SPACE CENTER, Houston – NASA said Tuesday night that it had recovered a videotape showing four of the Columbia astronauts in the last minutes of their flight just before things went awry.

The 13 minutes of tape, which includes the space shuttle’s flight over the Pacific just before problems developed, shed no light on what went wrong, said an official close to the investigation into the Feb. 1 disaster. The astronauts are seen doing routine tasks in the cockpit, like putting on their gloves, and casually chatting, the official said.

The tape ends, because it was burned, four minutes after the start of Columbia’s atmospheric entry while the spaceship is still above the Pacific and flying normally. The first sign of trouble shows up in temperature monitors in the left landing gear compartment another four minutes after the end of the tape, the official said.

Reportedly seen on camera are the pilots Rick Husband and William McCool, flight engineer Kalpana Chawla and Laurel Clark. The three other astronauts were on the lower deck. Neither the official nor a NASA spokeswoman knew where, when or how the tape was found, but it was thought to have been recovered in Texas sometime in the past week.

Board members knew about the videotape for the past several days but did not discuss it at the weekly news conference Tuesday afternoon, the official said, because they wanted to give NASA time to show it to the astronauts’ families. NASA plans to release copies to the news media later this week.

Earlier Tuesday, the accident investigators said they wanted to know more about a mysterious object that almost certainly fell off the shuttle and was flying alongside the spacecraft during its second day in orbit.

The object orbiting near Columbia was never noticed during the flight. After the shuttle’s destruction over Texas, the Air Force Space Command began analyzing radar data that might shed light on the disaster and noticed the object.

Initially, NASA said it suspected the object might be frozen waste water dumped overboard or an orbiting piece of space junk that the shuttle happened to encounter.

But Air Force Brig. Gen. Duane Deal, a board member, discounted both possibilities Tuesday and said the object almost had to have come from the shuttle itself.

“You or I could invent a dozen scenarios,” Deal said. “It could have been something loose that separated, it could have been something inside the payload bay.” It also could have been part of the left wing, where all the overheating and other troubles developed during re-entry.

He described the object as about 1 foot by 1.3 feet in size and said it was flying in tandem with Columbia one day into the mission. It was within 50 feet of the shuttle and, within that first day, started separating farther and farther away until it burned up on re-entry three days later, he said.

“It’s not like my friend Rick Husband rendezvoused with a piece in orbit,” Deal said, referring to Columbia’s commander. “It was something that more than likely came loose.”

The composition of the object is unknown, but it was lightweight and not dense, Deal said. Lab testing is planned by the Air Force and NASA to determine the material, based on its reflectivity.

Columbia had just gone through a major maneuver in orbit Jan. 17, about 24 hours into its flight, when the object popped out of nowhere, Deal said. That suggests it could have broken loose from the shuttle during the maneuver.

Following the accident, Space Command staff went through reams of data to track the object until its atmospheric re-entry Jan. 20. Nearly 3,200 radar observations were made of Columbia during its 16 days in orbit.

“It’s been the most laborious examination that’s ever taken place in the history of Space Command, looking at every single one of those observations,” Deal said.

Because the astronauts did not do a spacewalk and did not have many windows, they would not have noticed the unidentified object, Deal said.

Meanwhile, a piece of a thermal tile, believed to be from the top of the left wing, remains the westernmost piece of debris found yet — and probably the earliest known fragment from its breakup.

The board’s chairman, retired Navy Adm. Harold Gehman Jr., said the fragment came from the upper surface of the left wing near the fuselage. It was found in West Texas, about 300 miles west of Fort Worth.

Gehman said he does not know how badly damaged the fragment is and stressed that it is too early to draw any conclusions from it.

But he held up pictures of another tile fragment found about 30 miles west of Fort Worth. It was dark gray or almost black with orange specks and extremely rough surfaces — heat damage that is much more severe than what is normally seen from shuttle tiles.

Engineers do not yet know whether the damage occurred during or after the breakup of Columbia, Gehman said. It is so badly damaged that investigators do not even know what part of the shuttle it came from.

Of the more than 8,100 pieces of shuttle debris recovered, about 5,300 have been identified, Gehman said.

Both NASA and the investigation board believe any wreckage west of Texas could provide hard evidence about what was happening to Columbia as it descended on its way to a Florida landing. The shuttle was 16 minutes away from touchdown when it disintegrated over Texas, killing all seven astronauts.

The 10-member board suspects the left wing was breached, allowing superheated gases to penetrate during re-entry. A central focus of the investigation is whether any of the debris from liftoff 16 days earlier caused or contributed to that breach.

Board member Scott Hubbard, director of NASA’s Ames Research Center, said computer analyses show that a hole of 20 square inches would account for the rapid 60-degree rise in temperature detected in Columbia’s left landing gear compartment during the final few minutes of flight.

What needs to be done next is a more sophisticated and complex analysis in which the hole is moved to various wing locations, he said.

Among the early tentative findings: the tires in the left landing gear compartment likely did not explode, though there was some disturbance going on in that area; the ship’s hydraulic systems failed in the final seconds of the doomed flight and the hydraulic fluid dumped out somewhere; and even though the power and guidance systems were still working up until the total loss of data and the fuselage was still intact, there were no signals from the left wing.