In the investigation of a selective action of homogeneous X-radiation on the allantoic membrane of the chick (Moppett, 1929), a constant time of exposure was adopted since one cannot assume that the " time factor " is expressed by a linear function. It was observed at an early stage that a considerable fluctua tion in current supply did not materially alter the biological reaction if the normal time of exposure were given. An experimental investigation was carried out using a wave-length of 0-5 A. which produces a hypertrophic reaction with an exposure of \ hour and an atrophic reaction in 1J hours, when other con ditions are similar. The reaction shows up abruptly when the above values are exceeded, suggesting the breakdown of a process of repair and the phenomenon is aptly described by the term threshold dose.Typical experiments are given in the following table (I) and the results are represented by the letter A for atrophy, H for hypertrophy, and HA for an intermediate stage. Experiment 147 may be taken as a standard in which a certain quantity of energy is given over a period of 1 hour, the reaction being HA. In experiment 135, approximately the same dose was given over a period of 2 hours and the reaction was increased to A +. The reliability factor 5/5 indicates that five consistent pairs of experiments showed an increased reaction in association with a longer time of exposure. Experiment 142 illustrates the fact that an increased reaction is obtained even if the exposure is given inter mittently over a long period of time.The above variation is opposite in sign to what may be termed a repair factor which is illustrated by experiments 306 and 308. A reaction to homo *