Summary: Immune sera produced by the injection of 6 times crystallized duck ovalbumin and hen ovalbumin into rabbits were at first rather specific, those produced by the former apparently absolutely so, as was a serum produced by injecting a hen with duck albumin. Rabbit serum from later bleedings reacted equally with both proteins, but from such sera the reactivity to the heterologous antigen could be removed by specific adsorption, leaving that to the homologous antigen practically unchanged. It was found by the method of optimal proportions, that a titration of antibody in such an adsorbed serum indicated that there remained after adsorption with the heterologous antigen as much antibody against the homologous antigen as was there originally. These facts are thought to support the idea that in one molecule of a pure substance there may exist several antigenic determinants of diverse specificity.