R. Goldschmidt worked on intersexuality in the Lepidoptera for many years up to 1934 and the moth Lymantria dispar was his chief experimental material. According to him, an intersexual insect develops for a time as one sex and then changes to the other, though the chromosomal sex remains that of the original zygote. If the change takes place early enough in development, e.g. at the formation of the gonads, the whole insect appears to be sexually converted, whereas, if it occurs later, only those structures formed towards the end of development, e.g. the wings, will be affected. Goldschmidt held that, within races, a fixed dosage of female determinant, carried maternally in the cytoplasm or in the Y chromosome and elaborated into the cytoplasm, outweighs the effect of a single dose of male determinant carried in one X chromosome in the heterogametic sex, here the female. In the homogametic sex the double dose of X chromosomes is balanced against the single dose of female determinant received from the female parent, and results in a male. Moreover, a certain minimum excess in either direction is required for normal sex determination. He held that, while the relative values of the sex determinants always conform to this plan, their absolute values may differ from one geographical race to another. Consequently, intersexuality due to lack of correct balance between the sex determinants may arise in different ratios when distinct races are crossed. Its degree and type, whether male converted towards female or the reverse, are controlled by the races and sexes used. Since sex abnormalities have appeared, but only sporadically, in more recent genetic work involving race crosses in mimetic butterflies, we decided to reassess Goldschmidt's results. On repeating those of his crosses that he regarded as the most fundamental, between German and Japanese material, we bred a number of intersexes, but there were marked discrepancies between his and our overall findings. The matter is discussed and it is shown that the accuracy of the Goldschmidt hypothesis can now be tested in much more detail by using the heterochromatin body in the larva as a prospective marker of chromosomal sex in the adult.