‘Lumping and splitting’ is a process that goes on all the time in science, as categories are grouped and regrouped, divided and dispersed. In medicine, diagnoses are changing constantly as new evidence accrues and old knowledge proves inadequate. At the same time the diagnoses are being included with, or separated from, other diagnoses according to their apparent similarities and contrasts. All of this is an attempt to provide a taxonomy of illness, a coherent schema so that illnesses can be classified with some degree of coherence. It is an ever-evolving dynamic process, since new information is continually revealing fresh relationships or exploring false similarities.

In 1900, paranoia was a well-accepted diagnosis. By 1950 it was vanishing from the psychiatric textbooks. In the year 2000 it will again be a widely recognized illness. How can this be? Much of the answer is to do with inappropriate lumping and especially to do with the over-enthusiastic and slipshod diagnosis of schizophrenia in the mid-twentieth century, particularly in the United States.