AMID THE conjecture and rumor surrounding the O.J. Simpson murder case, reporters have repeatedly pointed to one apparent beacon of certainty. A typical headline, from New York Newsday on June 24, roared, "DNA: The Test That Could Nail or Clear O.J." Columnist Jim Dwyer wrote, "The blood will speak louder than all the rumors."

The Simpson affair is but one of a long list of criminal cases to which DNA testing is said to hold the answer. The public has come to see DNA testing as an infallible mix of science and magic, with the power to exonerate or incriminate, able to prove someone guilty or innocent of a horrible crime even months after the incident.

The truth is quite different. Scientists are deeply divided over how to interpret the genetic signals gleaned from a splash of blood or a dash of semen. And because scientists are at odds, the courts are confused over how to tackle this gee-whiz technology that holds so much power, but also so much danger. Science, with its slow grope toward knowledge, is colliding with the law, which likes its truths quick and definitive. Judges and jurors are being hit with inordinately complex science and statistics based on molecules no one can see and biology no one fully understands.

DNA is deoxyribonucleic acid, the substance that determines every person's biological characteristics. The color of your hair. The curve of your smile. Whether you will die young from Huntington's disease. And, in part at least, how smart you are, how musical, maybe even how kind. Aside from identical twins, no two people have the same DNA. It's the very basis of our individuality, appearing in nearly every cell in the body. So when scientists a decade ago began figuring out how to match DNA from crime-scene evidence with the DNA of a suspect, crime-fighting had found a gleaming new tool.

"It is the single biggest advancement in forensic science in 100 years, since fingerprinting," says Paul Ferrara, director of Virginia's Division of Forensic Sciences. "It provides us a degree of sensitivity and specificity far in excess of any of the traditional old techniques."

British scientists won splashy headlines when they first used DNA typing in 1986, to solve a paternity case and a double murder. Since then, the technique has proliferated; 5,000 DNA tests are conducted annually in the United States, half of them in the FBI's Washington laboratory.

DNA technicians don't even necessarily need blood or semen for their tests. Saliva from a postage stamp will sometimes do, or saliva left by a rapist on a victim's skin.

One Virginia rapist was so spooked by DNA that he made his victim sit in scalding hot water, an apparent attempt to wash away the semen. Other rapists have taken to using condoms. Sometimes the mere announcement that a DNA test is in the works terrifies a suspect into pleading guilty.

In a 1989 Maryland case, prosecutors used DNA to convict a killer even though they never found the body -- a near-impossible legal feat. The state's attorney proved that minuscule blood droplets on the furniture of a man's home came from his dead wife, destroying his claim that he knew nothing of her disappearance.

But there is a huge catch, a gap in this dazzling display of truth-detection. Scientists cannot actually examine a suspect's entire genetic makeup, since that consists of billions of units. Instead, they pick four or so segments of DNA and compare them to similar segments from body fluids at a crime scene.

If the segments don't match, the suspect is cleared. But if they do -- and this is the big twist -- experts must calculate the odds that the match is merely a coincidence, that the DNA at the scene actually belongs to someone else. While everyone's full DNA sequence is different, four segments could conceivably be the same for two or more people.

That probability calculation is at the heart of the DNA debate. After all, if a jury is told there's a one in 75 chance that a match between a suspect and blood from the crime scene is coincidental, that's very different from being told there's a one in 2 million chance.

But experts differ wildly in making those calculations. Some believe certain DNA patterns occur quite frequently in some racial groups; that would vastly increase the chance of a coincidental match.

The scientists' dissension over how to handle DNA testing is mirrored in the legal community.

"Every month you see a new decision coming down, and they're about 50-50 as to whether it's admissible or excludable," says Georgetown law professor Paul F. Rothstein, an evidence expert. "Despite all the wording and all the decisions, it comes down to a gut feel as to whether a particular jurist feels it's reliable enough evidence."

In the O.J. Simpson case, tests are being conducted on blood from the former football star's white Bronco, his driveway, his house and two gloves. Defense lawyers demanded part of the blood evidence so that they could conduct their own tests instead of relying on the prosecution's. Judge Lance A. Ito ruled that Simpson's lawyers could split the blood samples, as long as they used the equipment and procedures of the prosecution's lab. But the defense team called that unacceptable and boycotted the DNA testing altogether. If the tests do link Simpson to the murders, jurors will likely be treated to a long debate over the probability that the match is coincidental.

For years, a scientific technique had to be "generally accepted" to win entry to the courtroom -- a tough standard that can rule out DNA testing. But the Supreme Court last year changed the rule, deciding that a new technique must only follow "scientific method." That should open the door a tad wider to DNA in federal courts; but state courts, where most rape and murder cases are heard, remain split. Judges are anxious because jurors seem spellbound by genetic data -- especially when numbers like "one in a million" are tossed around -- and may downplay other evidence.

James E. Starrs, a George Washington University professor of law and forensic science who's participated in many DNA trials, describes juries enthralled by color charts and graphs on DNA.

"I've never seen a jury so alert -- not understanding, but alert, following the Yellow Brick Road," Starrs said. "I don't know of any other instance in forensics where the jury is just overwhelmed with a visual and pictorial presentation."

"I don't think of myself as a dumbbell, but it's too much to handle," Starrs confessed after attending a recent seminar on DNA testing. "My mind is still screaming. We're not being given this on a silver platter. It's a very rusty platter. The legal community can't cope."

All this makes for a swampy legal battleground. As prosecutors, led by the FBI, campaign to promote DNA testing, defense lawyers protest that DNA labs need stronger safeguards against mistakes. And, of course, defense lawyers are seizing on DNA to prove innocence rather than guilt. New York law professors Barry C. Scheck and Peter J. Neufeld recently created the Innocence Project, dedicated to using DNA to free wrongly convicted prisoners. More than a dozen have been exonerated in recent years, and Scheck and Neufeld are looking at 200 more cases. Typically, they use years-old semen or blood from a crime scene to prove that a prisoner was not the actual perpetrator.

Much of the current DNA fight focuses on libraries of genetic samples taken from convicted criminals. These databanks allow police to instantly compare DNA from a crime scene with the DNA of thousands of convicts, much as fingerprint records are now used. Congress is about to approve a bill creating a nationwide DNA library, and about half the states are assembling such databanks. They've already begun paying off.

Last year in Virginia, for example, a man broke into the home of a 63-year-old woman at 4 a.m. and raped her, making sure she never saw his face. Police found a small semen stain on the sheets, ran it against the state DNA databank, and quickly had a suspect: Jackie Crumity, released from prison four months earlier after serving 15 years for rape.

But some squirm at the prospect of vast storehouses of genetic records used by the government to create a powerful presumption of guilt or innocence. Among other things, the databanks could invite unsavory genetic research.

"The state bills permit any use of the libraries for 'law enforcement purposes,'" Scheck says. "That's sort of like saying 'national security.' If somebody wanted to do research on the violence gene -- or the sexual assault gene or the addiction gene or the psychopath gene -- they might very well try to get these samples."

For those who believe that things can go horribly wrong with DNA testing, proof is provided by Fred Zain, a former DNA technician at West Virginia's criminal laboratory. He managed to bamboozle the state's entire justice system, and he got away with it because DNA is so tough to second-guess in court.

In 1987, Zain testified that semen from a crime scene was that of a defendant named Glen Woodall -- yet five years later another DNA test proved Woodall innocent. The West Virginia Supreme Court ordered an investigation of Zain and found he'd had been falsifying DNA results for years, apparently on a misguided one-man crusade against crime.

Zain's misdeeds, the court declared last November, "are shocking and represent egregious violations of the right of a defendant to a fair trial. They stain our judicial system and mock the ideal of justice under law." West Virginia has been forced to review more than 130 cases in which Zain testified.

No doubt it's natural that a powerful new technology would provoke controversy as it jumps from the laboratory to the courtroom. But there is something different about DNA, something that raises emotions and fears to a higher pitch. It may be simply that we are uncomfortable with the idea of a physical basis for our individuality, with the notion that a splatter of microscopic molecules determines much of who we are.

DNA is perplexing for another reason too. We like the notion of a truth-seeking device that rises above human uncertainty. But we are afraid to leave justice outside the realm of human judgment. DNA testing is a reminder of how tough it is to have it both ways.

Naftali Bendavid is a reporter and editor for Legal Times.