An estimate from pedigree data of the gene frequency for colour blindness.

There are several possible approaches to the measurement of the population gene-frequency for colour blindness, and each approach has its particular advantages and disadvantages. The first is the direct approach of a population survey and especially, since the disease is commoner in males, a measurement of its incidence in men. In the event of full penetrance and ascertainment of the XY genotype, the mathematical treatment is commendably simple, for the incidence in men is numerically equal to the gene frequency. The weakness of the method, however, is its dependence upon the presumption of full penetrance and this difficulty is manifest in the varying estimates yielded by different methods at different times. Bell (1933) quotes a survey in 1892 which yielded an estimate of 218 per cent. male incidence; Gates (1946) quotes another early survey of 14,846 persons where the male incidence was 416 per cent., and other more recent surveys give estimates in European populations of 6 63 per cent. (Grieve, 1946), 7 49 per cent. (Vernon and Straker, 1943), 7 95 per cent. (von Planta, 1928) 8 per cent. (Wright), 8*07 per cent. (Waaler, 1927) 9 7 per cent. (Wieland, 1933), and 10-07 per cent. (Schiotz, 1922). Detection of colour blindness by Ishihara charts is no doubt responsible for recent estimates being higher than earlier ones. But even allowing for this and for the fact that there seem to be geographical variations in the incidence of colour blindness (see Vernon and Straker, 1943), the variations among the different estimates of incidence are sufficient to cause some doubt about the validity of direct conversion of the male incidence to a genefrequency estimate. Nor indeed would a higher degree of consistency necessarily validate such a conversion; used alone the method does not provide any objective means of assessing its own propensity to overor under-estimate. The second approach depends upon the sex ratio of the disease. While the male incidence of a sexlinked disease is proportional to the gene frequency (g), the female incidence is proportional to its square (gs) (Knox, 1958; Dahlberg, 1947), and provided