On January 31, 1994, at 1809 central standard time, a Fokker F28 MK 0100, N1437B, sustained minor damage during takeoff from Amarillo, Texas. The crew of four and 24 passengers were not injured. Visual meteorological conditions prevailed for the American Airlines Flight 851 scheduled Title 14 CFR Part 121 operation to Dallas/Fort Worth (DFW) Airport, Texas. An aircraft holding for takeoff on runway 22, reported to the Amarillo Air Traffic Control (ATC) Tower that the right main outboard wheel departed the aircraft. During the departure climb the tower informed the crew. The crew reported to ATC that all cockpit indications were normal and that they would continue the flight to the destination. At 1913, the crew landed the airplane on runway 35 left at the DFW Airport. The airplane was towed to the ramp without further incident. Examination of the aircraft components on February 1, 1994, by company maintenance personnel at DFW, and a Federal Aviation Administration (FAA) airworthiness inspector (enclosed report), revealed the outboard wheel bearing had failed in a manner that the bearing cage lost its ability to retain the rollers. Loss of the cage and rollers allowed clearance for the wheel to migrate outboard. The spacer was fused to the inner race of the outer bearing. On February 2, 1994, the components were reexamined by a FAA inspector and maintenance personnel at the company maintenance and engineering center in Tulsa, Oklahoma. The maintenance director for the facility commented that both bolts used to lock the nut to the axle were in place. On February 16, 1994, the components were examined by the investigator-in-charge, a FAA inspector, a company engineer, and the Aircraft Braking Systems representative. Physical evidence was consistent with a bearing failure. A ridge was seen on the damaged spacer. The engineer (enclosed report) reported that maintenance personnel may have installed the spacer backwards. He further stated that by installing the spacer backwards, damage to the grease seal could have occurred, allowing grease to migrate out, leaving the bearing without sufficient lubricant. The engineer was unable to demonstrate installation of the spacer backwards. The components were forwarded to NTSB metallurgists for examination. Metallurgical examination of the inside diameter of the spacer revealed heavy circumferential score marks indicating the spacer was rotating on the axle. The inside diameter surface also disclosed what appeared to be a ridge. "However, no measurements determining a position of the ridge and, therefore, the orientation of the spacer could be made." Examination of other components from the outboard wheel bearing disclosed evidence of moderate to severe overheating.