IN AN EARLIER publication in the ARCHIVES OF GENERAL PSYCHIATRY1 we reported the occurrence in human serum of an antibody which hemolyses rabbit erythrocytes. We reported that when a large number of persons, including patients at this hospital, patients in general medical-surgical hospitals, and healthy controls, were studied, a bimodal distribution of hemolysin level was found. Over 70% of schizophrenics but less than 25% of controls gave sera with titres above the low point of the distribution. During two years of work on this system, these findings were extended, with a final N of 616 and a χ2 of 111 differentiating schizophrenics from all other groups.
Two other laboratories were unable to confirm our observations. At a third laboratory, differences in hemolysin titre were found to be related to a difference in time of mixing serum and erythrocytes and in the subsequent