The probability of a twin birth depends strongly and almost equally on the age of the mother and on the number of children she has had. Even when these variables are held constant, the frequency of twinning varies between countries or geographic regions and between years or decades. This variation has been thought perhaps to result from economic and environmental factors, but another explanation can be suggested.

Bulmer (1959) found that twins are frequent among the earliest conceptions after marriage. Eriksson and Fellman (1967) found twins to be frequent among illegitimate births and suggested that the mothers of twins conceive more easily than other women; in other words, that they are more fecundable. We can present some new evidence for this hypothesis, and we believe that high fecundability in the mothers of twins can account for variations in the twinning rate.

Fig. 1 shows the wide range of twinning rates in the US white population, in different parts of the country and for three periods, centering on 1937, 1954 and 1964. The nine geographic divisions are shown in descending order according to the twinning rates in 1937. The two regions at the right, with the lowest rates in 1937, had nearly stable twinning rates over the three time periods. They are the oldest and most urban parts of the country.