Dashing hopes of finally launching shuttle Atlantis on a critical space station assembly mission today, one of the four low-level engine cutoff - ECO - sensors in the hydrogen section of the shuttle Atlantis' external tank failed to perform properly during initial tests after being submerged in supercold propellant. NASA managers said Saturday they would proceed with launch today if and only if all four sensors and associated instrumentation worked flawlessly and at 7:25 a.m., today's launch attempt was officially called off.

"Confirmation now that we have scrubbed for the day," NASA commentator George Diller said about a half-hour after the sensor problem became apparent. "That direction coming from LeRoy Cain, chairman of the Mission Management Team, and Doug Lyons, the STS-122 launch director."

Fueling began at 5:56 a.m. and by 6:48 a.m., the loading process transitioned to "fast fill." A few minutes later, Diller reported what engineers initially believed was good news.

"The liquid hydrogen sensors have been cycled and they are all confirmed to be working properly," he said. "Once again, at this time, all four liquid hydrogen sensors are working. This was somewhat predicted based on past tanking attempts when we had a problem with the sensors and then tanked a second time and the sensors have worked properly. So right now, this is consistent with our experience base. And all four hydrogen sensors are currently operating."

But moments later, sensor No. 3 apparently failed.

"One of the liquid hydrogen sensorsw, No. 3, has failed," Diller said. "Standing by for further action. ... This is not good news."

Engineers continued filling Atlantis' external tank to collect additional data. NASA has a flight rule that permits a launch if three of four sensors are working properly, but NASA managers said Saturday any problem with the sensors today would result in a scrub. And after additional data collection, that's exactly what happened.

Three of the four sensors, including No. 3, did not respond properly during fueling Thursday, derailing the first attempt to launch Atlantis. Because the system is suspect - and because repairs are not possible before the end of the current launch window Dec. 13 - NASA's Mission Management Team required all four sensors and associated instrumentation to be working properly today for the countdown to proceed.

The goal of Atlantis' mission is to deliver the European Space Agency's Columbus research lab to the international space station along with French astronaut Leopold Eyharts, who will replace Expedition 16 flight engineer Dan Tani.

Three spacewalks are planned to prepare Columbus for installation; to install two experiments on its hull; and to replace a spent nitrogen tank used to pressurize the station's ammonia cooling system. The nitrogen tank and a failed gyroscope will be moved back to the shuttle for return to Earth.

But NASA only had until Dec. 13 to get Atlantis off the ground because of temperature and power issues related to the angle between the sun and the plane of the space station's orbit. The window reopens Dec. 30, but any delay past Dec. 13 would push the flight into January to avoid countdown software issues due to the end-of-year rollover.

Engineers cannot access all of the ECO sensor circuitry at the launch pad, raising the possibility of a rollback to the Vehicle Assembly Building to remove the orbiter from the external tank for more extensive troubleshooting. But no such decisions have been made as of this writing.

The ECO sensors are part of a backup system intended to detect low fuel levels in time to prevent the shuttle's main engines from draining the tank after some other problem that might cause the powerplants to use up hydrogen faster than expected. Running out of fuel during engine operation likely would trigger a catastrophic failure.

In the wake of earlier problems with the cutoff sensors, NASA implemented a flight rule change that would permit a countdown to proceed if three of four sensors were operating normally and certain other conditions were met.

But engineers do not yet know what caused two sensors to "fail wet" during fueling Thursday or why a third sensor did the same during de-tanking operations. Because the system is suspect, the flight crew office proposed proceeding with launch Sunday if, and only if, all four sensors are operating normally.