One of the positive recommendations of the Committee on Methods of Testing Hearing by Bone Conduction (chairman, Stacy R. Guild, Ph.D.)1 is "that tests of the hearing by bone conduction be made for as great a range of frequencies as possible with the available instruments, always with due regard to the possible sources of errors." In this paper we report the results of physical tests to which two common types of commercially available bone oscillators for audiometers were subjected in an investigation of some sources of errors. According to definition, the measuring of the threshold of hearing or any other sensation above threshold by means of an audiometer should be done with a tone of defined purity and of defined intensity. This holds for air as well as for bone conduction. The investigation was therefore concerned with two problems: First, can the intensity of the bone-conducted sound be determined sufficiently