The interpretation of twin data.

ONE OF THE PROBLEMS that led to the development of biometry by Galton and Pearson was the interpretation of data on physical measurements of man. In spite of this early interest in the study of quantitative attributes in man, which constitute the common and the majority of important human characteristics, human genetics has been almost totally preoccupied with rare pathologies and so-called single gene effects, (Dobzhansky, 1955; Dobzhansky and Wallace, 1954; Mather, 1953; 1'enrose, 1949; Snyder, 1954, 1955) effects which can be conveniently studied by traditional pedigree analysis and genealogical methods. The pedigree methods are based on the replacement of continuous attributes by discrete attributes, e.g. PTC tasting (Harris and Kalmus, 1949), and by the assumption that the approximate discrete variable is controlled by the phases at usually a single locus or, rarely, a very few loci. This methodology has been very successful in achieving some understanding of so-called abnormalities and in general cases in which the continuous distribution has distinct modes. The type of attributes we are considering here do not fall into this category. With an attribute like height, for instance, the distribution in the population is not at all bimodal and exhibits in some respects an appearance of blending inheritance even though we are convinced that no such process happens. Similarly the treatment of environment in pedigree analysis is considered to be discrete, as, for instance, a malarial versus non-malarial environment. Undoubtedly some aspects of environment can be so classified but discontinuous or multimodal distribution is the exception rather than the rule. In man it is the environment which must be changed to permit different genotypes with limited or particular norms of reaction to produce the desired phenotypes, (Osborne and De George, 1959a). This requires an understanding of the interaction of heredity and environment for which we must study quantitative variables employing methods other than pedigree analysis. In theory at least, an efficient method for appraising the heredity-environment problem in man, particularly with respect to complex or quantitative inheritance, is by the study of twins (Verschtier, 1939). The twin study method began in 1875 with Galton's recognition of the useful-