ASHINGTON, March 31 — Newly recovered data from a recorder on the shuttle Columbia raise doubts that the much-discussed protective tiles were the site of the flaw that led to its destruction, the chairman of the panel investigating the accident said today.

Instead, the data focus attention on the reinforced panels at the leading edge of the shuttle's left wing, said the chairman, Harold W. Gehman Jr., a retired admiral.

The new information, gleaned by 100 NASA engineers who spent the weekend laboring over data recovered from an old-fashioned magnetic tape, does not undermine a leading theory of the accident — that damage to the left wing, perhaps from debris that fell off the shuttle's external fuel tank during liftoff, doomed the orbiter.

But if the problem is eventually demonstrated to have been in the panels, made of reinforced carbon-carbon, then a hurried analysis of possible tile damage by Boeing, a NASA subcontractor, during the Columbia's 16-day mission, would appear to have been misdirected. During the mission, engineers on the ground worked frantically to determine whether the tiles could have been damaged enough to endanger the orbiter, but spent little time discussing the panels on the wings.

Still, if officials had asked for a photographic examination of the tiles by ground-based telescopes or spy satellites during the mission, that might have revealed the damage to the leading edge.

What could have been done to repair the part, or reduce the load during re-entry, is not clear. But pinpointing the location of the breach now would help NASA engineers determine what improvements can be made to the surviving fleet.

Admiral Gehman also said today that the new data hinted that the shuttle already had severe damage when it began its re-entry, and not a minor flaw that was made worse by re-entry. Engineers had theorized that minor damage to the thin layer of protective silicon carbide on the panels could have allowed hot oxygen to begin eating away at the leading edge, but Admiral Gehman's comments suggest that this is unlikely.

Damage before re-entry is likely, he said, because the data show extreme heating taking place early, while the force of air passing over the wing was still quite weak. Admiral Gehman spoke in a conference call with reporters this afternoon.

The new information "certainly leads us away from things like the tile and the landing gear doors," he said. He added, "It kind of leads us toward the leading edge."

The recorder, which contains data from hundreds of sensors that measure heat, vibration and strain, was recovered two weeks ago near Hemphill, Tex. Admiral Gehman said it showed that drastic heating began in the left wing five minutes after the Columbia entered the outermost traces of the atmosphere. That is about three minutes before an automatic system on the Columbia radioed to the ground that heat was building up on a component in the well for the left main landing gear.

Engineers now believe that the Columbia did not have "some kind of a weakness, that some part of the orbiter was ripped off when it entered the atmosphere," Admiral Gehman said, because the air pressure is so low at the altitude where the heating was observed, probably 270,000 to 280,000 feet.

"This looks to us like it probably had a pre-existing condition," he said. "The orbiter probably had this before she ever started this entry."

Investigators are working first with readings from 420 sensors, including 35 heat sensors. Two that have attracted immediate attention are near Panels 9 and 10 on the forward edge of the left wing. A sensor directly behind Panel 9, attached to its support structure, began heating up at 8:44:09 Eastern time on Feb. 1, about 14 minutes before the orbiter broke up, Admiral Gehman said.

The temperature on a sensor behind the leading edge rose to 450 degrees Fahrenheit, and then the sensor, or the wiring that connected it to the recorder, failed. Admiral Gehman said he assumed that the temperature began far below zero.

In contrast, the temperature in the left main landing gear rose to 140 degrees, from 80.

In such a thin atmosphere, he said, heat transfer is slow, and "you've got to have a ton of heat" to cause such a temperature rise in a sensor.

A drawback in using heat sensors in a crash investigation, he pointed out, is that they do not indicate what direction the heat is coming from. But data from strain gauges, which has yet to be analyzed, could help show what parts of the wing were breaking up first, he said.

"I would hope we'll get enough of these little signposts that they'll line up and point us toward something," he said.