To the Editor.—
Cloninger et al1 claim convincing evidence that schizophrenia is a discrete clinical disorder. They cite three justifications for their belief: (a) bimodality was obtained in a discriminant function score of symptomatology; (b) on a simple sum of five symptoms, the probability of subsequent diagnosis as schizophrenic was nonlinear, with a sharp increase between one and two symptoms; and (c) only when subjects were dichotomized at the same place on this symptom score between one and two symptoms did a familial relationship in schizophrenia emerge.The alternative possibility is that schizophrenia is a dimensional illness, with only those at the extreme coming to medical attention. If schizophrenia is a dimensional disorder, rather than a discrete entity, each of the three findings noted above could emerge. The defects in the evidence lie in the choice of symptoms and selection effects in the sample studied.The simplest way to demonstrate