NASA officials yesterday postponed the flight of the space shuttle Discovery until late next week -- at the earliest -- to give engineers time to find and fix a mysterious anomaly that caused a hydrogen fuel sensor to malfunction on the launchpad Wednesday, scrubbing the first shuttle flight in 21/2 years.

Twelve teams of troubleshooters from around the country began a race against time to solve the sensor problem in time for Discovery to launch by the end of July. If their work fails to produce a solution by July 27, NASA may have to delay the flight until the next launch window opens in September.

"The simple things we did quickly did not provide any resolution of the problem," Wayne Hale, deputy shuttle program manager, told reporters in a televised news conference. "We are going forward on a day-by-day basis, and as soon as we find the problem we will fix it, and as soon as we fix it we will be four days from launch."

Hale said Mission Commander Eileen Collins and Discovery's other six astronauts are remaining at the Kennedy Space Center over the weekend on the chance that the launch could take place at the end of next week, "but that would require a near-term lucky find."

Still, he added, "we are not pessimistic about making the July window. We are here for the duration."

Discovery, with all seven astronauts aboard, was fueled and ready on Kennedy Space Center's Launch Pad 39B less than three hours from liftoff when the launch team ordered a test of four hydrogen sensors located on the bottom of the shuttle's 1.6 million-gallon external fuel tank.

The sensors, about the size of small candy bars, are used to trigger a main engine cutoff if they detect that the tank is running out of fuel prematurely. The suspect sensor failed to respond to a test calling for a "dry" reading and instead locked in the "wet position." Under NASA's launch policies, a malfunction in any one of the four sensors requires the launch be scrubbed.

Discovery's long-awaited flight was the first since the shuttle Columbia disintegrated during reentry Feb. 1, 2003, and the scrub was a sharp disappointment for NASA as it seeks to open a new era in space travel aimed at returning humans to the moon by 2020 and eventually sending them on to Mars.

The agency made many major changes to the shuttle to enhance its safety features after the Columbia disaster. John Muratore, manager of shuttle systems engineering, said side effects from these changes are "prime candidates" as possible causes of the malfunction in a sensor system that had flown almost without mishap for more than 20 years.

Muratore is heading the teams of troubleshooters, who have been given tasks including analyzing everything new on the spacecraft, checking thermal conditions aboard the orbiter and reviewing the history of the telephone-book-size "black box" that serves as the sensor system's controller.

The job promises to be difficult. Problems with the hydrogen sensors first arose during a countdown simulation in April when the external fuel tank was filled with liquid hydrogen and liquid oxygen, chilled to several hundred degrees below zero.

Two sensors misbehaved during this "tanking test," toggling intermittently between "wet" and "dry." Extensive analysis failed to reveal the cause, so engineers changed everything in the system -- the controller, the wiring and even the sensors. Wednesday's fuel tank was not the same as the one tested in April.

"It is not at all clear that the problem [Wednesday] is linked to the problem we had in the first tanking test," Muratore said. "On the other hand, it wouldn't surprise me if it was."

Unable to determine what was wrong, engineers decided to try to fly in July with an "unexplained anomaly." Hale said planners are now committed to finding the cause of the problem and fixing it, but he did not rule out the possibility that they may end up facing the same hard choice they had earlier.

"If you come to the end and it's still unexplained," he said, "then you have to sit down and have a discussion about the rationale to go fly."

Discovery rests on the launchpad as NASA engineers look for the reason behind a malfunctioning fuel sensor.