In contrast to allotypy, the anti-idiotypic sera precipitate the corresponding immunizing sera, but not the sera taken in the immunizing rabbits prior to their immunization against S. typhi, nor the immunizing sera absorbed with the somatic antigen of S. typhi, demonstrating that idiotypes are antibodies.

The idiotypic specificities of the antibodies of one rabbit against S. typhi are not detected in the antibodies of the same rabbit against another noncross-reacting Salmonella (S. tranoroa) and vice versa; nor are they detected in the anti-pneumococcal antibodies of the same rabbit.

Each anti-idiotypic serum fails to precipitate anti-S. typhi sera of rabbits other than the immunizing one except for certain extremely faint reactions, the significance of which has not been established. The idiotypic specificities of anti-S. typhi antibodies of three rabbits were not found in anti-S. typhi antibodies of their parents. This lack of a sign of hereditary transmission of idiotypic specificities contrasts with allotypy. The apparent role of random chance in the determinism of the idiotypic patterns or of the idiotypic determinants has been discussed.

Unless it were admitted that antibodies with similar functions do not exist in different individuals, idiotypy apparently adds an order of magnitude to the antibody variability which had been previously envisaged. In one given individual, the heterogeneity of the idiotypic specificities seems to be less extended than that of the antibody functions. The possible relationships between these two levels of molecular variability and between the corresponding levels of cellular variability have been discussed.