Discussion.-Burnet and Fenner7 list a number of essential immunological ob-servations which are not satisfactorily accounted for by the antigen-template theory. 1. The booster effect. A secondary stimulus with the same antigen provokes a more active production of antibody than does a primary stimulus. According to the natural-selection theory, this would be due to the fact that antigen injected secondarily encounters a larger concentration of specific antibodies in the circulation than were present at the time for the primary injection. More antibody molecules are therefore brought to the globulin-reproducing cells. 2. The change in character of the antibody produced in response to repeated inoculations of the same antigen. The main changes observed are from" low-grade" antibody of low combining capacity, produced in the beginning of an immunization course, toward" more avid" antibody of high combining power, produced later. 7-'0 Besides this improvement in quality, an increase has been observed in the range of cross-reactions with related antigens. The present theory explains this develop-ment by natural selection. At the time of the primary stimulus the antigen injected finds only few globulin molecules in the circulation, showing various degrees of affinity toward the antigen surface patterns. At the time of a later stimulus in the course of immunization, when these molecules have been replicated in large num-bers, the antigen will find a larger concentration of globulin molecules fitting all its surface patterns and will preferentially carry those which show the highest combining capacity to the globulin-reproducing cells. 3. The apparent exponential rise in circulating antibody during the first period of production. On the present theory this may be due to an autocatalytic replication of the specific globulin molecules and to a multiplication of the cells. 4. The continued production of antibody for long periods. The antigen-template theory could deal satisfactorily with this point if it could be shown that