The writer believes that the experiments above recorded have demonstrated the following points: 1. As has been previously shown by other workers, complement is fixed by the precipitate formed in the reaction between unformed proteid and its homologous antiserum, and it is the precipitate only which fixes, the supernatant fluid being devoid of fixing power except to a slight degree in tubes very rich in antigen. This latter fixation is probably due to a redissolving of the precipitin-precipitinogen complex which, as Gay has shown, does not lose its fixing power on resolution. The fixation of complement by mixtures in which antigen is so slight in amount that no precipitate is formed may be explained by analogy with colloidal reactions in which combination of colloids takes place without precipitation unless definite quantitative relations are maintained. 2. In contrast to the results above described, precipitating mixtures of bacterial antigen and antiserum show complement fixation both by the precipitate and by the supernatant fluid. 3. The complement fixation by the supernatant fluid of such bacterial precipitin reactions, with the precipitate removed, is, in the presence of washed bacteria, equal to that exerted by the same amount of the original serum. The complement fixation of the precipitates, here, as in serum-antiserum tests, is proportionate to the bulk of the precipitate. 4. The precipitin body is not removed from a bacterial antiserum by treatment with washed whole bacteria, whereas the cell sensitizer or amboceptor is removed by such treatment. The writer believes that the work recorded above justifies Gengou's conclusion that an albuminolytic sensitizer is formed in response to immunization with unformed proteids. This sensitizer, or amboceptor, is, however, distinct and independent of the sensitizer that reacts specifically with either bacterial or other cells. It would appear that immunization with unformed proteids calls forth only the formation of these albuminolysins, while immunization with cells, in our case the typhoid bacillus, calls forth both the cytolytic and the albuminolytic sensitizers. The albuminolytic sensitizers are not carried down mechanically with the precipitates in the precipitin reactions. We see no reason for not considering them identical with the precipitins themselves, bodies which combine with the dissolved antigens and, as in other colloidal reactions, lead to precipitation when definite quantitative relations are observed. The undiminished precipitating value of a bacterial antiserum after considerable amounts of bacterial amboceptor have been absorbed out of it, would point against the identification of agglutinins with precipitins. However, this question must be subjected to further study. It is thus clear that only the cytolytic sensitizer can enter into combination with the washed bacterial cell, while it is probable that both cytolytic and albuminolytic sensitizer may combine with the dissolved ingredients of a bacterial filtrate. Whether or not they combine with the same constituent of such an antigen cannot be decided by our experiments.