In connection with the connecting rod, emphasis must be laid on the importance of proper radii at all changes of section. The connecting rods for the first few Liberty engines were machined with sharp corners at the point where the connecting-rod bolt-head fits on assembly. On the first long endurance test of a Liberty engine equipped with rods of this type, failure resulted from fatigue starting at this point. It is interesting to note that every rod on the engine which did not completely fail at this point had started to crack. The adoption of a 1/32-in. radius at this point completely eliminated fatigue failures on Liberty rods. CRANKSHAFT The crankshaft was the most highly stressed part of the entire Liberty engine, and, therefore, every metallurgical precaution was taken to guarantee the quality of this part. The material used for the greater portion of the Liberty crankshafts produced was nickel-chromium steel of the following chemical composition: Carbon, 0.350 to 0.450 per cent; manganese, 0.300 to 0.600 per cent; phosphorus, 0.040 maximum per cent; sulphur, 0.045 maximum per cent; nickel, 1.750 to 2.250 per cent; chromium, 0.700 to 0.900 per cent. Each crankshaft was heat-treated to show the following minimum physical properties: Elastic limit, 116,000 lb. per square inch; elongation in 2 in., 16 per cent, reduction of area, 50 per cent, Izod impact, 34 ft.-lb.; Brinell hardness, 266 to 321. For every increase of 4,000 lb. per square inch in the elastic limit above 116,000 lb. per square inch, the minimum Izod impact required was reduced 1 ft.-lb. The heat treatment used to produce these physical properties consisted in normalizing the forgings at a temperature of from 1,550 to 1,600 deg.F., followed by quenching in water at a temperature of from 1,475 to 1,525 deg.F. and tempering at a temperature of from 1,000 to 1,100 deg.F. It is absolutely necessary that the crankshafts be removed from the quenching tank before being allowed to cool below a temperature of 500 deg.F., and immediately placed in the tempering furnace to eliminate the possibility of quenching cracks. A prolongation of not less than the diameter of the forging bearing was forged on one end of each crankshaft. This was removed from the shaft after the finish heat treatment, and physical tests were made on test specimens which were cut from it at a point half way between the center and the surface. One tensile test and one impact test were made on each crankshaft, and the results obtained were recorded against the serial number of the shaft in question. This serial number was carried through all machining operations and stamped on the cheek of the finished shaft. In addition to the above tensile and impact tests, at least two Brinell hardness determinations were made on each shaft.