In three decades of giving so-called "lie detector" tests, you'd think Ken Simmons would have seen everything; but after reviewing a string of reports in the controversial Michael Francke murder case, he's stunned.

"I've done polygraphs for 29 years, and I haven't seen anything like this," says Simmons, a longtime Oregon State Police polygrapher who is considered by law-enforcement officials and defense lawyers to be among the best in his field. "This is unbelievable.... This is really bizarre."

In January 1989, someone killed Francke, the head of Oregon's prison system. Prosecutors pushed a theory that it was a random car burglary and convicted a small-time drug dealer, Frank Gable.

Gable, however, maintains his innocence and, through an intermediary, has recently released records of the murder investigation to reporters ("The Murder That Would Not Die," WW, Nov. 23, 2004). The documents raise several questions about the investigation, particularly the probe's heavy reliance on polygraphs.

"Some of the techniques that are used here are simply unacceptable," says Simmons, now in private practice, who reviewed a dozen polygraph reports in the case records at WW's request. "These just aren't things we would do [today]. I'm surprised they would be done then."

Polygraphs are not really "lie detectors," which is one reason most courts won't allow them as evidence. The machines measure changes in pulse, perspiration and blood pressure but can be misled, particularly by addicts whose brains have been fried by methamphetamine or crack cocaine. There are also plenty of examples of people who flunked even when telling the truth.

"I do polygraphs for a living, and I have a lot of faith in them," says Simmons. "But when they're done badly, I have no faith in them." He places most of the Francke-probe polygraphs he reviewed in the latter category, saying that although he has no reason to think Gable is innocent, "I am very confident that I would not want to base any prosecution on these polygraph results."

In the Francke case, state polygraphers tested about 100 people and eliminated dozens of potential suspects based on their results. Later, in September 1989, when the case was at a standstill, detectives used Gable's failure of two polygraphs to make him a prime suspect.

According to reports by Gable's investigators, at least five people polygraphed by police as potential witnesses said later that police used the devices to shape their statements.

John Kevin Walker, a top prosecution witness, told a defense investigator after the trial that cops cited his initial polygraph to claim he was lying and implied he would be charged as an accomplice if he didn't tell him what they wanted to hear. "I told the truth. They came back, said, 'No, you're not tellin' the truth," Walker told the investigator.

A friend of another key witness, Mark Gesner, told investigators Gesner described the exact same experience.

Yet another key witness, Jodie Swearingen, also told a defense investigator that police subjected her to repeated polygraphs until she told them what they wanted to hear. Simmons and other polygraphers told WW that a witness would normally be given three tests at most--after that, the machine becomes easier to beat. But after reviewing the reports on Swearingen, Simmons said she was tested 22 times in less than five months. This was particularly surprising to Simmons since he administered one of Swearingen's exams--causing him to conclude that the teenage meth addict was a poor subject for any polygraph test.

Simmons, who is chairman of the state's polygraph advisory committee, says he thinks his former co-workers at the State Police used the polygraph so heavily only to draw out as much information as they could. But, he said, it's possible that the repeated tests might unintentionally have shaped witnesses' testimony if they figured out what police wanted to hear. "Anything is possible," he said.

Dale Penn, the former Marion County district attorney who oversaw the Gable prosecution, disagrees. He says it is the heavy use of polygraphs that convinces him the Gable conviction was sound. For instance, all 20 prison officials tested about the murder passed tests, which tells Penn there was no vast conspiracy to kill Francke. "If you call a polygrapher, they'll tell you the accuracy is 90 percent," he says.

Simmons is not the first ex-cop to raise questions about the Francke probe he participated in. According to Tuesday's Portland Tribune, Randy Martinak, a longtime Oregon Department of Justice investigator who worked on the case, testified during an earlier Gable appeal in 2000 that he thought the state had prosecuted and convicted the wrong guy. The provocative statement went unreported in the media at the time.