All straightening operations on the Liberty crankshaft which were performed below a temperature of 500 deg.F. were followed by retempering at a temperature of approximately 200 deg.F. below the original tempering temperature. Another illustration of the importance of proper radii at all changes of section is given in the case of the Liberty crankshaft. The presence of tool marks or under cuts must be completely eliminated from an aviation engine crankshaft to secure proper service. During the duration of the Liberty program, four crankshafts failed from fatigue, failures starting from sharp corners at bottom of propeller-hub keyway. Two of the shafts that failed showed torsional spirals running more than completely around the shaft. As soon as this difficulty was removed no further trouble was experienced. One of the most important difficulties encountered in connection with the production of Liberty crankshafts was hair-line seams. The question of hair-line seams has been discussed to greater length by engineers and metallurgists during the war than any other single question. Hair-line seams are caused by small non-metallic inclusions in the steel. There is every reason to believe that these inclusions are in the greater majority of cases manganese sulphide. There is a great difference of opinion as to the exact effect of hair-line seams on the service of an aviation engine crankshaft. It is the opinion of many that hair-line seams do not in any way affect the endurance of a crankshaft in service, provided they are parallel to the grain of the steel and do not occur on a fillet. Of the 20,000 Liberty engines produced, fully 50 per cent of the crankshafts used contain hair-line seams but not at the locations mentioned. There has never been a failure of a Liberty crankshaft which could in any way be traced to hair-line seams. It was found that hair-line seams occur generally on high nickel-chromium steels. One of the main reasons why the comparatively mild analysis nickel-chromium steel was used was due to the very few hair-line seams present in it. It was also determined that the hair lines will in general be found near the surface of the forgings. For that reason, as much finish as possible was allowed for machining. A number of tests have been made on forging bars to determine the depths at which hair-line seams are found, and many cases came up in which hair-line seams were found 3/8 in. from the surface of the bar. This means that in case a crankshaft does not show hair-line seams on the ground surface this is no indication that it is free from such a defect.