The personal equation often makes a statistical study of a clinical problem in medicine difficult. Much discrimination is needed in order to obtain two strictly comparable groups of patients in order that one of them may be used as a control. Proper precautions must be taken in advance in collecting data in order to prevent the results from being obscured by variables which may make the study useless. Having these facts in mind, the Poliomyelitis Committee of the New York Academy of Medicine attempted to estimate the value of convalescent human poliomyelitis serum1the efficacy of which had been established on the basis of Flexner and Lewis'2studies on experimental animals. Unfortunately, earlier work on human beings3had been done without proper control cases, and, with few exceptions, paralytic as well as preparalytic patients were treated. Later it became obvious that the disease in the paralytic stage