Fellman and Eriksson (2008) wondered whether there is variation in the probability that a human birth will be male. Accordingly, they examined the sexes of sibs of index MM, MF, and FF twin pairs. For this purpose they used 19th century German data published by Geissler and Lommatsch. In these data, the sex ratio (proportion male at birth) of sibs of MM pairs was significantly high; while that of MF pairs was normal and that of FF pairs was significantly low (as contrasted with contemporaneous live birth sex ratios). Accordingly Fellman and Eriksson concluded that there is, indeed, variation across couples in the probability of producing a son. Here it will be noted that though there are external grounds supporting this conclusion, there is nevertheless some reason to suspect a form of systematic error in the data cited by these authors. (In Geissler's data, there is very substantial unexplained variation of sex ratio of the sibs preceding index twins by the sex and birth order of the twins). Both these points will be addressed here. In addition, evidence is adduced that (1) the sex ratio of MZ twins is low, and (2) the sex ratios of DZ twins and of their sibs are high. Lastly, appeals are made for (a) data that would test the reliability of the data of Geissler and Lommatsch on the point questioned above, and (b) data on the sex ratios of offspring of twins by the sex and zygosity of the twin parents.