By MATTHEW L. WALD

WASHINGTON, Oct. 15 With hearings into the crash of American Airlines

Flight 587 set to open later this month, investigators have turned up new

information that has led them to focus more intensely on pilot training and

performance as they try to explain the disaster that left 265 dead in Queens

last November.

The new information, investigators say, includes testimony that the co-pilot

at the controls of the plane had a history of overreacting to wake

turbulence, which the Airbus A-300 encountered shortly after taking off from

Kennedy International Airport on Nov. 12. Investigators have also discovered

that the plane's rudder was prone to make large movements with very little

effort by the pilot, meaning a pilot could inadvertently swing the plane

wildly from side to side.

Officials at the National Transportation Safety Board, who are preparing to

convene hearings on Oct. 29, are still far from concluding that the crash

was caused by pilot error; they will not establish a probable cause for

months, and the role of the co-pilot's actions is still not completely

clear.

When taken together, though, the new information has greatly increased

officials' interest in the actions of the co-pilot, Sten Molin, as they try

to understand how, for the first time in 40 years, a commercial jet lost the

vertical portion of its tail, dooming all on board and five people on the

ground in Belle Harbor, Queens.

Investigators are focusing on the manipulation of the plane's rudder by Mr.

Molin, who was the flight's first officer. The rudder is attached to the

vertical portion of the tail.

Before Mr. Molin flew A-300's he flew Boeing 727's for American, and after

the crash, one investigator said, a captain with whom Mr. Molin had flown in

1997 notified investigators that he was concerned with the way Mr. Molin had

reacted to wake turbulence on the 727's. According to the investigator, who

spoke on the condition of anonymity, the captain said Mr. Molin was "awfully

aggressive." According to another investigator, the captain said Mr. Molin's

handling of the 727's was "abrupt."

Investigators say the plane in this crash left Kennedy 2 minutes and 20

seconds behind a Japan Airlines 747-400, slightly longer than the minimum

that should separate planes as they take off, and ran over its wake.

An airplane leaves two wakes, each a horizontal tornado beginning at the

wingtip. The A-300 hit the first uneventfully, but seven seconds after it

hit the second, which was 93 seconds after liftoff, it lost the vertical

portion of the tail. Investigators believe Mr. Molin pushed the rudder as

far as it would go in each direction twice, a movement that one investigator

said could "break the airplane."

Among the issues that will be explored at the hearings is the performance of

what is known as the plane's automatic rudder limiter, a device that is

supposed to prevent the rudder from being moved farther than is safe at its

air speed at the time.

To the surprise of pilots, the accident has shown that the rudder limiter

may prevent the rudder from being pushed too far in a single direction, but

does not protect against damage if it is pushed in alternating directions.

Shortly after the crash, the safety board, an advisory body, and the Federal

Aviation Administration issued a warning against flipping the rudder back

and forth.

A second question concerns the ease with which the rudder could be

manipulated. So little force is required that it would be hard to make only

a partial movement of the control, some experts said, especially since it is

moved by the pilot's leg.

The rudder's function is to let a plane land or take off in a cross-wind,

and to hold a straight course if an engine fails. Generally, pilots of big

jets seldom use the rudders in flight, but American trains them to do so in

certain circumstances.

Partly in response to a previous crash involving a Boeing 737 belonging to

USAir near Pittsburgh, American puts its pilots through a course called the

advanced aircraft maneuvering program. In a simulator, pilots are trained to

recover when the airplane is at various extreme angles, and they are taught

to use the rudder in the maneuvers.

Now, investigators are considering whether the sensitivity of the rudder

control may mean that the rudder will usually move abruptly. Some experts

think the pilot may have pushed the rudder all the way in one direction,

realized that he had gone too far, pushed all the way back in the other, and

then repeated the process in an oscillation that destroyed the plane in

flight.

Another question, about the tail itself, seems to have faded. Immediately

after the crash, investigators focused on the tail because it is made of

carbon-fiber composite and not the traditional aluminum, a material with

which they and the airlines have more experience. There was some thought

that the tail might have been more vulnerable to wear and tear.

Because the hearings have not begun, people involved in the investigations

said they could not speak for attribution.

At this point, investigators believe that the tail functioned as designed,

although the witness list for the hearing includes specialists who will

testify about whether the design itself is adequate.