By his estimate, Pennsylvania dog handler John Preston worked more than 100 criminal cases for Brevard County police and prosecutors from 1981 to 1984, earning tens of thousands of dollars to link suspects to crimes.
Since then, Preston has been exposed as a fraud, and Florida has freed three imprisoned men since 1987, including accused rapist Wilton Dedge in 2004 and accused murderer William Dillon this year. Both spent more than two decades in prison before modern DNA tests showed that they couldn't have touched the evidence to which Preston's dogs had linked them.
Which raises the questions: How many others did Preston help convict in the early 1980s? And how did bad evidence from his tracking dogs influence the outcome of cases?
Of 48 grand jury indictments in Brevard during those four years, FLORIDA TODAY found 16 murder and capital sexual assault cases in which the dog handler testified or provided key evidence. Archived trial records revealed that Preston's tainted testimony mostly helped -- but sometimes hurt -- prosecutors:
After Dedge's 2004 release, Brevard-Seminole State Attorney Norman Wolfinger -- a public defender during the Preston era -- asked his staff to review all cases of people still imprisoned because of Preston.
"The system is set up to make it difficult to convict an innocent person. This should rarely happen," said law professor Charles Rose of Stetson University in St. Petersburg. "But DNA evidence shows what we've said all along, that people are human, and humans are fallible."
Wolfinger ordered another review after a FLORIDA TODAY report on June 21 identified others convicted with Preston's help.
Brevard investigators first employed Preston in 1981 on the recommendation of an Orange County sheriff's deputy who had used him in the murder investigation of an Eatonville postmistress.
Preston died last year.
Trial records show that Preston used two methods to link suspects to a crime:
The Brevard County Sheriff's Office brought Preston in on the 1981 double-murder and sexual assault of two teenage Titusville girls. Witnesses said Mark Wayne Jones picked them up at a Cocoa Beach nightclub and heard him talk about killing them. After sniffing one victim's shoe, Preston's dog Harass II alerted to her scent in Jones' 1973 Ford Pinto.
Preston didn't testify in court, but then-Assistant State Attorney Dean Moxley mentioned the dog evidence at the hearing in which Jones pleaded guilty to murder to avoid the death penalty, court transcripts show.
Subsequent convictions only bolstered the dog handler's credibility as an expert witness, trial records show. Moxley told a judge in a 1982 murder case that Preston had helped seal convictions in 10 other cases -- five of them murders -- and was recognized as an "expert tracker" by the 18th Judicial Circuit. Preston once testified that he had worked more than 100 cases in Brevard and more than 50 elsewhere in Florida.
Nearly from the start, defense lawyers -- in many cases, public defenders -- tried to discredit Preston as a witness by attacking his credentials and his dogs' training and inconsistent performance. They criticized Preston for leading his dog, giving foot signals, standing near scented items in lineups in which one item was obviously different from the rest.
Phyllis Riewe, an assistant public defender in 1981, was first to move to suppress Preston's evidence, in the rape trial of Frank Berry. She said the dog naturally picked an old nightgown from a lineup that otherwise included four brand-new nightgowns and couldn't help but pick the knife bearing the victim's blood from a lineup with other knives washed clean.
The judge denied that motion. Berry was sentenced to 124 years in prison.
"You need a mountain of evidence to show this witness was not credible," said Rose, the Stetson University professor. And even so, he cautioned, "a dirty witness doesn't mean an innocent defendant."
Prosecutors have a responsibility to ensure that evidence doesn't merely conform to preconceived notions, said George "Bob" Dekle, a criminal law professor at the University of Florida and a former prosecutor for the Third Judicial Circuit of Florida.
"You have to be wary," Dekle said. "You have to make dang sure it's the right evidence."
Moxley continued to employ Preston as an expert, even as other prosecutors and law enforcement officers began to suspect that Preston was a fraud, trial records show.
Sam Bardwell, who prosecuted Frank Berry, testified in another man's appeal in 1987 that he had begun to suspect Preston after the dog handler refused to participate in a "scientific experiment" Bardwell recommended.
In another case, Preston's dog failed three times to call attention to anything in a "scent lineup" before it seemed to identify the murder weapons. Then it urinated on several towels marked as evidence from the crime scene.
Preston's credibility collapsed in 1984, costing prosecutors a murder conviction. A jury acquitted defendant James Elmen, who later, in 1985, raped and killed a Jacksonville woman.
Convicted murderer Juan Ramos was acquitted on appeal in 1987.
Testifying at Ramos' appeal, ex-prosecutor Bardwell said discrediting Preston was a high point of his career. That testimony, plus detailed accounts of how Preston manipulated scent tests, helped unravel the state's case against Ramos. "You couldn't see any competency in the action of the dog," Bardwell said.
At the time, Moxley said, he had no reason to doubt Preston's credibility."If you believe you have evidence that is not reliable or tainted, you do not use it," Moxley said. "Today, I wouldn't use Preston."