It looked like an open-and-shut rape case. The accuser had made identification. The suspection had previous rape conviction, his wallet was found at the scene and he refused to take a polygraph test.

But when FBI agents gave the accusing women a polygraph test, the examiner concluded that she was "practicing deception." Confronted with the expert's conclusion, she admitted she had lied and the charges were dropped.

"Without the polygraph examination and the victim's subsequent admissions that she had lied, it was very likely that prosecution would have been undertaken against the subject, and in all probability a conviction would have been obtained," an FBI memo said. And given the suspect's previous conviction, a long prison term appeared likely.

The case illustrates a significant but paradoxical fact: the polygraph -- or "lie detector" -- still is not considered 100 percent reliable in scientific terms and its results still are not admissible in court, yet the use of the polygraph by federal investigators is increasing and many consider it a crucial tool in developing criminal cases.

It was more than 40 years ago that FBI Director J. Edgar Hoover, exasperated by the machine's shortcomings, scribbled an angry memo to agents in Florida ordering them to "throw the box into Biscayne Bay." And even its most enthusiastic advocates concede that the polygraph today is essentialy the same sometimes-fallible device it was then.

What accounts for the booming interest in the polygraph is a growing realization that, while not infallible, the polygraph is sufficiently good at detecting deceptions -- about 90 percent reliable, users say -- to make it a valuable tool in shaping the directions and focus of investigations.

The training of a skilled corps of examiners, coupled with more realistic appraisal of the polygraph's potential and limits by top law enforcement officials, also has contributed to its increased use.

As used by the FBI, the polygraph measures three factors: the subject's blood pressure and pulse rate, stomach and chest breathing and galvanic skin reflex, or changes in resistence to electricity.

The object of the hookups is to measure changes in physiological functions controled by the subject's nervous system -- changes produced by the psychological stress he experiences, particularly when lying.

Before a test is conducted, the subject must sign a consent form stating that he is taking the test voluntarily and acknowledging that he has the right to refuse to answer individual questions. If the subject is a suspect and not a witness, the consent form includes the standard Miranda warnings that law enforcement officials must read to suspects before questioning them.

Bell P. Herndon, who recently retired as assistant FBI director in charge of the laboratory division, recalled that Hoover never was an enthusiastic believer in the device.

Nevertheless, the late 1950s and early 1960s, the FBI used the polygraph more and more. But in 1964, after extensive hearings, a House Government Operations subcommittee issued a blistering report on federal use of polygraphs. Through the report focused largely on Pentagon reliance on the device, Hoover ordered his agents to stop using it.

Experience during the Nixon administraion, however, rekindled FBI interest in the polygraph, and the bureau decided to send selected agents to the 14-week-long school conducted by the Army at Fort McClellan, Ala.

In assessing the polygraph today, James W. Greenleaf, assistant FBI director in charge of the laboratory division, and Ronald M. Furgerson, the special agent in charge of the polygrapoh unit, stress that they do not see the machines as "20th century witchcraft or a panacea," terms used to describe the device by early proponents.

Instead, they see the polygraph as "saving an awful lot of investigative time and in helping to direct an investigation."