Umbilical cord blood was collected from foetuses at 12–22 weeks gestation delivered by hysterotomy, and from term foetuses delivered by elective Caesarean section or normal vaginal delivery. Plasma testosterone was measured by the radioimmunoassay method of Hotchkiss, Atkinson & Knobil (1971) as modified by P. Rowe (in preparation) using an antiserum prepared against testosterone-3-oxime bovine serum albumin. Chromatography was omitted as it was found to be unnecessary. This antiserum cross-reacted with 17β-hydroxy-5α-androstan-3-one (27·5%) but cross-reaction with C18, C19 and C21 steroids known to be present in human umbilical plasma was minimal (e.g. oestriol, 0·05%; oestradiol-17β, 0·02%; androstenedione, 0·01%; dehydroepiandrosterone, 0·05%; progesterone, 0·02%).
The concentration of testosterone in male and female foetuses is listed in Table 1. The blood from mid-pregnancy foetuses was collected by dividing the cord and allowing both ends to drain, care being taken to avoid contamination from liquor amnii or maternal blood. Blood from