Umbilical arteries were not found in some fetuses from female rats given retinoic (vitamin A) acid by gavage on day 10 of gestation. It was felt that fetomaternal exchange in them must have been maintained either by the vitelline placenta alone, or by the allantoic placenta supplied by branches of the vitelline artery. In control rat embryos the vitelline and umbilical arteries communicated until day 11 after which the two circulations became separate. In young from females given retinoic acid during gestation, however, this communication (vitelloumbilical anastomosis) persisted and the definitive umbilical arteries failed to develop completely. This resulted in the normally formed single umbilical artery of the cord and its branches to the allantoic placenta receiving blood from the aorta through the vitelline or superior mesenteric artery. Thus, although the intra‐abdominal portions of the umbilical arteries were missing, the allantoic placental circulation was maintained. The vitelline placental circulation was normal. In other experimental embryos persistent primary umbilical arteries were frequently seen alone or accompanied by definitive umbilical arteries. In normal human embryos a vitelloumbilical anastomosis apparently does not form and the single umbilical artery seen in fetuses with symmelia is most likely a persistent primary umbilical artery, as suggested by Dawson. However, the possibility of a vitelloumbilical anastomosis occasionally arising as an abnormality in human embryos cannot be overlooked.