Adding to the pressure, the space agency routinely hired and trained far more astronauts than could ever be accommodated on scheduled shuttle flights, the agency's inspector general reported earlier this year.

Consequently, astronauts now waited years longer than their predecessors for a shuttle flight. In the interim, they trained, handled engineering jobs and performed public relations functions.

In public, they were careful to display all the scripted spontaneity of Disneyland tour guides. About five a year resigned.

During the recovery operation, the astronauts took charge of everything the Columbia crew had touched, worn or used during the mission, including the twisted wreckage of the compartment that had sheltered them.

They sequestered the crew module wreckage in a locked corner of the reconstruction hangar at the Kennedy Space Center in Florida.

Most accident investigators were refused access. Computer files containing information about the module were encrypted. Photographs of the wreckage were locked away or kept on secure computers.

In all, searchers recovered about half of the crew module, according to the agency's internal reports.

It had been ripped apart by aerodynamic stress over about a half a minute — "tormented," one investigator said.

NASA launched an internal investigation of the crew wreckage and, for a time, kept it secret from everyone else involved in the reconstruction effort and the independent accident investigation.

No one would say whether the special handling of crew-related debris was driven by a sense of delicacy or shame.


Lacking comprehensive data from Columbia's onboard electronics, NASA accident investigators in February and early March had to rely on engineering intuition and technical analysis — informed guesswork.

Investigators were intrigued by a blurred image of the shuttle taken by two off-duty Air Force officers at the Starfire Optical Range in Albuquerque.

A volunteer — Julian Christou, a research specialist at the Center for Adaptive Optics at UC Santa Cruz — sharpened the picture through days of intensive computerized image enhancement, using techniques developed to clarify images of distant galaxies.

Even with his best efforts, the image of Columbia remained a smudge, but it revealed signs of an unusual disturbance around the leading edge of the left wing. It could have been caused by a crack, a dent or a tear in its skin.

Engineers at NASA's Langley Research Center looked at the data and wondered how that could match the only clues they had to work with: Columbia's last seconds of telemetry signals transmitted to Mission Control in Houston.

The signals showed four failing sensors in the wheel well and abnormal temperature readings from two sensors along the back of the left fuselage.

What damage near the front of the craft would cause a flow pattern that would affect temperatures at the rear?

"Whatever that damage was, it was moving the flow field around," said aerodynamics expert Bill Scallion, who has been with NASA since it was founded. "You get a tremendous amount of heating when you come in at 25,000 feet a second."

They tested their ideas with scale models of the shuttle in Langley's hypersonic wind tunnels among the groves of pin oak and pine outside Hampton, Va.