On November 21, 1996, at 0901 hours mountain standard time, a McDonnell Douglas (MD) prototype experimental helicopter, MD-600N (NOTAR), N630N, landed hard during a local test flight at the airport in Flagstaff, Arizona. Visual meteorological conditions existed at the time. The aircraft sustained substantial damage and the pilot was not injured. This aircraft was involved in a similar accident with the same pilot at the same location during flight tests on 11/04/96 (LAX-97-L-A034). According to the operator, the test pilot was performing a series of height velocity curve autorotations at maximum gross weight (4,100 pounds) at various altitudes and at zero airspeeds between 800 to 1,000 feet above the ground (agl). During touchdown on the fourth autorotation, the helicopter contacted the runway and displaced both skids with the right skid separating from the aircraft at the brace assembly connecting bolt hole. The fuselage was buckled and cracked along the right side and the bottom of the fuselage. The investigation revealed that the three previous test point autorotations were conducted at: 1,000 feet agl at 40 knots airspeed; 1,000 feet, 0 airspeed; and at 850 feet, 0 airspeed. There were no reported problems with these tests. According to MD engineers, the first two test points had "mild touchdown rates" (less than 1.5 g's and less than 5 feet per second). The touchdown rate for the third data point indicated a 1.75 g or 5.2 feet per second. It was discovered (after the accident) that the actual entry for this test was begun at 810 feet, instead of the 850 foot intended altitude. On the fourth test point autorotation, the test pilot was to be at an entry altitude of 800 feet agl and zero airspeed. Prior to entry, the pilot radioed a standard 3 second call to the ground crew that he was about to begin the test run. According to MD ground test data personnel reviewing the instrumentation plots, the entry did not occur until 8 to 12 seconds after the pilot's initial call. During this time, the aircraft had drifted down approximately 70 to 730 feet agl. A review of the video recording indicated the helicopter contacted the runway with the aft portion of the skids. The touchdown rate was about 4.0 g's or 13.5 feet per second. The design limit for this landing gear (skid) system was 6.5 feet per second. According to the pilot, the autorotation looked very similar to the previous data point until touchdown. At touchdown, the right gear collapsed and the helicopter dropped onto its right side. He stated the touchdown speed was approximately 45 knots. According to the test data, the horizontal speed did not get above 60 knots indicated airspeed, whereas, the other previous data points had horizontal speeds over 64 knots. The test data for this autorotation indicated a speed of 52 knots at touchdown. The investigation revealed the helicopter was equipped with a standard barometric pressure altimeter and a radar altimeter that was recently calibrated to a + or - 10 feet. Neither of these altimeters provide a digital readout and the altitude seen by the pilot is the needle position (analog) on the gauge. According to MD, there were no other precision instruments available to assist the pilot with altitude readout. Ground personnel monitoring the altimeter strip chart, which they had known to be inaccurate, noticed the aircraft "drift down" about 50 feet and assumed the pilot had already lowered the collective to begin the test after he had made the 3 second call. They did not inform the pilot of their observations. As a result of this accident and others, MD reduced the maximum operating gross weight to 3,650 pounds and installed a digital altimeter in the cockpit. The McDonnell Douglas Helicopter Company has claimed an exemption from public disclosure of the information contained in their accident report as privileged and confidential.