James Watson's 16 percent ancestry

New York Times blog item on James Watson:

The math is puzzling me. How can a person have 16 percent anything ancestry? And wouldn’t having a great-grandparent who was African give Watson a 12.5 percent African ancestry?

Eh, 1/16 (6.25%) and 16% are pretty interchangeable for most common folk.

Dunno, but most amusing story since Strom Thurmond’s illegitimate daughter.

It could be a great grandparent (12.5%) and a great-great-great grandparent (3.125%), or any other imaginable combination that gets to approximately 16%. Also, my intuition is that these tests aren’t very precise. They’ll give a nice set of relative numbers, but possibly plus or minus a few percent.

It is possible he would have ancestory coming from different lines, not necessarily from a single source. For instance, suppose his he has one great-grandparent, one great-great-great grandparent, that would come out to 15.6%. If you throw in a great-great-great-great-great-great grandparent, it will get to almost exactly 16% (16.015625%, actually). Also, as one might notice, having two great-great grandparents in different lines is the same as having one great grandparent in one line. IOW, the number of combinations that could approach 16% is limited by how far back you go.

Of course, whether or not 16% is what he actually meant, is another question. I’d tend to agree with Earl Snake-Hips Tucker. The article quotes “someone who had a great-grandparent who was African” but the 16% is not quoted, so it’s entirely possible (and IMO, likely) that the author is just an idiot and doesn’t know the difference between 1/16th and 16%.

And, for a different perspective, here’s an article that’s skeptical of the whole process. At the base of this all, geneticists have identified specific sets of genetic markers that have some correlation to modern populations living in a specific area. And, further confounding any specific predictions, the reassortment of these markers that happens in human lineages is also a random process. So test like this may be able to place you in a specific group, but statements of ancestry get weaker the further back you go.

But is James Watson Sicilian?

That article doesn’t really address the process in question, though. While mtDNA analysis (which your article criticizes) is often trumped up to be more than it is, that’s not what we have for Watson-- we have his entire genome. You can do a lot more with that than you can do with mtDNA. Claims that you “come from” such and such an area because of your mtDNA are flawed because that only means one of your remote ancestors (among hundreds) probably came from that area.

I am still suspicious of the claim being made about Watson, only because there are no details given. And these markers are better used to characterize whole populations than individuals, since we’re talking about the relative probability of having such markers not thatall individuals from X population always has such markers.

As a genealogist, I am also suspicious of the findings. For Watson’s ancestry to be 16% African, one of his grandparents was 64% African, or two of his grandparents were 32% African, or, very unlikely, all four of his grandparents were 16% African. Or some combination of the above that totals 64%.

Yet all four of Watson’s grandparents are listed as “white” in the U.S. Censuses, 1910 to 1930.

1. Thomas G. and Nellie Watson on his father’s side. Thomas was born in Illinois, his parents in Indiana and Minnesota. Nellie was born in Wisconsin, both her parents in New York.

2. Lauchline A. and Elizabeth Mitchell on his mother’s side. Lauchline was born in Scotland as were his parents. Elizabeth’s parents were from Ireland.

I’m having trouble believing that anyone who was between 32% and 64% black African could pass as “white” in those days.

I’ll look for Thomas and Nellie’s parents in the U.S. Census. If they are also listed as white, I’m calling bogus on the 16% African ancestry.

Remember that genetics aren’t all clear-cut in nice little powers-of-two slices. Yes, a person will, on average, be 25% related to any given grandparent, but depending on the way the cards get shuffled, it could be anywhere from 0% to 50%. In fact, a person could be anywhere between 0% and 50% related to any given ancestor further back than parents. The percentages close to the average are more likely, but they’re not guaranteed.

If you’re going to take that link as meaningful, I’d rely more on the part where it says it would be like one of his great-grandparents was African. In that case, his grandparents would have to be:

3 White, 1 50% White/50% Black
2 White, 2 75% White/25% Black

Or something along those lines. I think the second scenario is not out of the question, but I still don’t buy the analysis without more info on how the person came to that conclusion.