Discussion and Summary: Experimental evidence has been presented confirming previous observations that skin hypersensitiveness can be set up in a healthy animal by injection of heat killed tubercle bacilli. It is true that a large amount of the material must be injected in order to accomplish results similar to those obtained when living organisms are used, but it must be remembered that dead organisms do not propagate and that the antigen is soon completely used up. On the other hand, living organisms grow and continuously renew the supply of antigen. The skin hypersensitive state established by killed bacilli is not as transitory a phenomenon as has been thought by many, but may persist for a year or longer. The degree of immunity established by this method of sensitization cannot be ignored as insignificant. From our experiments we cannot agree with Uhlenhuth, Selter, Bruno Lange, Dold and others that there is no appreciable protection. We believe that the failures of these investigators to demonstrate protection have been due partly to the large number of infecting organisms and partly to the method of infection used in testing the immunity. We do not believe that an absolute immunity against virulent tubercle bacilli can ever be obtained no matter what method is used, i.e. the use of living virulent, living avirulent or dead organisms. But we may ask, do we need such an absolute immunity? Tuberculosis is rarely an acute disease. In the majority of instances it is benign and in many cases is a slowly progressive disease whose activity can be easily arrested. So far as the dosage of infection in infants is concerned, we feel that excessive doses are uncommon. The mere fact that a large number of children become infected without any clinical manifestation of disease leads one to believe that the numbers of bacilli entering the body during infancy or adult life are relatively few and that there must be a normal, possibly non-specific resistance which by some method can be reinforced and the body become more resistant to small numbers of organisms. In infancy and in childhood when the dosage is excessive, the normal mechanism of resistance is completely overcome and progressive fatal disease results. On the other hand, if the organisms are few in number, a period of approximately three months may elapse before the child becomes positive to the tuberculin skin test. This has been confirmed by animal experimentation. Guinea pigs infected with a few bacilli (5 to 10) become positive to the tuberculin skin test in 70 to 90 days, a period of time approximating that observed in ordinary human infection. From this we may assume that it is unreasonable to test the immunity by inoculating the sensitized animals with very large numbers of infecting organisms. Let us for a moment consider what has been the usual procedure for infection in the past. A suspension of tubercle bacilli was weighed and various dilutions made. The dosage was determined by weight1 and from 0.01 to 0.001 mgm. of organisms was used to test the immunity of each animal. Comparing this method and the amount with those described in this study, it can be seen that formerly each animal was inoculated with a sufficient number of organisms to kill thousands of guinea pigs. It is not surprising that there was no demonstrable immunity in animals sensitized with dead tubercle bacilli. By careful regulation of the dosage of infecting organisms we have proved to our satisfaction that animals sensitized with dead bacilli and then infected with 440, 4,400 or 44,000 virulent organisms, outlived the controls. Furthermore, the sensitized animals which were infected with 44,000 organisms outlived the controls which were inoculated with only 440 organisms,—a hundredfold protection. Our experiments corroborate those of H. Langer, who found that within three months after infection none of his sensitized guinea pigs had died from tuberculosis, while during this period 33 per cent of the controls died from definite tuberculosis. At the end of the ninth month 66 per cent of the sensitized and 100 per cent of the control animals had died from tuberculosis. The best results were obtained in the experiment where the infection was made by intracardiac inoculation. It seems that hematogenous infection did not take place in the sensitized animal as readily as in the controls. In all probability the bacilli came more directly in contact with the defensive mechanism present in the sensitized animal. This method of infection is under consideration at present. Perhaps the discrepancies reported on the value of immunity obtained with dead tubercle bacilli can be explained if one considers the phenomenon of microbic dissociation. We may recall Tillett's (13) experiments with “R” and “S” colonies of Pneumococcus III, where he found that when living “R” colonies were inoculated into rabbits the natural immunity was reinforced. This could not be obtained with heat killed “R” colonies. On the other hand, heat killed “S” colonies accomplished the same purpose as the living “R.” This probably suggests that the antigenic properties of the two colonies were different. Probably the same may be the case with the tubercle bacillus. Recently Petroff (14) and also Petroff, Branch and Steenken (15) have pointed out that in every culture of tubercle bacillus there are two or more types of colonies. These colonies not only presented different structural form and other biological characteristics, but chemical analysis also revealed that the two colonies differed considerably and that one grew better in acid media and the other more profusely in the alkaline range. The type of colony in a culture will depend largely on the reaction of the medium. From this it may be assumed that the antigenic properties of the two colonies very likely will be different in quality and quantity. The suspensions...