« DYSLEXIA ETCETERA | Gene Expression Front Page | Science & brown writers »
September 12, 2003


In an earlier post here I discussed the curiously neglected implications of cuckoldry for estimates of heritability.

There is uncertainty over the incidence of cuckoldry in the past. I have just come across some data which suggest an incidence of about 5% in America in the first half of the 20th century. This is high enough to be worth taking into account.

For details proceed...

The data are reported in Anne Anastasi, Differential Psychology, 3rd edition, 1958, based on a study by C. Cotterman and L. Snyder in J. Amer. Stat. Assoc., 1939, 34, 511-23. Cotterman and Snyder studied the ability to taste phenyl-thiocarbamide among 800 families. Most people find this chemical tastes intensely bitter, but some cannot taste it at all. The ability to taste is inherited as a simple Mendelian dominant. Non-tasters are homozygous for the recessive non-taster allele. Two non-taster parents therefore should not have any taster children. Yet out of 223 children born to such couples, 5 (2.24% of the total) were tasters.

Prima facie, this sets a lower limit of 2.24% to the incidence of cuckoldry (barring rare events such as new mutations). The full incidence would be higher, because even with cuckoldry not all of the offspring of non-taster women would be tasters. Taster offspring would always be produced when these women mated with a homozygous taster male, and half the time (on average) when they mated with a heterozygous taster, but never when they mated with a non-taster. Assuming random mating in the population w.r.t the taster gene, the proportions of the genotypes are p-sq., 2pq, and q-sq., where p is the proportion of the taster gene in the gene pool, and q = (1 - p). Also assuming random mating w.r.t. the taster gene in adulterous relationships, the proportion of tasters among the offspring of non-taster females produced by cuckoldry should be p-sq. + pq. As q = 1 - p, this reduces to p. The value of p in the study population can estimated from the fact that 86 of the 800 families (10.75%) contained two non-taster parents. Since the proportion of non-tasters in the population is q-sq., the proportion of couples in which both partners are non-tasters should be q-to-the-4. This gives a value of about .573 for q and .427 for p. The full incidence of cuckoldry implied by these data is therefore 2.24%/.427 =5.25%.

I need hardly say that no firm conclusions can be based on this single set of data. The total number of individuals involved in the study is quite large (several thousand) but the crucial group of taster children born to non-taster parents is very small. Quite apart from sampling error, even a small proportion of misclassification, of either parents or children, would invalidate the conclusion. But it is consistent with other moderate estimates of the incidence of cuckoldry in modern western populations.


Posted by David B at 09:40 AM

A recent Mexican study found big differences between classes. Upper class people have very low cuckoldry rates, lower class people have high rates. Ties into lack of paternal investment in the lower classes.

There's also adoption. Plus a certain number of people grew up thinking their grandparents were their parents and their 15 year older mother was their big sister.

But, the legitimacy rate might well be higher. The legitimacy rate that Bryan Sykes found for Englishmen named Sykes doing Y chromosome tests, assuming one progenitor around 1300, was about 99% per generation.

Posted by: Steve Sailer at September 12, 2003 08:10 PM

Yes, I mentioned the Sykes study in my earlier post. Of course, an average of 1% over 20 or so generations leaves room for some variation in rates over time. I would guess that cuckoldry was more difficult to get away with in a small medieval village, where everyone knew each other's business, than in a modern city.

Posted by: David B at September 12, 2003 11:51 PM