The question of the relationship of the virulent diphtheria bacillus to avirulent forms of similar morphology and fermentative properties has always attracted the interest of bacteriologists. Practically all careful observers agree that avirulent bacilli can never become virulent either in vivo or in vitro. It has been also the experience of most observers that no virulent strain can be made avirulent by laboratory methods. Crowell (1926) and Cowan (1927) have however reported that they had succeeded in isolating avirulent variants from a pure culture of a virulent strain. From the nature of Crowell's experiment, which only records one such change, it seems possible that he may have chanced on a true biological (i.e. uncontrolled) variation, but Cowan suggests that it might be possible to isolate with some degree of regularity “rough” avirulent colonies from virulent cultures. As Cowan points out the procedure is both lengthy and difficult. My own experiments in this direction have been entirely negative, although I have on many occasions tried to isolate avirulent variants from virulent cultures by colony selection and other methods. I have not even been successful with the cultures with which Miss Cowan worked and which she very kindly put at my disposal. Individual colonies of the Park 8 strain have been also tested on more than 700 occasions and have always proved fully virulent when pure. Passage both through a large variety of artificial media and through animals failed to engender avirulent variants.