The Skull Wars

Chapter 3

The Skull Wars

Boas and the immigrant children

The same cephalic index that nineteenth-century craniometrists used to sort populations into fixed, ranked types (Chapter 2) became, in the hands of the German-American anthropologist Franz Boas, a tool for testing whether those types were fixed at all. Between 1908 and 1910, Boas and a team of assistants measured roughly 18,000 European immigrants to the United States and their American-born children — Sicilians, Neapolitans, Hebrews, Bohemians, and other groups then treated by immigration authorities as biologically distinct "races." In his 1912 report, Boas found that children born in the United States, especially those born several years after their parents arrived, showed cephalic-index values shifted measurably away from their parents' — and, notably, the head shapes of different immigrant groups seemed to be converging toward each other in the American-born generation rather than staying fixed apart.¹ Boas read this as evidence against the reigning assumption of physical anthropology: that head form was a stable, heritable racial marker unaffected by environment. If a single generation of American diet, health, and upbringing could shift the very measurement craniometry treated as destiny, then "type" was not fixed — it was plastic, responsive to the conditions a body grew up in. The finding was enormously influential, cited for the rest of the twentieth century as one of the empirical blows against biological race theory.

The study was not a standalone academic project. It was one volume of the 41-volume Dillingham Commission report Congress convened to settle the era's immigration debate with data — and the same commission also published a Dictionary of Races or Peoples that ranked ethnic groups by supposedly fixed hereditary traits. It was that document, far more than Boas's contrary findings, that fed the racial thinking restrictionists carried toward the national-origin quotas of the 1924 Immigration Act.²

Boas was not the first to have a racial-measurement project turn on its own premise. In the 1870s and 1880s the German pathologist Rudolf Virchow — a towering figure in physical anthropology — organized a survey of the hair, eye, and skin color of roughly six million German schoolchildren, in part to map supposed "Jewish" and "Germanic" types. The data refused to sort that way: Virchow found no clean racial patterning, and he went on to publicly ridicule the emerging cult of a pure "Nordic" or "Aryan" type as mysticism rather than science. The legacy is more tangled than a clean hero-of-science story, though: the historian Andrew Zimmerman argues that the survey's very practice of tabulating "Jewish" and "German" children as distinct, countable categories helped normalize racial classification in Germany, whatever Virchow's own anti-racist conclusion.³

The stakes of that plasticity finding turned lethal within Boas's own lifetime. On May 10, 1933, Nazi students threw his books onto the "un-German" bonfires burning across German university towns; Columbia's student newspaper ran the 74-year-old anthropologist's reaction under the headline "Nazis Burn Boas' Books Today; Anthropologist 'Not Interested,'" noting only that he disagreed with Hitler on the supremacy of the Nordic race.⁴

A 2002 statistical challenge

For nine decades, Boas's cranial-plasticity finding sat largely unexamined by the statistical methods anthropology later developed. That changed in 2002, when Corey Sparks and Richard Jantz returned to Boas's original data — by then digitized — and reanalyzed it using modern quantitative-genetics techniques, including pedigree-based heritability estimates Boas did not have available.⁵ Sparks and Jantz reported that, under their reanalysis, the shifts Boas had attributed to environmental plasticity were smaller and less consistent than Boas's original interpretation suggested, and that cranial form showed higher heritability — more resistance to environmental influence within a generation — than the 1912 paper implied. Published in the Proceedings of the National Academy of Sciences, the paper argued that a landmark finding, cited for ninety years as settled, had never been rigorously stress-tested against the statistical tools available by the early 2000s.

A 2003 defense

The following year, Clarence Gravlee, H. Russell Bernard, and William Leonard published their own reanalysis of the same immigrant dataset, reaching a different conclusion.⁶ Applying different statistical and computational methods to Boas's data — and arguing that Sparks and Jantz had characterized Boas's original claims more strongly than Boas himself had stated them — Gravlee, Bernard, and Leonard reported that their reanalysis supported Boas's core conclusion: cranial form in the immigrant children's generation did shift in ways consistent with meaningful environmental plasticity, not merely statistical noise or a heritability the earlier team had underestimated. Where Sparks and Jantz read the data as undercutting Boas, Gravlee, Bernard, and Leonard read essentially the same numbers as vindicating him, once the methods and the historical framing of his original claims were handled differently.

The debate continues

Sparks and Jantz did not concede the point. In a further 2003 response, they defended their original statistical approach and pressed their case that the immigrant data, read carefully, still showed less plasticity and more heritability than Boas's legacy claimed.⁷ The exchange did not resolve into a declared winner — it settled into an ongoing methodological disagreement about how to model heritability and environmental effect from century-old pedigree data, a disagreement that continues to surface in the physical-anthropology literature. That is not a flaw in the science; it is what science on a hard measurement question actually looks like while it is still being worked out, as opposed to a question that was settled and later shown to be fraudulent or pseudoscientific.

The pattern of reanalyzing an old immigrant dataset was not new even in 2002. Harry Shapiro had measured Japanese immigrants and their Hawaii-born children in the 1930s, and when J. W. Froehlich reworked that Hawaii data in 1970, he found real anthropometric shifts across generations in a wholly different population — but ones that leveled off after a single American-born generation in men, more consistent with a heritable ceiling than with unlimited environmental malleability.⁸ The same argument, on a different ocean, had already landed on a similarly split verdict.

Same measurements, opposite conclusions

The throughline is worth sitting with. The cephalic index that Retzius introduced and that nineteenth-century craniometrists used to rank populations into a fixed hierarchy (Chapter 2) is the same measurement Boas later used to argue against fixed racial types altogether — and it is the same measurement still being argued over today, on entirely different grounds, by anthropologists using tools Boas never had. Chapter 2's story was about a measurement toolkit built to serve a predetermined conclusion, and it was pseudoscience from the start. This chapter's story is different in kind: legitimate scientists, using the same raw dataset, applying different but defensible statistical models, and reaching different readings of what that dataset shows. The lesson is not "measurement is unreliable" — it is that a single number, however carefully collected, does not interpret itself, and how a discipline chooses to model heritability, sampling, and environmental effect can shift what the same measurements appear to prove.

Further reading