Do men named Smith and Tailor sort into different sports? Testing a family-name theory

During the Middle Ages, a lot of surnames came straight from what a person did for a living. A blacksmith's family became the Smiths; a man who cut cloth passed the name Tailor down to his sons. Working from that custom, and from the fact that these names moved from father to child across many generations, Bäumler put forward an idea in 1980: maybe the bodies went along with the names. His genetic-social theory proposed that physical traits like height, weight, and overall build might be distributed unevenly between populations of men carrying the German names Schmied (smith) and Schneider (tailor).

The reasoning runs like this. Forging iron at an anvil is heavy work that rewards a tall, strong frame, and over centuries men suited to it would have ended up in that trade and kept the name. Sitting at a bench with needle and thread asks nothing of the kind. If the trait really did ride along with the name through inheritance, you would expect the two surnames to point toward two different body types even today.

Stemmler and Bäumler turned that into a prediction about elite athletes. In the power-and-size events of track and field, the ones that call for strength and height, there ought to be a relatively larger share of men named Schmied than you would find in the endurance events. And in the lighter, stamina-driven disciplines, the name Schneider should turn up more often. So the Smiths cluster toward the throwing and heavy events, the Tailors toward the distance running.

They did not stop at those two statements. The study also checked the mirror-image claims: fewer Schneiders than chance would give you among the heavy events, and fewer Schmieds than chance among the light ones. Each of these was set against a null hypothesis that said name and discipline have nothing to do with each other.

To test all of this they leaned on methods for picking apart contingency tables, including configural frequency analysis, a chi-square decomposition, a biprediction setup, and the Kimball test. The analysis was run on two separate samples drawn independently of one another.

Both samples pointed the same way. Every one of the hypotheses, the main predictions and their complements alike, found support in the data.