No detailed and comprehensive system of subclassification of chronic schizophrenia has been generally adopted in this country. Most psychiatrists are content to use the broad categories of simple, hebephrenic, catatonic, paranoid and undifferentiated schizophrenia. Placing a patient into one of these subgroups entails a judgment which is based upon all the information at the clinician's disposal. Such data are necessarily unstandardized; the method of collection varies from doctor to doctor and from patient to patient, and the criteria of classification are not uniformly agreed upon. No objective tests are available which would serve to check the accuracy of the decisions taken. It is inevitable, in these circumstances, that clinical systems of subclassification should not be much used for scientific purposes. The basic essential of a classifying instrument is that it should produce the same results in the hands of different investigators, and that its standards of measurement should not vary from one occasion to another. No study so far published has attempted to deal with this problem, which becomes more serious as the elaboration (and possibly the clinical usefulness) of the classification increases. Leonhard's system (1, 4, 5) for example, would gain enormously if its reliability could be demonstrated.