Who Shall and Shall Not Have a Place in the World?

In the seventh essay in the Legacies of Eugenics series, Lily Hu asks whether the racialist and eugenicist roots of statistics can be cordoned off from “proper” science.

By Lily HuFebruary 13, 2025

• Science & Technology

• Politics

• History

This is the seventh installment in the Legacies of Eugenics series, which features essays by leading thinkers devoted to exploring the history of eugenics and the ways it shapes our present. The series is organized by Osagie K. Obasogie in collaboration with the Los Angeles Review of Books, and supported by the Center for Genetics and Society, the Othering & Belonging Institute, and Berkeley Public Health.

¤

IN MAY 1921, the organizers of the Second International Congress of Eugenics assured the public that its conference, to be held later that year in September, would be taking a measured approach to “the topic of human racial differences.” All sides of the matter would be soberly considered; no conclusions would be reached in a hurry. The public could feel secure that, as the statement brightly declared, “certain prejudices directed toward existing races will be removed when allowance is made for the influence of their social and educational environment, and their fundamentally sound and strong racial characteristics are brought to light. On the other hand,” the announcement more somberly continued, “limits to development of certain races and the inalterability through education and environment of the fundamental characteristics of certain stocks will be considered.”

By the time the first meeting of the congress rolled around, a scant five months later, the science seemed to have been settled—and it apparently gave reason for considerable alarm. In his welcome address, Henry Fairfield Osborn, president of the American Museum of Natural History and soon-to-be co-founder of the American Eugenics Society, described unalterable “hereditary handicaps” that were, alas, embedded in the makeup of certain races:

In the United States we are slowly waking to the consciousness that education and environment do not fundamentally alter racial values. We are engaged in a serious struggle to maintain our historic republican institutions through barring the entrance of those who are unfit to share the duties and responsibilities of our well-founded government. The true spirit of American democracy that all men are born with equal rights and duties has been confused with the political sophistry that all men are born with equal character and ability to govern themselves and others, and with the educational sophistry that education and environment will offset the handicap of heredity. South America is examining into the relative value of the pure Spanish and Portuguese and of various degrees of racial mixture of Indian and Negroid blood in relation to the preservation of their republican institutions.

Later scientific sessions at the Second International Congress would validate Osborn’s concerns about the many dangers of racial “intermixture.” The essay “The Problem of Negro-White Intermixture and Intermarriage” by Frederick L. Hoffman, former president of the American Statistical Association, found that “the mixed race is physically the inferior of the white and the pure black” and that “morally, the mulatto can not be said to be the superior of the pure black.” At the First International Congress on Eugenics, held a decade earlier in 1912, Hoffman had warned of declining birth rates among “native stock” and the much higher birth rate among foreign-born mothers.

It’s hard to read these remarks today without wincing at the overtly racist language being passed off as scientific taxonomy. But of course, in the late 19th to early 20th centuries, all of this was scientific taxonomy. None of it was paradoxical, nor were the labels “race science” and “scientific racism” contradictions in terms.

Recent histories have chipped away at the notion that “race science” or “scientific racism” was a niche endeavor, or the province of the rare oddball racist fanatic. Rather, statistics as a field was inherently and unselfconsciously a racist discipline. As Khalil Gibran Muhammad documents in The Condemnation of Blackness: Race, Crime, and the Making of Modern Urban America (2010), as Tukufu Zuberi traces in Thicker Than Blood: How Racial Statistics Lie (2001), and as Aubrey Clayton explores in this series and elsewhere, the racism inherent in early statistical science was not aberrational. It was, to use the language of philosopher and historian of science Thomas Kuhn, “normal science,” just business as usual.

That said, in considering the legacies of such thought today, shouldn’t it be relevant—and exonerative even—that statistical analyses about race are now, in the 21st century, predominantly deployed for egalitarian ends? Racial disparities are, after all, increasingly treated by the social sciences as social disparities, rather than as markers of innate physical difference. As such, they are understood to present a serious indictment of our society and not of a subset of human beings. From this vantage, statisticians eager to defend their discipline might cry foul at attempts to draw conclusions about statistical methods from abhorrent aspects of the discipline’s history. They might, for instance, claim that the accusation partakes of the genetic fallacy—insofar as it tries to conclude, from the origin of statistics, its reliability as a field of study—and they might, too, say that it neglects the progressive nature of quantitative social science on the topic of race today.

This is the usual impasse that historical critique reaches every time unfortunate, sordid details surface about some seemingly innocent institution or field of study. Despite the findings of so many committees and task forces assembled by corporations, nonprofit groups, and universities in recent years to investigate their institutions’ ties to slavery—and despite claims of the supposed continued relevance of these findings—they have nevertheless seemed safely ensconced in history. This explains why, on the one hand, unearthed histories of racism, eugenic thinking, and the like are acknowledged by social scientists and statisticians today to be an embarrassment but, on the other hand, still fail to transform the field’s methods and analyses. Only properly methodological critiques could, so it seems, threaten the discipline.

This standard response hinges on a distinction between what is supposedly “external” to scientific inquiry and what is properly “internal” to it. The historicist challenge, by contrast, collapses that distinction: it insists that the social enterprise of statistics—including its founders, its institutions, and their entanglements—bears on the core of statistics itself: its methods, analyses, and results. If the historicist critics are right, then failures of both morality and rationality in our scientific past might not have been vanquished once and for all. They would have afterlives within the discipline that we now call “quantitative social science.”

What precisely would this connection be? What grounds this continuity? Aubrey Clayton suggests an answer in his essay in this series when he tells us that, for the eugenicist dealing in correlations, “proof was never an option.” There is always distance to be traversed from observation to conclusion. “The process of data interpretation is never-ending,” he writes, “with no one having the final word.” This is by now a truism about all scientific inquiry, and it is, as I will show in the rest of this essay, what yokes together, no doubt uncomfortably, the more egalitarian racial statistics of our time with the racist statistics of Hoffman’s.

¤

Rather than start by asking how 19th- and 20th-century statistics about race became racist, we might first ask how such statistics became science. Part of the challenge is to see race science not as an aberrational departure but as bona fide science—not in terms of being true or insightful about the nature of reality but in the more prosaic sense of being a real product of scientific inquiry. Racist statistics traveled through official channels of scientific knowledge production. It was the work of men (indeed, almost exclusively men) of specialized technical training. It was debated among experts, subject to critical peer review, and published by professional organizations in public scholarly venues. Its evidence and theories were variously challenged, discarded, resurrected, revised, and refined. Passing through this gauntlet of modern scientific scrutiny, its basic premises remained largely intact decade after decade. Taking this robust persistence seriously means grappling with the success of race science as science. Its means contending with its real epistemic merits despite its irrational basis.

To this end, Muhammad’s Condemnation of Blackness focuses on a specific figure from the early years of American statistics: M. V. Ball. A prison doctor, Ball publicly sparred with Hoffman over statistical analyses of racial disparities. Muhammad unearths archival evidence of how Hoffman, a representative of the statistics establishment, prevailed over Ball (who serves as a surrogate for the more progressive quantitative social science of today). Their exchange speaks to the historical and philosophical question of how Hoffman’s racist statistics successfully became “science” while Ball’s nonracist statistics failed to do so. How is it that Ball, who would eventually emerge on the right side of history, was in his own time on the wrong side of science? How did Hoffman, by contrast, gain scientific credibility despite being in fundamental error? From here, we are led to ask: What if being scientifically right does not entail being morally right? What would this mean for the modern conviction that quantitative analyses of racial disparities are an important tool in a broader project of liberation? What would it mean for the relationship between social justice and social science more broadly?

In brief, the story Muhammad tells is this. In 1894, in the pages of The Medical News, a weekly journal, Ball, a doctor at Eastern State Penitentiary in Philadelphia, responded to a study of racial disparities by another prison doctor, R. M. Cunningham. Cunningham’s work had shown stark racial differences in prison mortality rates, which he took as evidence for the prevailing theory that Black people were inherently more infirm than white people. In his letter to the editor responding to Cunningham’s view, Ball offered an opposing theory—that the observed disparities were due not to “physical distinctions” or “so-called hereditary and racial characteristics” but to the “differences in the environment [and] to the social differences which surround” them—that is, to the intense poverty, wretched living conditions, extreme political hostility, and broad exclusion from public life that characterized the Black experience after the Civil War. Entirely unmoved, Hoffman, writing in defense of Cunningham’s original study, refuted Ball with every tool in the expert’s kit—from new data and analyses to extensive appeal to various disciplinary authorities—concluding in his reply: “On the strength of the facts that are here brought together I cannot but repeat that the [racial] differences are physical and permanent.”

Ball’s subsequent reply to Hoffman contained more data, statistics, and analyses of his own. Although Ball does not appear to have pursued work of his own on racial statistics, the exchange in the pages of The Medical News indicates that he was unassailably scholarly and scientifically astute, measured and constructive. To truly assess the question of inherent Black infirmity and criminality, Ball writes, “we must compare individuals of the same class, in order to obtain proper statistics for medicine as well as for sociology.” Ball’s criticisms of his interlocutors are grounded in a concern for proper scientific methodology: “Figures in themselves mean nothing; they must be carefully analyzed and studied in connection with social conditions. To say that one city has a larger death-rate than another may be a statistical truth, but it is not a truth from which alone we can draw conclusions.”

Broadly understood, Ball’s responses read like a primer on careful statistical reasoning. A properly scientific orientation to the problem calls for data on Black and white individuals who share similar social and material conditions. Data that do not properly eliminate external sources for the observed disparities are prone to yield mistaken conclusions of inherent racial differences. As a general matter, analyses that rely heavily on statistics must justify the data on which the statistics are based. Why are these the right data from which conclusions should be drawn? In what ways might the data prefigure those conclusions? Do alternative ways of slicing up these data support alternative hypotheses? To the extent that scholars like Hoffman fail to take up this prior set of methodological questions, they cannot hope to answer the empirical question at hand. And their studies are in turn, as Ball puts it plainly, “not scientific.” He continues, “As stated before, this is not a race question: it is simply a plea for the sociologic analysis of statistics.”

These are eminently reasonable entreaties. And yet it is Hoffman who will get the final word, in the form of a 329-page book published two years later by the American Economic Association. The Race Traits and Tendencies of the American Negro (1896) was heralded as “by far the most thorough and comprehensive treatment of the Negro problem, from a statistical standpoint, which has yet appeared,” according to Black mathematician and later dean of Howard University Kelly Miller in the opening lines of his (still highly critical) review of the book—a work that “seems destined to awaken in the field of science and exact inquiry” a great public interest.

¤

When we say now, more than a century, that Ball was “right” in his dispute with Hoffman, we mean it in two ways. First, we mean that he was morally right to underscore entrenched social disadvantage and injustice rather than innate pathologies when discussing the problem of racial inequality. And it was morally right to use these points to advocate for social policies geared toward the alleviation of poverty and the improvement of public health, rather than toward support for eugenics programs. Second, we mean that Ball was right as a scientific matter: contrary to prevailing expert opinion, his skepticism about Cunningham’s and Hoffman’s statistical analyses was well-founded. There is no scientific basis to support the claim that Black populations are inherently more criminal or physiologically infirm than other racial groups. To the extent that observed disparities are at all telling of criminality, they result from social conditions, just as Ball had suggested. Factor these in, and the spiritual and physical “risks” presented by “racial traits” disappear entirely.

According to a certain common picture of science, these two ways of seeing Ball as “right” are linked, and not by accident. Hoffman and his ilk were racial ideologues, and their conclusions were driven primarily by commitment to that ideology, which prevented them from properly discharging their basic duties to scientific inquiry. Hoffman failed to properly consider the evidence in its entirety, attend to the full set of possible alternative explanations, and engage in critical self-reflection to scrutinize the assumptions he made when connecting evidence to explanation. It follows, the common interpretation goes, that he was never really doing science at all. Ball, by contrast, was not so weighed down by the political agendas that often cloud scientific judgment, and his comparatively more scientific—or equivalently, from this perspective, his less ideological—way of approaching the question is what directed him to the “right” conclusions.

In this version of the exchange, Ball represents the virtues of statistics and quantitative social science when performed rigorously and responsibly, while Hoffman represents all of its vices. This is obviously a vindicatory telling, as well as one that’s rather optimistic about the relationship between social science and social justice. Jettison bad racial ideology, this argument suggests, and proper scientific method will eventually correct its course and allow the truth to come out. It follows, then, that the reign of eugenicist racial statistics during the late 19th and early to mid-20th centuries was a mere blip in the broad march of scientific progress—no doubt a profoundly dangerous blip but a blip nonetheless. In fact, the story is even more vindicatory than that: not only does the project of statistical analysis about race emerge unscathed, but we also might even say that it took good racial statistics to overcome bad racial statistics.

The problem with such vindicatory interpretations is not that they are vindicatory per se. Rather, it is that a reading of history that takes vindication as its aim tends to minimize genuine tensions and controversies in the past in order to bolster the status of those beliefs and practices that “won out” according to the standards prevailing today. Thus, our present-day interest in defending proper scientific analyses of racial disparities, seeing them as deeply incompatible with essentialist theories of racial difference, drives us to interpret Hoffman’s work as—of course!—riddled with errors in reasoning, and Ball’s as a standard-bearer of statistical wisdom.

But to the extent that there is a genuine tension between scientifically credible conclusions and morally and politically right ones, the reading that takes Ball to have simply bested Hoffman on methodological grounds too easily dispenses with it. After all, it was Hoffman’s analyses and conclusions that bested Ball’s when it came to his era’s scholarly uptake and scientific consensus. It was Hoffman who was seen as the leading expert on the topic of racial disparities and whose statistical analyses would soon culminate in a study called the “most thorough and comprehensive treatment” of the subject. Ball, by comparison, was an amateur, a medical doctor by training who in all likelihood only ever dabbled in statistics and appears in the discipline’s annals merely because he took it upon himself to write a letter to the editor submitting his own pet theory. In today’s parlance, he was quite literally a “reply guy.” These facts must be explained—or, at least, explained away.

¤

If we set aside the too-convenient view that Ball simply outstripped Hoffman in scientific prowess, we might find ourselves puzzled by their divergent conclusions. How is it that one man’s “striking proof” can so easily be another man’s evidence to the contrary?

Statistics, as they like to say, is a damned lie. But this turning of the historical tables is in fact endemic to scientific inquiry. It is rooted in underdetermination—the fact that there is always, indissolubly, a logical gap between observation and hypothesis. Observations do not automatically constitute themselves as evidence; rather, they are imbued with evidential meaning only when combined with background assumptions. This is what Kuhn meant when he argued that data were always “theory-laden.” Thus, the “same” data may lead rationally to divergent hypotheses.

So, for Hoffman, the stark racial disparities in mortality and imprisonment rates—which prevailed nationwide, across jurisdictions North and South, rural and urban, and across differences in age and sex—were evidence of the “fundamental and permanent rather than social and transient” differences among races. The background assumptions here are that (1) various sources of racial difference, fundamental and permanent, or social and transient, do exist, and (2) there is no common social denominator that might present an alternative nonracialist explanation of the disparities. The latter assumption is the one Ball targeted as dubious in his original letter to the editor. For him, not assenting to this assumption rendered the numbers and figures cited by Hoffman evidentially inert, not proof of anything.

Since assumptions are essential to scientific inquiry, the question is not whether to make them but rather which ones are reasonable. But there’s a hitch: background assumptions cannot be sized up in the way direct objects of inquiry are, for that would invite an infinite regress of checking assumptions in order to check assumptions in order to check assumptions. As a result, assumptions are often held to the far less exacting standard of a gut check. The problem is that, in the case of a dispute over the nature of racial disparities, the gut check seems to beg the question at issue—i.e., what are the potential causes of observed disparities among racial groups? If an answer to this question is a prerequisite to any statistical analysis of the nature of racial difference, then reasoning becomes dangerously circular, and this, of course, cuts both ways: against Ball as well as against Hoffman.

The practical consequences involved in begging this question emerge more clearly when we recall that scientific inquiry is not a private but a social affair, and thus the background beliefs and assumptions of the individual expert must contend with those that populate the minds of other scientists (as well as the general public). A conclusion that relies heavily on a premise that is precisely the matter up for debate will fail to convince anyone who was not already convinced. Meanwhile, those not inclined to see it that way will likely find the reasoning plainly fallacious, which leaves their preferred hypothesis wide open for confirmation.

This helps to explain why, in an ideological environment permeated with beliefs about racial difference, it was Hoffman’s rather than Ball’s analyses that gained traction. It explains Hoffman’s success as being due not just to the fact that his conclusions affirmed many people’s preexisting social and moral views but also to the crucial fact that, because it cohered better with their background beliefs, it seemed to them more persuasive on scientific grounds. Here politics conditions not just scientific production but also its uptake, in just the way that inferential reasoning always bears the mark of the background against which it operates. The widely held assumptions that (1) there exist significant physical differences across races and (2) there are no systemic social or environmental differences in how Black populations and their white compatriots experience American life together make rational the inference that the Black population was indeed inherently less healthful and more criminal than the white population. From this vantage point, Ball’s insistence that more data should be collected before drawing this conclusion was less indicative of an exacting researcher’s scrupulous attention to all aspects of the evidence, and more the mark of a dogmatist’s persistent denial of basic scientific facts. Compare, for instance, the ceaseless filibustering of contemporary climate denialists, demanding that more data be collected before we can draw a connection between fossil-fuel combustion and global temperature changes. The rationality of rational inference lies in the eye of the beholder.

W. E. B. Du Bois had already noted this dynamic in his 1898 essay “The Study of the Negro Problems,” writing that “the scope of any social study is first of all limited by the general attitude of public opinion toward truth and truth-seeking” and that “the present condition of the public mind […] renders it almost impossible to get at the facts and real conditions.” But where Du Bois points to entrenched opinion as the primary obstacle facing statistical analyses of race, I view the “present condition of the public mind” as an ineliminable feature of scientific hypothesis confirmation itself. Insofar as the American public was not ready to recognize the natural equality of races as a genuine possibility, so Du Bois thought, no amount of statistical analyses of racial disparities would be able to vindicate the conclusion. For those negatively predisposed to the idea, evidence to the contrary seemed only to mount. For others, the new data and alternative explanations put forth by dissenters like Du Bois and Ball, and the ensuing rounds of argument, reinforced their sense of the issue’s abstruse complexity, and of its uncertain resolution. At some point, the mere fact of an issue’s having remained up for debate for so long itself becomes a pernicious ideological force, lending credence to even those hypotheses that are ultimately found to be false. All the while, the project of Black justice post-emancipation is postponed yet again, waiting for the results to come in.

¤

Unlike the vindicatory story, this version is a more skeptical account of the contribution of statistical analyses to racial justice. It is skeptical on three fronts: first, early statistical analyses in support of racial equality were not obviously more “scientific” than the analyses of their opponents; second, the structure of hypothesis confirmation itself imposes barriers to the persuasive power of racial statistics within scientific (and public) discourse; and third, this dynamic imposes limits on the broader political potential of scholarly debate about race. All of this cautions against the conviction that racial statistics, if only “done right,” will readily lend itself to the fight for racial justice. For what it’s worth, empirical evidence suggests that normative political views are often resistant to revision in light of new facts.

It is undeniable that the racial statistics of the late 19th and early 20th centuries was politically potent indeed, legitimating a politics of racial inequality centered on notions of innate Black pathology. This presents a puzzle: if racial statistics wielded in service of white supremacy can grow to become politically powerful, why can’t a different racial statistics that aims at racial equality be similarly productive? Put differently, if the risks of underdetermination, circularity, and begging the question are truly internal to scientific inquiry about the nature of racial disparities, then why don’t they serve to undercut equally all such analyses, eugenicist and egalitarian alike?

This is a big question, and its obvious answer—that the broader American public remains steadfastly racist well into the 21st century—should certainly not be understated. But this judgment encourages an overly simplistic picture—that American racism is a constant social fact onto which the science of the day inevitably grafts itself. This view overlooks the extent to which racialist and eugenicist theory is productive, not only built on the backs of preexisting racism and science (which the theory advances and enhances) but also conjoining them within a larger, mutually reinforcing network of political and scientific ideas.

Whereas conventional views see political ideology as anathema to proper science, the eugenics program both avows politics and is itself an essentially political project that rests on a set of scientific premises. The idea that improvements to the future of humanity can and should be pursued by altering its genetic heritage—culling certain segments of the population while encouraging others to reproduce in greater numbers—constitutes a powerful (if sickening) vision. It spells out in very concrete terms who shall and shall not have a place in the world. The assumption that there are genetic differences, indeed a genetic hierarchy, among the races is absolutely foundational to this vision. Racial statistics is part and parcel of the eugenics project, not merely neutral data that could either support or undermine it. Within this context, racial statistics come imbued with racialist meaning, at once plugging into an existing ideology and actively constructing it.

By contrast, today’s scientific consensus largely disapproves of such explicit importing of political beliefs into research on race. Even if individual researchers privately hold views about what racial disparities mean for a society riven by racial injustice, such commitments are not permitted to penetrate their work as science. Recall Ball’s insistence that “this is not a race question: it is simply a plea for the sociologic analysis of statistics.” The implication is that one need not make assumptions about social justice to settle the matter—in fact, one might fare better without such assumptions. If statistical science is properly conducted, the truth will eventually, inevitably win out.

I have already given some reasons to doubt this happy result. Insofar as the conclusions one draws about racial disparities depend heavily on one’s prior beliefs, then what counts as rigorous analysis and methodologically proper inquiry will vary by audience. But there’s yet another reason to be skeptical: a posture of neutrality can itself have serious political consequences. To the extent that no statistical analysis gives the final word, verdicts about race remain open to further revision. Of course, in a society where beliefs in racial hierarchy are deeply entrenched, findings of racial equality will almost certainly be challenged. And if such challenges are regarded as the result of genuine attempts at inquiry, then it follows that continual vacillation between competing explanations must be allowed and even encouraged. Any resulting sense, among the public at large, that the cause of racial equality rests on shaky or unsettled empirical grounds is simply a cost that must be paid to ensure the integrity of true science. To my mind, this price is too steep, and not worth paying.

But more importantly, this orientation toward research on race misguidedly strives to overcome precisely what so many early statisticians understood so well and seized upon: that the question of racial disparities is an inherently political issue. In the years following the Civil War, everyone knew that racial statistics was first and foremost a political enterprise that bore immediately on the question of what was to be done about emancipated Black people. It was neither the case then, nor is it the case now, that interest in these questions was motivated by a purely scientific curiosity about what the world is like. Rather, these questions have always been motivated by concrete political concerns about a society structured and stratified by race. All people who live in this society have material stakes in how this question is answered, and this plain fact is why the issue will always have an audience and be a source of political conflict—and, in turn, a focus of social-scientific debate. This explains why eugenic and racialist thought is continually being revived, even while its scientific basis is repeatedly debunked.

Moreover, any attempt to install an intellectual division of labor—first the science, then the politics—cannot actually strip the scientific work of political content, because statistical meaning-making about racial inequality is necessarily political. Decisions about what evidence to marshal and what hypotheses to consider inform all statistical analyses, and, in this context, these decisions implicate background beliefs regarding race and racial difference.

There has always been an air of embarrassment about the fact that scientific findings inevitably depend on background beliefs and assumptions, which are not themselves amenable to empirical validation. This anxiety follows from the view that the more independent of assumption a scientific finding is, the stronger it must be. This approach might well apply to scientific issues about which people do not already hold firm beliefs—e.g., what happens when black holes collide or how cells keep time. But race is clearly not one of those issues. Every person comes to social-scientific research on race preloaded with a robust set of beliefs about the subject. Scientific conclusions are either successfully woven into that belief set or else rejected by it.

This suggests that those who seek to reclaim racial statistics from its eugenic past are unlikely to make inroads among the unconverted simply by presenting findings of social inequality. Instead, a truly emancipatory social science will need to engage competing racial ideologies directly by presenting a holistic alternative picture of social reality. This involves bringing to bear political views about how society works and weaving together disparate matters of fact regarding racial difference into a broader theory that explains why the world looks the way it does. If part of what has made eugenic science so enduringly compelling is that it presents its audiences with both “scientific” data and conclusions and a political narrative with which to interpret those data and conclusions, uniting them into a normative picture of social reality, then the only alternative that can challenge such a robust system of thought is an equally powerful, all-encompassing political vision regarding where we came from, where we are now, and where, in light of these facts, we go from here.

¤

Featured image: Samuel Cowperthwaite. The State Penitentiary, for the Eastern District of Pennsylvania, 1855. Library of Congress (2021670376). CC0, loc.gov. Accessed February 10, 2025. Image has been cropped.