At least 12 of the 15 drivers killed in major auto racing since 1991 -- including seven of the eight killed in NASCAR and all three in 2000 -- died of injuries caused by violent motion of their inadequately restrained heads in crashes.

At least nine, including four of the last five, suffered a syndrome called basal (or basalar) skull fracture. At least eight died specifically of it: NASCAR's J.D. McDuffie in 1991, the U.S. Auto Club's Jovy Marcello and NASCAR's Clifford Allison in '92, Formula One's Roland Ratzenberger in '94, the Indy Racing League's Scott Brayton in '96, CART's Gonzalo Rodriguez in '99 and NASCAR's Adam Petty and Kenny Irwin in 2000.

The ninth, CART's Greg Moore in '99, likely would have died of basal skull fracture had he not suffered other lethal injuries to his head and neck.

Yet basal skull fractures and related intense, whiplashlike injuries to race drivers remain widely misunderstood. Many physicians, including some pathologists, still believe basal skull fracture must involve an impact of the head against something solid, such as a roll bar.

Indeed, in the New Hampshire state medical examiner's death certificates on Petty and Irwin, the descriptions of the causes of both deaths began, "Blunt impact of the head . . ."

"That's a classic view of it," says Dr. John Melvin, a Detroit-based biomechanical engineer who is one of the world's leading authorities on racing injuries. But "the head doesn't have to hit anything. We're very convinced that this basal skull fracture occurs very early, due to the violent whipping motion of the head, before the head can hit anything."

Melvin's research with computerized crash dummies in racing simulations began in 1992. But he has studied basal skull fracture for nearly 30 years -- first as a professor at the University of Michigan for the federal Occupational Health and Safety Administration, then for 13 years at General Motors, and now at Wayne State University's center for the study of racing injuries.

Here is a layman's synopsis of what occurs, based on the studies of Melvin and Dr. Robert Hubbard, a professor of biomechanical engineering at Michigan State University. Hubbard, who has taught in MSU's engineering and medical schools, has developed a head-restraint device -- the HANS -- scientifically proven to prevent basal skull fracture and related head-whip injuries.

When a car crashes, usually against a concrete wall, the "rapid deceleration" -- the sudden stop -- creates surges of energy, many times the force of gravity, called "G-spikes."

Inertia, the tendency of mass to continue in motion at a sustained rate, tries to keep the body moving forward in the car. But the driver's safety harness (a system of lap and shoulder belts) stops the pelvis first, then the torso.

That leaves the head, which at 10-12 pounds is one of the heaviest parts of the body, in motion -- and in violent velocity relative to the car.

"The neck is now acting as the restraint, or the tether, for the head," Melvin says. And when the head reaches the end of its tether but strains to continue moving under tremendous "G-loading," the fatal damage occurs "right then and there," Melvin says.

The forces are trying to pull everything in the head and neck through the top of the skull. But the top is strong enough that whensomething has to give, the base -- the weakest part of the skull -- usually cracks. If it doesn't, then the neck usually breaks and/or severe brain-stem stretching occurs.

Basal skull fracture is "a structural failure of the base of the skull," Melvin says, "which, in and of itself, shouldn't matter much from the standpoint of threat to life."

But at the base of the skull, major blood vessels are grouped near the hole where the spinal cord begins to extend down from the brain stem. "The fracturing process itself, that crack going through the bone, can cause the blood vessels to rupture," Melvin says.

So drivers bleed to death, often within seconds.

"It's not a brain injury at all," Melvin says.

Many pathologists still think the head recoils violently, but Melvin's studies show it doesn't. And even if it did, it wouldn't matter. After the instant of greatest hyperextension of the head, and the accompanying whipping motion, it's too late.

High-tech Indy and Formula One cars carry on-board computer chips that measure and record the level of G-spike the cars undergo during crashes. There is not yet a precise measure for G-loading on drivers' bodies, but scientists believe it to be as much as 50 percent greater than the G-spike on the car itself.

NASCAR doesn't use crash recorders on its cars.