The injection into one rabbit (with Freund's adjuvants) of a specific precipitate made with antibodies from the serum of another rabbit is usually followed by the appearance in the serum of the first rabbit of antibodies which precipitate the serum of certain rabbits but not of others. It was found that the antigen (or one of the antigens) concerned in the reaction of these anti rabbit serum antibodies with rabbit sera had an antibody function, and was therefore a protein. It was concluded that at least one serum protein antigen, the specificity of which so far has been considered to be uniform throughout the animal species, can instead be present in different individuals as different forms or allotypes with somewhat different antigenic specificities. A large number of rabbit sera were allowed to react with a large number of rabbit immune sera. The gel method of immunochemical analysis made it possible to enumerate the allotypes that took part in each reaction. In addition, the technique mainly used (simple diffusion in separate tubes) made it possible to recognize the presence of one given allotype by the mere aspect of the precipitation zone in the reaction of one suitable immune serum with any serum in which the concerned allotype occurred. Neighboring reactions of sera, in contact with each other and with the suitable immune serum, in suitable cells easily constructed in the laboratory, were carried out occasionally and, each time, their results agreed with the previous identification. The analysis of the reactions in tubes lead to a list of seven allotypes designated by a, b, c, d, e, f, and g, of which two or more (e not included) were contained in almost every serum. The specific conditions necessary for antibody formation against an allotype are its absence from the serum of the immunized animal and, except in the case of cross-reactivity, its presence in the immunizing material. When these necessary conditions are fulfilled for several allotypes at the same time, their competition in the immunization seems to favor the allotype present at the highest concentration. The individuality of six of the listed allotypes has been discussed independently of the part of their specificity that may be common to all the allotypes of one given protein antigen in all the individuals of the same animal species. A cross-reaction of the anti f rabbit antibodies with allotype g has been observed. When two allotypic specificities were detected in one serum, attempts were made to find whether they were carried by two allotypes, i.e. by two distinct kinds of molecules, instead of being the manifestation of two "allotypic patterns" present on the same molecules. The presence of several allotypes in the immune sera made it often impossible to find definitive answers in this regard. However, for a limited number of cases of two allotypic specificities present in one serum, it could be demonstrated that at least a large proportion (if not the totality) of the two allotypes were independent of each other. No sign of a systematic coexistence of two allotypic patterns on the same molecules has been observed to date.