First they had to convince themselves that the idea was not ridiculous. Then they had to figure out how to do it. Then they had to invent the parts. Then they had to put them in space. And now, 40 years and $700 million later, they are just about ready to start.

Just about. NASA's Gravity Probe B, perhaps the most exotic space science project ever attempted, was launched into orbit April 24, but a series of technical glitches has delayed the formal beginning of its experiments for a month from the planned inception. These hurdles now appear to have been overcome, and the science phase should kick off in a week.

"The honest answer is we thought we knew everything in advance, and some things weren't as well planned as they should have been," said C.W. Francis Everitt, the painfully honest Stanford University physicist who has mid-wifed Gravity Probe B since birth. "The amount of fiddling we've been doing has been amazing."

Gravity Probe B proposes to prove two tenets of Einstein's Theory of General Relativity. The first, called "geodetic precession," or the "geodetic effect," holds that the gravity of large bodies such as Earth warps space and time near the body, much the way an iron ball lying on a rubber sheet stretches the sheet and causes nearby objects to roll toward it.

The second, known as "frame dragging," theoretically occurs when the rotation of a large body "twists" nearby space and time. Think of a silk scarf spread out on the floor, put an iron ball on top of it and turn the iron ball, twisting the scarf.

Demonstrating these effects in the real universe, however, is harder, for the angular distortion in space-time caused by Earth is predicted to be microscopic: 6,614.4 milli-arcseconds per year for the geodetic effect; 40.9 milli-arcseconds per year for frame dragging. A milli-arcsecond is about 1/4,000,000 of a degree of arc.

To measure this, a scientist needs only a telescope attached to gyroscopes. Point the telescope at a star, known as a "guide star," and lock on it so the telescope always has the star in its sights. Then spin up the gyros, so that their axes are also pointing at the guide star.

Then wait. If the instruments are properly calibrated and unaffected by friction, heat, movement, magnetic fields or other outside influences -- and, of course, if Albert Einstein was right -- then the gyros will drift over time to follow the space-time distortion.

If Einstein wasn't right?

"Well, at this point people would be very surprised if it finds something seriously unexpected," said Princeton University physicist Joseph H. Taylor. "In fact, it would be so surprising it probably wouldn't be believed."

The knock on Gravity Probe B is that theoretical physics, and some experiments, over the last 40 years have validated the two predictions to such a degree that time may have made the experiment moot.

"But the important point is that, in the end, our understanding of nature has to rest on experimental proofs, and that's what this is," Taylor continued. "There's no question that people will be very interested in the results, and if it achieves what it is supposed to do, the results will be cited for years."

The idea for Gravity Probe B was first entertained at the dawn of the space age by Stanford physicist Leonard Schiff, and, independently, George E. Pugh of the Defense Department. Space, they believed, offered for the first time the pristine medium necessary to set up a disturbance-free experiment.

Schiff and Stanford colleague William Fairbank began brainstorming the concept in 1959 and recruited aeronautical engineer Robert H. Cannon Jr. into joining them. In 1962, Fairbank induced the Britain-born Everitt, then a postdoctoral student at the University of Pennsylvania, to come have a look.

"I was the first full-time person on it," Everitt said in a telephone interview from Stanford, where he oversees the project. "There is a distinction between showing that a project is not ridiculous and then proving it can be done. After a couple of years, we had satisfied ourselves that it wasn't ridiculous."

Everitt, now 70, originally thought he would stay four years, but he never left and now supplies the project's institutional memory along with his expertise. Cannon is still alive, but Schiff died in 1971 and Fairbank in 1989.

In 1964, the team got its first installment of NASA money to start solving engineering challenges for which no answers then existed: how to build a perfectly spherical and therefore friction- and torque-free gyroscope; how to measure the direction of spin; how to spin up the gyros without permanently disturbing them. This, Everitt said, "is by no means obvious."

They also needed a tracking telescope "three orders of magnitude better than anything that had gone before," Everitt said. "We had to invent that." In the early 1970s, NASA conducted a formal "mission definition" of the project and found that nine separate new technologies would be needed to realize it.

The key decision in engineering Gravity Probe B was to embed the probe in a cement mixer-sized thermos bottle called a dewar, invented by Lockheed Martin Corp., and fill it with liquid helium cooled to -456 degrees Fahrenheit, less than 2 degrees above absolute zero.

"You want a mechanically stable system, and when you take it to low temperatures, the [heat] distortion problem, which is a tremendous one, disappears," Everitt said. "The Earth's magnetic field can also be a disturbance, but low temperatures . . . allow you to mitigate that as well."

The ping-pong-ball-size gyros on Gravity Probe B are the most perfectly spherical objects ever made, carved from cold-resistant fused quartz and coated with the metal niobium, a superconductor at low temperatures whose magnetic field lines up perfectly with the gyros' spin axes and allows the instruments to measure the expected drift.

As the liquid helium boils off, the spacecraft uses it in microthrusters that spin up the gyros and keep everything positioned properly. Everitt estimates it will take about 13 months for all the helium to boil off -- and that will end the experiment. Everitt said he would like to have at least 11 continuous months of data and preferably more.

All the major systems worked flawlessly after liftoff until the team tried to lock on to the guide star, IM Pegasus, about 300 light-years away in the northern sky. Dust motes "spoofed" the lenses of two wide-angle telescopes designed to locate the guide star for the tracking scope.

Then a couple of the microthrusters malfunctioned. And each time something was fixed, other things needed to be recalibrated. "It's been going through some typical early mission challenges," said Rex Geveden of NASA's Marshall Space Flight Center, the project's manager. "The mission is complex, and it's been worrisome, but we're exactly where we want to be now. We're very confident."

Ping-pong-ball-size gyros, the most perfectly spherical objects ever made, are critical to the experiment.