1. It is well established that p, the probability of a male birth, declines with birth order. It is suggested here that this decline occurs within individual sibships (Poisson variation). 2. Other workers have also offered good evidence that (at least in some samples) p correlates positively between adjacent births within sibships (Markov association). 3. The present paper suggests that in addition, some couples have higher values of p than others. In other words, there is Lexis variation in p between couples. It is estimated that this Lexis variation augments the variance (contingent on the Markov process above) by a value of less than 0-002. 4. Data on sex ratio by the sex composition of the pre-existing children are interpreted to support the author's hypothesis that coital rate of parents is related to the sex ratio of their children. 5. However if the decline in sex ratio with maternal age were entirely due to the concomitant decline in coital rate, the variation in p between couples seems greater than could be accounted for by the observed variation between couples in coital rate. It is concluded that sex ratio depends on other factors besides coital rate. 6. It is noted that (when sibship size and birth rank are controlled) MF birth intervals are particularly long, and FM birth intervals particularly short, MM and FF intervals being of intermediate length.