When George Washington alleged "I cannot tell a lie," he was probably lying. Catching him at it, however, would not have been easy.

For years, researchers have known that just about everybody lies, but they have also shown that people can detect other people's lies only a little over half the time--slightly better than chance. Statements such as "My child never lies to me" or "I can always spot a liar" are mostly lies.

But not every time.

In the most recent issue of the journal Psychological Science, a study by California researcher Paul Ekman and two associates suggests that by choosing the right experts, lie detection scores can rise substantially above the norm.

Ekman, a psychologist at the University of California at San Francisco, believes that 85 percent to 92 percent of liars give themselves away by making 35 different errors, singly or in combination. These "cues" to lying can include facial movements, voice tics, nervous gestures and the like.

And in a test administered to six different sets of professionals, Ekman and his co-researchers found that a group of 23 federal law officers--principally CIA agents--could spot the liars 73 percent of the time.

A second group of 107 clinical psychologists with an abiding interest in lying research scored 67.5 percent, while 43 Los Angeles County sheriff's personnel with reputations as good interrogators scored 66.7 percent.

"The fact that some groups do do better tells us that you don't need [to know] all the measurements," Ekman said in an interview. "In an ordinary interview situation, it is possible to do significantly better than chance."

Other experts, however, are not ready to concede that Ekman's results necessarily mean that examiners will be able consistently to assemble groups of overachievers.

"Average accuracy is 54 percent, and two-thirds get between 50 and 59 percent," said University of Virginia psychology professor Bella DePaulo, who is making a review of more than 100 studies on lying. "Groups who get in the high 60s and low 70s [are] impressive, but not unprecedented."

DePaulo said that in one of her own studies of federal law enforcement officers, she found that old pros didn't do any better than recruits at spotting liars: "They just thought they did."

Ekman agrees that police "everywhere" believe "70 to 80 percent of people are lying," and put great faith in "the school of hard knocks," showing that they, like everyone else, "wholly overestimate their abilities."

Still, Ekman and his associates believe they can teach lie detection, and they conduct perhaps 15 seminars per year for judges, police officers and other groups of truth-seekers. Their approach is to build a framework in which questioners can consciously use techniques that they may already know instinctively.

"People who are highly accurate mention many of the things that we have discovered in our research," Ekman said. "They have figured out some of the things we teach without their being taught to them."

Cues include a "lack of coherence" in answers, words "that don't fit the facial expression" and "gestures that don't fit the voice," Ekman said. "You have to give people videotape and let them see it."

And you have to listen to all of it, said J.J. Newberry, a former Bureau of Alcohol, Tobacco and Firearms agent and a leading Ekman disciple who teaches seminars on lying. "There is a cluster of things that a person does before he lies. There is no Pinocchio response."

Favorite film clips for Ekman include then-British Prime Minister Margaret Thatcher, lying when she denied granting permission to torpedo the Argentine cruiser General Belgrano outside a British-imposed exclusionary zone during the Falklands War.

"It's what we call a flutter--her eye blinked very rapidly five to six times in a row. It's a sign that you're thinking a lot about what you're saying when you say it," Ekman said. "If my wife asks me what my car is doing parked in front of a Holiday Inn at 2 p.m., I should know that answer without having to think."

He noted that Thatcher later admitted lying about the General Belgrano, as did Kato Kaelin when the prosecution asked him during the O.J. Simpson trial if he had a book contract--"His voice got much softer suddenly."

Ekman never saw Simpson himself testify. He thought Clarence Thomas "probably did more than he said, but less than she [Anita Hill] claimed." He saw President Clinton's deposition in the Paula Jones case, but will "never talk about a politician while he's in office." He saw Oliver L. North lying before the Iran-contra committee and never had a clue.

"North is really great, totally convincing," Ekman said. "There are some people who are professional performers--in theaters, sales and politics. One of the reasons they are difficult to catch is that they have the ability to believe their role as they are playing it."

Lie detection, like the use of the polygraph, also has other shortcomings that make it unsuitable as a courtroom tool. Researchers agree it is easier to detect a high-stakes lie than an "everyday lie," but as the stakes rise, questioners have to guard against the "Othello response"--confusing lying with the terror of a truthful person.

"If I had a seven-foot guy in a room with me who was going to kick hell out of me, I'd tell him anything he wanted to know," Newberry said. "I've never seen a result that was any good from physically abusing anybody."

It is also difficult to construct convincing tests. In the study reported in Psychological Science, Ekman had "liars" argue opinions on controversial subjects that were opposite their true beliefs and gave financial rewards for success. By contrast, a suspected murderer may be facing the electric chair if he makes a mistake.

In fact, though she is skeptical that lie detection can be taught, DePaulo has discovered that detection improves with familiarity, particularly among women.

"There is no difference if the women and men who are trying to catch the lie do not know the liar," De Paulo said. "But if you know the person, women get better. The men still have no clue."

Still, DePaulo suggests that "context" is important. If law officers are "good at detecting the lies they're supposed to detect, what does it matter if they go home and believe their teenage son when he says he's going to the library on Friday night?"

CAPTION: Know the "cues" for lying: A true smile, left, must have movement from the muscles around the eyes, while a false smile, right, is mouth only.