The National Transportation Safety Board said yesterday that a crack in an engine part that led to the crash-landing of a United Airlines DC-10 jumbo jet in Iowa last year probably could not have been detected by metallurgical tests now used on operational engines, and urgently recommended that new tests be developed for engines still in service.

The board said the defect in the en- gine part could have been detected at the time of manufacture -- and the crash possibly prevented -- if another test performed during assembly had been used properly.

The board's recommendations covered the General Electric CF6-6 engines which were used in the DC-10-10, the early version of the McDonnell Douglas jumbo jet. A total of 137 DC-10-10s remain in service. The remaining 290 later-model DC-10s use different engines.

Spokesmen for General Electric and the Federal Aviation Administration said they were not informed in advance of the board's recommendation, which was released late in the day. "We give safety board items a high priority, and we will study this one," said an FAA spokesman.

The part found to have caused the Iowa crash was the fan rotor disk at the front of the engine, which holds the fans that draw air into the engine. The disk, made of titanium steel, developed a microscopic crack because of a metallurgical anomaly, the safety board said, and it blew apart when the crack finally gave way.

United Flight 232 crashed July 19, 1989, at Sioux City, Iowa, killing 112 people, after the DC-10's tail-mounted engine exploded in flight, sending shrapnel through the vertical tail fin and disabling the hydraulic system that operates the jet's controls. However, 184 people survived the flight because the crew was able to steer the plane by manipulating the power in the two undamaged wing-mounted engines and almost succeeded in landing the plane.

Engine failures, while not common, occasionally take place when turbine blades deeper in the engine give way. Usually, these failures are "contained" within the engine, although they occasionally penetrate the engine casing. However, the Sioux City crash involved the only known failure of a fan rotor disk.

The board, in a recommendation labeled "Class I, urgent action," said that two current tests -- one involving ultrasonic waves and the other fluorescent light -- may not be adequate.

A United employee performed the fluorescent test on the disk that gave way just 760 flights before the Sioux City crash, but did not detect the crack although it was large enough by then to have been detectable. The board said this led them to believe that the test is inadequate.

The ultrasonic tests were ordered on all CF6-6 engines after the Sioux City crash.

The board said that General Electric and the FAA should develop other tests, perhaps involving use of electrical currents.

The FAA also should change its method of certifying all rotating parts of aircraft engines, the board said, and should "no longer assume they are free of defects when new."

In another recommendation involving manufacture of the titanium disks, the board said that inspections should be performed after the disk is machined into its final shape, not when it is in its intermediate machined shape, as is done now.

The board "believes that the metallurgical anomaly that led to fatigue cracking and separation of the disk could have been detected during manufacture" if the test had been performed at the right stage.