On April 23, 1999, at 0645 hours local time, Samoa Air flight 906, a Dehavilland DHC-6-200, N719AS, veered off the runway and collided with a ditch and an embankment while landing at Fitiuta, Tu'a, American Samoa. The aircraft sustained substantial damage; however, neither of the 2 flight crewmembers nor any of their 12 passengers were injured. The aircraft was being operated under 14 CFR Part 121 by Samoa Aviation, Inc., when the accident occurred. The flight originated from Pago Pago, American Samoa, at 0609 as a regularly scheduled non-stop domestic passenger flight to Fitiuta. Visual meteorological conditions prevailed at the time and a composite IFR/VFR flight plan was filed. Upon arrival at Fitiuta airport, the captain overflew the uncontrolled airport to assess the current wind conditions. The windsock indicated winds of about 10 knots from 150 to 160 degrees. Based on those indications and past experience, the captain selected his landing direction and entered the downwind portion of the traffic pattern for runway 12 at an altitude of 1,000 feet agl, expecting a quartering crosswind from the right. On final approach, the flaps were lowered to 30 degrees and the before landing checklist was completed. At this point, the first officer reported a slight tailwind. The captain acknowledged the report, and continued the approach at "blue line" until touchdown abeam the taxiway. As the aircraft began its rollout on the runway, the captain pulled both power levers into beta and the aircraft began drifting to the right almost immediately. He corrected with rudder and braking but was unable to maintain directional control. The aircraft subsequently ran off the right side of the runway, across a ditch, and into an embankment. During the crash sequence, the nose section separated from the aircraft. The aircraft came to rest abeam the windsock. The captain shut down the aircraft and verified that none of his passengers had been injured. During the shutdown, the first officer noticed that the left beta light was not illuminated while the right beta light was illuminated. The passengers were deplaned and escorted away from the aircraft, then transported to the terminal. After deplaning, the first officer noticed that the surface winds were not blowing as indicated by the windsock, and were in fact a 60-degree tailwind for runway 12. Upon a closer inspection of the windsock, they noted that the pivot point at which the windsock was attached to the pole was rusted and the windsock would not rotate. According to the aircraft manufacturer, the landing roll distance for the aircraft's landing weight and a 10-knot tailwind would be 1,600 feet without the propellers in beta. A Federal Aviation Administration airworthiness inspector examined the aircraft after the accident. He reported that the beta pin at the cam cluster had backed out of position on the left engine. The beta pin, cotter pin, and washer were found in the bottom of the left engine cowling. The cotter pin had been cut but the severed piece was not recovered. The beta pin, washer, and cotter pin were transported to the Safety Board's Materials Laboratory for further examination. According the Materials Laboratory Factual Report, a scanning electron microscopic (SEM) examination revealed that the undamaged portion of the cotter pin's fracture surface exhibited elongated dimples, typical of overstress separations under direct shear loading conditions. Review of the maintenance records disclosed that the left engine had been changed 2 days prior to the accident. Interviews and written statements were obtained from the mechanics and the inspector involved in the engine change procedure. The lead mechanic and RII inspector reported that on the evening of April 21, 1999, he had inspected the beta pin during the No. 1 engine reinstallation and found that the pin was in place, and properly safetied and cotter pinned. Following the inspector's signoff of this area of the engine change, the two mechanics performing the work discovered that the teleflex cable was too short for the beta valve to be flushed. Together they informed the director of maintenance of the discrepancy they had found. He walked out to the aircraft and performed the required adjustments on the teleflex cable at the back and front himself, which required removing the beta pin installation safety wire. When he had finished re-rigging the teleflex cable he instructed both of the mechanics to complete the safety wiring of the engine installation. According to statements from the mechanics, one started at the front of the engine and the second at the back. The second mechanic, who had previously safetied the beta pin portion of the installation, believed that the back of the engine was completed and signed off by the inspector, and he did not look at the beta pin. Following completion of the work, the inspector was notified and he inspected the remainder of the installation, believing that he had already inspected and signed off on the beta pin portion of the work. After the inspector finished the inspection, the mechanics replaced the panels and prepared the aircraft for a leak check and ground run. Both checks were completed successfully and the aircraft was parked for the night.