1. Selection for sex ratio in mice was ineffective. The heritability of sex ratio under the conditions of the experiment was not above about 5 per cent. 2. Statistical analysis of sex ratio in large numbers of mice of several stocks revealed no significant heterogeneity between families within stocks. 3. Selection for sex ratio in Drosophila melanogaster was effective in one strain selected for excess of females. In a second experiment selection was not effective in either direction. The presence of a sex-linked lethal in the first experiment is postulated. The heritability in the second experiment was not above 3 per cent. 4. There was significant heterogeneity of sex ratio between families in the first Drosophila experiment but not in the second. 5. The absence of detectable multifactorial genetic variance affecting sex ratio suggests that sex ratio may differ from better known metrical characters, such as body-size and bristle number, in the underlying genetic control. But this difference, it is suggested, need not be real. The large amount of error variance inherent in the measurement of sex ratio misleads one into expecting a commensurate amount of real variation, and we have no logical criterion for deciding how much real variation to expect.