An electronic glitch in the equipment that is supposed to allow a pilot on the ground to fly a full-sized jetliner by remote control has forced postponement of a federally financed airplane crash here.

The crash, a five-year, $11.8 million effort by the Federal Aviation Administration and the National Aeronautics and Space Administration, will test several design concepts to see if planes can be made safer so passengers will have a better chance of surviving a crash.

The major experiment will be to test the behavior of a fuel additive that is supposed to prevent or dramatically retard a fireball after impact. Other experiments deal with new types of seats, basic structural components such as floors and bulkheads, or walls, and the effectiveness of fire-retardant fabrics in the passenger cabin.

The crash was scheduled for early Saturday but was postponed late Thursday when NASA and FAA officials decided that they needed more data before entrusting the plane and its cargo of 75 heavily instrumented, human-like dummies to remote control for the first -- and last -- time.

FAA and NASA safety experts regard the crash as a one-time golden opportunity to learn a great deal, and they want their airplane to crash the right way in the right place, a carefully selected corner of Rogers Dry Lake here. The agencies hold that humans should be able to survive the crash as designed.

M. Russell Barber, NASA's program manager, said the electronic problem was discovered during a test Monday while the plane was on the ground. For 10 seconds, he said, the remote-control signal to the airplane was interrupted. If that occurred in the air, it would force NASA to crash the plane the "wrong" way to keep it from straying into inhabited areas.

Barber said he believed that the problem was in a digital decoder on the airplane and that it was repaired. Nonetheless, he said, "we want a higher level of confidence" before the plane is turned loose under remote control. Late Thursday he ordered another flight test of the repaired system with a crew on board.

That next manned test flight of the plane, a 24-year-old, four-engine Boeing 720, is scheduled sometime next week. The crash will be rescheduled depending on the results of that flight.

John Reed, the FAA's project manager, said in an interview that he regarded the remote control system as "a major experiment." The 720, which looks much like the more familiar Boeing 707 and would normally carry about 135 passengers, is the largest aircraft anyone has tried to fly by remote control, he said.

NASA has considerable experience in flying smaller craft by remote control, and it designed the control system for this experiment. A pilot on the ground "flies" an electronic cockpit that resembles a real cockpit, and the control inputs are transmitted electronically to the plane's autopilot.

NASA and FAA employes have worked for months outfitting the plane with electronic equipment and the instrumented human-like dummies. The information those instruments feed crash-proof, on-board recorders and ground-based telemetry receiving equipment is expected to give safety researchers years of work. The event has the official label of "Controlled Impact Demonstration" (CID), but several people have claimed credit for renaming it "Crash in Desert."

Officials here consider it unlikely that the crash will be late next week, because Edwards is the alternate landing site for the space shuttle Discovery, scheduled to return to Earth Friday. All of this vast desert base's facilities will be turned over to shuttle operations.