That portion of horse and human serum globulin precipitated by 33 per cent saturation with ammonium sulfate and precipitated on subsequent dialysis was taken as euglobulin; and the fraction precipitated between 33 and 50 per cent saturation and remaining in solution on subsequent dialysis was taken as pseudoglobulin. The sera of rabbits injected with either of these antigens gave precipitation with both. However, two distinct and fraction-specific antibodies could be demonstrated by absorbing the sera with the one antigen, and testing the supernatant fluid with the other. The experimental results are adequately explained on the basis that there are at least two antigenically distinct globulins in serum which we may term globulin I and globulin II and which are largely associated with so called euglobulin and pseudoglobulin respectively. The ordinary methods of salting out and dialysis do not effect complete separation and each globulin preparation contains a trace of the other antigen. The antisera to these euglobulin and pseudoglobulin preparations therefore contain antibodies to both antigens. Each protein solution precipitates all the antibody specific for the one antigen and in addition, by virtue of the trace of contaminating protein, precipitates a portion, and only a portion of the antibody specific for the other antigen. The fact that antisera to whole serum contain these same fraction-specific antibodies suggests that this immunological specificity is an inherent property of two globulins present as such in serum and is not an artifact induced by their precipitation and purification. Lipoids extracted from the globulins by ether, petroleum ether, and alcohol give no demonstrable reaction with antisera to these globulins; antisera absorbed with a large excess of lipoid are not affected as regards their reactivity with the original protein; and globulins extracted with ether and petroleum ether at room temperature are not affected as regards their reactivity with antisera. It is concluded that the immunological specificity of the globulin fractions as evidenced by the precipitation reaction is not determined by lipoids associated with the protein.