Nine seconds after ignition, a small snap ring broke deep inside the space shuttle engine.

It may have been the simplest, cheapest component in the $8.9 billion shuttle program. The ring held a plug-like gas jet in place near a sensor that reads pressure in the combustion chamber.

Small, but vital. The failure of the snap ring during an engine test last month -- combined with a chance power failure at the shuttle test site -- triggered a series of mishaps that ended with a serious engine fire.

The engine's complex oxygen pump and control system were destroyed. "We lost most of it," said Walter Dankhoff, director of engine programs for the National Aeronautics and Space Administration.

The engine was a spare, not one of the the three main engines mounted in the shuttle Columbia, still under construction at the Kennedy Space Center at Cape Canaveal, Fla. And there is plenty of time to remove the snap rings from the flight engines and improve the design before the shuttle's first flight, now scheduled for next March.

But this accident is the latest in a long series of engine fires, faulty welds and parts problems that since 1977 have plauged the shuttle, the heart of the U.S. space program for the balance of the century.

The shuttle program already is two years behind schudule and 20 percent over budget, NASA still expects the shuttle program to put the next generation of military surveillance and civilian communications satellites into orbit and make possible space experiments on topics ranging from the origins of human life to the fate of the universe. There are a lot of eggs in the shuttle program's basket.

The latest mishap feeds the nagging worry that no matter how close NASA and its contractors come to perfection, there always will be a new undetected problem lying ahead to jeopardize the shuttle's mission.

The space shuttle is designed to be a crowning achievement of U.S. technology. Unlike the Saturn rockets that lifted astronauts toward the moon and then burned up in the atmosphere, the shuttle is designed to go into orbit, discharge its cargo of satellites and return to a gliding landing on earth. It is part rocket, part satellite, part aircraft.

To lift the 153,000-pound shuttle, the size of a DC9 arliner, NASA and Rocketdyne had to create a new engine, far advanced as compared to anything used in the moon program. Specially designed pumps push a torrent of liquid hydrogen and liquid oxygen to the combustion chamber, producing an explosive discharge of superheated steam through the engine nozzles at 1,000 pounds per second.

"These pumps we are talking about are roughly the size of trash can and the amount of energy being released during combustion is equivalent to 5 million horsepower. That gives you a sense of the magnitude of the challenge that NASA and its contractors are facing," said Eugene Conant of the Massachusetts Institute of Technology, head of an independent advisory committee on the shuttle engines, in testimony before a Senate Subcommittee.

The pumps are a telling example of NASA's dilemma in the shuttle program.

Denied a blank check for the shuttle program, NASA devised a reusable shuttle unlike earlier rockets that would carry large payloads into orbit relatively cheaply. But that requirement forced the design of light, compact, but durable parts -- often creating conflicting technological demands that pulled project engineers in opposite directions.

The tight budget has forced NASA to follow a riskier "success-dependent" strategy, as Conant calls it. In the moon program, spare parts were abundant and individual components were painstakingly tested under operating conditions before assembly. In the shuttle program, part-by-part testing has been reduced to save time and money. Instead tests were run on large assemblies containing many untested parts.

Had the engine been free of bugs, the strategy would have worked, but the engine was not, and the delays to redesign and construct replacements for faulty parts have added several years to the program, Conant and his committee say.

The fire that nearly destroyed the test engine last month is an example.

As Dankhoff explains it, the problem developed at the end of a tube that leads from the combustion chamber.At the end of the tube is a pressure plate that "reads" the pressure of the burning hydrogen gas. The physical force of the gas pressure is translated into an electrical signal that goes to the engine's programmed control computer.

Nearby, however, is a storage chamber for liquid hydrogen, cooled to zero degrees Fahrenheit. A tiny amount of hydrogen gas is permitted to seep from the chamber in order to control the pressure coming from the combustion chamber -- the amount is controlled by a plug-like part called a Lee Jet. The snap ring, like those used in portable dishwashers and other appliances, holds the jet in place.

When the ring broke, the jet slipped from its proper place. "The hydrogen wasn't dribbling in, it was flowing," said Dankhoff in an interview. The pressure sensor, feeling the unexpected weight of the hydrogen, began to "read" tremendous pressures coming from the combustion chamber and relayed this erroneous information to the control computer.

In response, the computer ordered the supply of oxygen to the combustion chamber cut back farther and farther. "The pumps and turbines were running at much lower temperatures and lower speeds than they were designed to," Dankhoff.

Finally, the pump failed, oxygen leaked and came in contact with hot metal, which ignited.

"Unfortunately, this happend in the same area as the electrical control panel. We lost all of it, I'm afraid," Dankhoff said.

The same accident could not have happened on the launch pad, he said, because the abnormally high reading from the pressure sensor would have been contradicted by a normal reading from a second, backup sensor. The computer would have heeded the normal reading and disregarded the other. That did not happen during the test because the power to the second sensor was off.

Had the accident happened on the launch pad, the flight would have been automatically aborted, Dankhoff said.

"We would have had an automatic shutdown after about seven seconds," he said.

The snap ring is being replaced and the fitting redesigned, to prevent the same combination of mishaps from causing a fire in flight.

Dankhoff says he is concerned, "but what makes me less concerned is that it wasn't a major functioning part. The accident happened in a secondary system. That doesn't justify the design, but it keeps it from being a major problem.

"It's clear the engine has greatly advanced the state of the art and has had the kinds of problems you'd expect. We feel we're over the hump on the basic problems," he said.

The various shuttle engines have so far compiled 84,542 seconds of firing time (a typical mission requires about 500 seconds of combustion), about half of that at the power level required for the first shuttle flight.

Why did the snap ring break? Dankhoff did not offer a detailed reason, but spoke of NASA's difficulty in getting reliable parts. "We have so many vendors and suppliers . . . Reaching back and assuring that we have good quality materials and parts are tough."

In one particularing troubling foul-up, for example, welds throughout the engine had to be inspected for craks after the discovery that a supplier had mistakenly shipped the wrong kind of solder to the contractors.

"In the end, it depends on the individuals (who produce the parts) and their motivation," he said.