At the present time the case for a genetic basis for schizophrenia rests on the compatibility of the pattern of elevated risks to relatives with certain genetic models, together with the continued exclusion of any particular environmental factor as a sufficient cause. Ordinary Mendelian models will not do for explaining the mode of transmission for mental disorders since we find almost no families with 25 per cent or 50 per cent segregation ratios, and the population rates are so high as to make each condition 'common' by the standards for Mendelian diseases. Dominant gene models with incomplete penetrance, genetic heterogeneity models, and polygenic models are all viable contenders in the explanatory arena.

A satisfactory corpus delicti for the biochemical geneticist or cytogeneticist is not yet at hand. This does not mean that specific genetic factors play no part, or even that we must compare progress in the genetics of mental disorders unfavourably with that in other fields. In the first Annual Review of Genetics Robertson (1967) pointed out that, after years of control over such characters as milk yield in cows and egg size in poultry, we are surprisingly ignorant of the real biochemical or physiological differences between inbred lines. One of the most exciting challenges of the near future will consist of the best ways to synthesize the empirical facts about twin, family, and adoptee studies so as to lead to testable hypotheses.

Methods developed by Falconer, Edwards, and Smith permit the estimation of the heritability of the underlying predisposition or liability to developing a mental disorder, once the assumptions about the model being appropriate are accepted. Independent estimates from various classes of relatives lead to the convergence on a heritability value of about 80-90 per cent for the liability to schizophrenia. A weighted polygenic model (cf. Thoday on bristle number in Drosophila) offers hope that some facets of the schizophrenic phenotype will be shown to segregate or to have detectable biochemical or neurophysiological consequences.

It is important to understand the implications of finding that a trait such as the liability to schizophrenia has a high heritability. In the samples so far studied, it means that environmental factors were unimportant as causative agents of the schizophrenias. However, and this cannot be emphasized too strongly, these data do not permit the conclusion that curative or preventive measures will be ineffective. As Falconer (1965, p. 69) has pointed out, 'The environmental factors proved to be unimportant are those operating in the population sampled and these do not include special treatments or preventive measures. No prediction can be made from a knowledge of the degree of genetic determination about the efficacy of curative or preventive treatments. All that could be said in such a case is that one will have to look outside the range of normal environments experienced by the untreated population.' (Italics added.)

The beauty of a diathesis-stressor theory, or philosophy if you will, is that it fills the chasm between geneticism and environmentalism. Our preferred model for construing the syndrome of schizophrenia permits the clearer separation of aetiological and phenomenological considerations. It comprises a network of events connected by sequential causal arrows. A chain of consequences is set into motion by a variable, polygenically caused predisposition and culminates in a set of symptoms recognizable as schizophrenia. Feedback loops and chance have important roles in the total picture. Our construction clarifies how psychotherapy or phenothiazines or a good mother may each contribute to symptom amelioration without necessarily casting light on aetiological questions. It is our hope that a heuristic genetic theory about the aetiology of schizophrenia will hasten the day which brings it under man's control.