ScienceDaily highlights an angle on the paper I blogged a few days ago, Chromosome And Surname Study Challenges Infidelity 'Myth':
"People often quote a figure of one in ten for the number of people born illegitimately," says Professor Jobling. "Our study shows that this is likely to be an exaggeration. The real figure is more likely to be less that one in twenty-five."
I didn't comment on this because there is research which contradicts the 1 in 10 number. There's a lot of variation worldwide, but among mainstream Western populations the misattributed paternity rates are on the order of 2%, not 10. Here are the rates from the paper:
This approach yields the following rates: Attenborough 1.29-3.39%; Haythornthwaite 2.07-4.54%; Herrick 1.00-2.47%; Stribling 1.00-2.87%; Swindlehurst 1.04-2.76%. However, it should be noted that if, in fact, these surnames had multiple founders, but only one founding lineage had survived to yield a sampled descent cluster...then the true nonpaternity rates would be lower than our estimates.
The logic here is rather straightforward. Remember that Y chromosomes are passed only from father to son. Take a theoretical paterlineage, and connect individuals vertically from father to son and iterate. Assume a 10% rate of misattributed paternity. This means that with no other information you are 90% certain that a man's putative father is their genetic father, and that his putative son is his genetic son. This implies that the putative relationship between a random grandfather and grandson is 81% likely to be in alignment with genetic realities. Now assume that a man has two sons, who each have two sons. If the probability of each grandson sharing the Y lineage with their paternal grandfather is 81%, then there is a 43% probability that all four grandsons are actually genetically related to their putative grandfather. Assumptions about independence are likely not valid; men who are cuckolded once are certainly more likely to be cuckolded later for whatever reason. But even with the simplifying assumptions you can see that a 10% misattributed paternity rate would result in a great deal of non-relatedness across a lineage. When looking at phylogenies such a high rate would stand out immediately. 10 generations is ~250 years. Again with the same assumptions the chances of a patrilineal descendant of a man being his genetic descendant is on the order of 40%. In other words, 250 years into the future the correspondence between theory and reality in terms of descent should begin to uncouple to the point where enormous numbers of men within the extended lineage are unrelated on the Y chromosome. The researchers report that this is not what you see, that in fact for many rare surnames there are high frequency lineages whose most recent common ancestor go back 500-1,000 years. This is not so for common surnames like Smith, but combined with contemporary cross-cultural non-paternity rates, it seems it is likely that the lack of relatedness has less to do with infidelity than the fact that a name like Smith was adopted by many unrelated individuals within that profession. Of course there are cross-cultural variations. That is why the French bill which required genetic tests to ascertain relatedness among immigrants drew such a sharp response from Africans. Some of this might be a different conception of family, but there's going to be rather obvious differences in paternity rates among Saudi women, whose men hold them chattel, and West African women who have long been independent economic actors. For example, DNA Tests Offer Immigrants Hope or Despair:
So when he became an American citizen and officials suggested taking a DNA test to prove his relationship to his four sons, he embraced the notion. Imagine, he marveled as a lab technician rubbed the inside of his cheek, a tiny swab of cotton would reunite his family. But modern-day science often unearths secrets long buried. When the DNA results landed on Isaac Owusu's dinner table here last year, they showed that only one of the four boys -- the oldest -- was his biological child.
Why has the 1 in 10 number become etched into public lore? Because it's a nice round number. Just like the "1 in 10 homosexual" number was long accepted.