On July 1, 1995, at 1150 eastern daylight time, a homebuilt VARI-EZE, CGMEZ, was substantially damaged during a forced landing to the Malone-Dufort Airport (MAL), Malone, New York. The private pilot was not injured. Visual meteorological conditions prevailed, for the personal flight that departed MAL, at 1130. No flight plan had been filed for the flight conducted under 14 CFR Part 91. In the NTSB Form 6120.1/2, the pilot stated that he had completed three loose formation passes over runway 23, with another homebuilt airplane. He added power at the completion of the last low pass, and noticed a vibration 1 to 2 seconds later. This was followed by a loss of propeller thrust and high engine RPM. The pilot secured the engine, and due to personnel to his left, and trees directly ahead and to the right, he initiated a turn back to the runway. At the completion of the 180 degree turn, he observed the airplane from the formation flight approaching head-on. The pilot then performed a landing to a rough field, adjacent to runway 23. During the landing roll, the main landing gear and left wing separated from the airplane. The pilot stated that initial examination of the wreckage revealed that the propeller shaft had failed at the junction of the middle bearing inner ring. The pilot further stated: Prior to removal of the PSRU [propeller speed reduction unit], a professional mechanical engineer and several other knowledgeable people inspected the failed PSRU. The PSRU vendor was also consulted by phone. It was agreed by all...the cause was bending induced fatigue failure, induced by thermal expansion of the PSRU back plate. This caused the front bearing to rise while the center and rear bearings were held rigid by the common housing. This induced a slight bend in the shaft with the stress focused right at the center bearing inner ring. The PSRU was sent to the Transportation Safety Board (TSB) of Canada, for further examination. The TSB of Canada report stated: ...the shaft failed in a progressive manner from fatigue cracking at the stress concentration provided by the change in shaft diameter...The material of construction was shown to be a unalloyed steel, in the normalized condition, resulting in a relatively low strength material for the application. The fatigue resistance of the part would certainly have been improved by the use of the intended AISI 4140 alloy material... The airplane was built in 1984, and flown 566 hours with a Lycoming, O-234 engine. In 1994, the Lycoming engine was replaced by a Subaru, EA-81. In addition, a synchronous belt PSRU was installed. This was a "two bearing" unit, that drove the propeller at a ratio of 1.84 to 1. The airplane accumulated another 125 hours with this installation. During the winter of 1994/1995, the PSRU was replaced with a one-of-a-kind unit. This unit utilized a third bearing to provide support for a longer propeller shaft. According to the pilot's statement, the new unit was ground tested for 8 hours, and flown for 49 flight hours, prior to the failure.