to wait 72 hours before they could even take a report."
Valerie's friend spent nearly a week searching for Valerie, pleading with authorities and trying to reach her parents in Oregon. When she finally did, the woman was "frantic because San Francisco would not let her file a missing-person report," Dee Dee Kouns said.
Valerie's parents arrived in San Francisco early the next morning and went straight to the Police Department -- where a detective dismissed their concerns. They say the detective, who now disputes their account, suggested that Valerie likely had run off to Las Vegas.
"He said to me, 'If you think something's wrong or she's dead, forget it,' " Dee Dee Kouns recalled. "And he laughed."
Outraged by the response, the couple hired private investigators to uncover information about the lives of the three ex-cons. They learned about a warehouse the men were renting, and began tracking their recent activities.
On Nov. 26 -- 17 days after Valerie disappeared -- Abbott and Hennessey turned up in Trail, B.C., where they shot it out with Mounties who ran a routine check of their car and discovered that Abbott was wanted in California. Hennessey was killed. Abbott was arrested.
Authorities determined that Thompson had also been in Trail, but had flown back to California before the shootout.
A search of the men's belongings turned up Valerie McDonald's ID card and address book and a leather jacket like the one she was wearing when she vanished, said Constable John Hudak, who was involved in the incident.
Gilbert W. Arias / P-I
Dee Dee Kouns looks at some of her favorite photos of her daughter, Valerie McDonald. "It's pretty clear what they (suspected killers) did to Val," she says.
Six weeks later, San Francisco police found evidence from robberies and plans for other crimes in the men's rented warehouse. They theorized that Valerie was killed because she stumbled across evidence of the men's criminal enterprises, but neither Thompson nor Abbott have ever been charged in relation to Valerie's disappearance or death.
Retired Inspector Anatole Balmy said San Francisco police could do nothing because they could find no body, determine where a murder might have happened or even prove Valerie was abducted.
More than 10 years would pass before any remains would be found, and another decade would pass before they were identified.
A long wait
Bob and Dee Dee Kouns had waited two decades for news about their daughter. By the time Ferry County authorities showed up at their home in suburban Portland in March 2001, the Kounses had become prominent victim advocates.
Over the years, they'd met murder victims' families, interviewed ruthless criminals and helped shape justice policies in Oregon. Little shocked them anymore.
But, when they listened to what the officers told them, the couple were stunned. They had come to believe Valerie would never be found.
"When we found out they had found her remains nine years prior, it drove me crazy," Dee Dee Kouns said.
Over 20 years, the Kounses had spent thousands of dollars and hours accumulating documents, tapes, receipts, letters, even journals to piece together a theory of what happened:
Valerie was abducted and killed in the warehouse. Her feet were cemented in a tub, her body loaded into the car and driven north. Somewhere along the way she was dumped in a stream.
They note that John Abbott, in a journal Mounties found after the shootout, had written:
"The Ice Maiden in her fallen beauty also what a dream.
Flying in the air flowing with the stream."
"It's pretty clear what they did to Val," Dee Dee Kouns said. "Especially now."
The theory makes sense to Sharp, who notes that investigators found only bones from her upper body.
"There were no legs," he said. "Someone possibly could've anchored them in concrete or something else. She could've been dumped in the river in Canada and floated down to Ferry County. Or, she could've been dumped here.
"Those are things we just don't have all the answers to yet."
Complicating the investigation is the status of the suspects: Abbott was deported to Britain after serving a prison term in Canada; and Thompson is now in a California prison for unrelated robbery and kidnapping convictions in 1983. Neither could be reached for comment for this article.
And while Valerie was obviously young and healthy when she vanished, examinations of her remains have been unable to determine a cause of death -- a necessary factor in any prosecution.
Both Sharp and the couple know that bringing Valerie's killers to justice is a long shot.
"There's been no information for 20 years," Sharp said. "The case sat cold, and we have a lot of work to do."
The lapse is maddening to Valerie's parents. If only the identification had come when her bones were found 10 years earlier.
"There would probably be a prosecution by now," Bob Kouns said.
©1996-2003 Seattle Post-Intelligencer