There has been quite a stink over the past few weeks about what color skin the Romans in Britain had.  The BBC put a dark-skinned Roman official in a children’s cartoon history program, and the denizens of social media were off to the races. [1] Mary Beard  and Neville Morley picked up the standard for the classicists. Among the alt-right antagonists was the pop market-analyst N.N. Taleb, who got famous for coining the term “Black Swan”, but seems not to have the chops to back up his reputation. The noise from the racists got so loud that the Faculty of Classics at Cambridge felt obliged to weigh in. This is kind of amazing to me. I was sure everyone knew that the Roman Empire stretched well into Africa and Asia.  They even had an emperor called “Philip the Arab“, for crying out loud!  When the Roman generals chose troops to occupy a far-flung province, prudence dictated that their preferred troops not have any language in common with the subject population except Latin.  So claiming that the Romans in Britain were all light-skinned seems unsupportable.

But let’s see what they’re saying.  As nearly as the the alt-right are willing to be understood, they’re basing their objections on a map of genetic markers in the current British population.  A 2015 study of the fine-scale genetic structure of the UK doesn’t show much sign of African genes.  They think this “hard science” is much more important evidence than squishy, historically-based evidence, even when the historians have eyewitness accounts.  The Guardian article I linked at the top lists some good reasons why genetic surveys might not be the best evidence for claims about ethnicity 2,000 years ago.

I’d like to add another reason:  There resemblance between the genetic survey’s clusters and the patterns of family names in the UK is not strong. Leslie et al. tracked autosomal DNA, not mitochondrial, so there should be strong parallels.  Surnames and genes are inherited from the same ancestors, after all. A map by from the Nature article looks like this:

Leslie et al. map of clusters

Clusters of genetic similarity.

In the course of tracking down hobbits, I found the work of James Cheshire and his collaborators,  which shows a strong relationship  between clusters of family names in the UK and the cultural/administrative regions of the country.  Here’s a map published by Cheshire, Longley, and Singleton in 2010.

Clusters of family names

There are some general resemblances. The big homogeneous blob in the East and South East is there, though family names don’t let it extend all the way to Northumberland.  I can see hints in the light-blue smear in the North West and the purple smear in the southern West Midlands. Below the coarsest level, though, the two distributions do not resemble each other very well.  In particular, the genetics suggests that the people of Pembrokeshire (the southern peninsula of Wales) are affiliated with the Scotch-Irish borderlanders.  Family names suggest they’re more like the West Midlanders.  And if there’s a family resemblance between Yorkshiremen and Cornishmen, it doesn’t show up in their names.

The conclusion I draw from this is that the genetics is pointing us in a common direction with external markers of family ties. There really is something there, and a salute to the geneticists who have managed to tease it out. However, the signals are accompanied by a lot of noise.  We can’t yet use genetic evidence with any precision.  When we have a person standing next to an Ethiopian legionary on Hadrian’s Wall and writing about it, it would be foolish to try to contradict him with our rudimentary genetic surveys.

Post-scriptum: I really enjoyed the line, “History is written by the winners; genetics is written by the masses.”


[1] Sorry.