The results obtained may be summarised in the following table comparing the filial and fraternal coefficients for autosomal and sex-linked variation. When dominance is complete the values given are referred to equal frequency of the allelomorphic genes, i.e.p = ½ = q.

Briefly, then, the contribution of sex-linked genes to the observed variance will tend to lower the correlation between father and son and between brother and sister. It will raise the correlation between father and daughter or between sister and sister conspicuously. These considerations emphasise the interest attaching to a separate investigation of fraternal correlation for sibs of all three sex classes in connection with studies on twins. If the correlation for dizygotic twins of like sex is higher than for dizygotic twins of unlike sex, the difference might be due to the greater similarity of environment for two individuals of the same sex, or it might be due to the contribution of sex-linked genes to the observed variance. In the latter case the correlation for boy pairs should be significantly lower than the correlation for girl pairs. The wide discrepancies in recorded data for fraternal correlations with respect to intelligence quotients might be re-examined from this standpoint. If an appreciable proportion of the total variance is due to sex-linked genes, it is of more importance that a boy should have a clever mother than a clever father.