Journal of Interdisciplinary History, xxxiii:1 (Summer, 2002), 21–46.

Howard Bodenhorn The Advantage: The Biological Consequences of Complexion in Rural Antebellum Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 For, as the whites have their blond and brunette, so do the have their chocolate, chocolate-to-the-bone, brown, low-brown, teas- ing-brown, yellow, high-yellow and so on. The difference on our side is so much more interesting. —Claude McKay, quoted in Joel Williamson, New People: and Mutattoes in the United States Langston Hughes, the most prominent writer of the Harlem Renaissance emphasized skin color throughout his ªction. At different times, he referred to African- as brown, light- brown, golden, yellow, high-yellow, almost white, blond, three- quarters pink, high-toned, coffee with cream, and cafe-au-lait. In Hughes’ ªction, complexion was paramount because it created in- terpersonal tensions, reºecting larger social dynamics. African- American men in Hughes’ ªction expressed a preference for light- skinned women, and dark-skinned women resented both the men who acted on that preference and the women who beneªted from it.1 Historians of race are quickto note that these tensions were not just the stuff of ªction. They regularly played out in real- world race relations. Leaders in the late nineteenth and early

Howard Bodenhorn is Associate Professor of Economics, Lafayette College, and Research Associate, nber. He is the author of A History of Banking in Antebellum America:Financial Mar - kets and Economic Development in an Era of Nation-Building (New York, 2000); “A Troublesome Caste: Height and Nutrition of Antebellum Virginia’s Rural Free Blacks,” Journal of Economic History, LIX (1999), 972–996. The author would like to thank Claudia Goldin, Whittington Johnson, John Komlos, Robert Margo, Michelle McLennan, Carolyn Moehling, Deborah Rosen, Joel Williamson, an anonymous referee, and the editor and seminar participants at Lafayette College for com- ments on earlier drafts. He is also grateful to the librarians at the Library of Virginia for assis- tance in locating materials and to Lafayette College and the Robert King Mellon Foundation for ªnancial support.

© 2002 by the Massachusetts Institute of Technology and the editors of The Journal of Interdisciplinary History.

1 Langston Hughes, Short Stories (New York, 1996). 22 | HOWARD BODENHORN twentieth century African-American community were more likely than the general population to be light-complected mulat- toes. Hughes was; W. E. B. Du Bois was; and so were Booker T. Washington, Eubie Blake, Paul Robeson, and Walter White. Wil- liamson argues that these men’s ascendency to leadership was not coincidental. They were the sons, grandsons, and great-grandsons of light-complected men who formed the African-American elite Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 before them. A fair complexion conferred a decided advantage in the late nineteenth and early twentieth centuries. Recent discus- sions suggest that the mulatto advantage continues into the twenty-ªrst century. In some circles, mulatto has become chic.2 Whether a light complexion conferred similar advantages be- fore 1865 remains unresolved. Most historians of the African- American experience accept a long-standing characterization of two Souths: Upper and Lower. Lower South whites, especially urban sophisticates in Charleston, Savannah, Mobile, and , drew ªne distinctions based on color. As in the West In- dies and , light-complected African-Americans were treated more generously than dark-complected peoples. Up- per South whites, on the other hand, are generally portrayed as es- chewing even the crudest color distinctions, seeing the world in blackand white. There was no middle ground in the Upper South. In their relations with the white community, Upper South mulattoes faced all the disadvantages of being blackand none of 3 the beneªts of a white heritage. THE MULATTO ADVANTAGE A close reading of the traditional interpretation of an Upper South–Lower South dichotomy reveals a fundamental contradic- tion. On one hand, historians argue that Upper South whites re- fused to recognize any color distinctions and pushed all African- Americans down into a mass of “blackness.” Yet, on the other hand, historians argue that complexion distinctions were vitally important within the African-American community itself. It

2 Joel Williamson, New People:Miscegenation and Mulattoes in the United States (New York, 1984), 145; Claudine Chiawei, Half and Half (New York, 1998). 3 Several historians have made or repeated the claim about light-complected African- Americans in the and South America. See, for example, Ira Berlin, Slaves without Masters (New York, 1974), 183–216; John G. Mencke, Mulattoes and Race Mixture:American Attitudes and Images, 1865–1918 (Ann Arbor, 1979), 10; Williamson, New People, xiv; Whitting- ton B. Johnson, Savannah, 1788–1864 (Fayetteville, 1996), 78; James Oliver Horton and Lois E. Horton, In Hope of Liberty:Culture, Community, and Protest among Northern Free Blacks, 1700–1860 (New York, 1997), 84–85. THE MULATTO ADVANTAGE | 23 seems unlikely that complexion-based preferences within the Af- rican-American community would have persisted without rein- forcement from the white community. If African-Americans esteemed “whiteness,” they are likely to have done so because lighter skin conferred some advantage in a white-dominated world. Indeed, complexion differences generated unmistakable socioeconomic differences in the antebellum Upper South. Mu- Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 lattoes climbed the agricultural ladder more quickly and accumu- lated considerably more wealth than blacks in antebellum and Virginia.4 This article utilizes a unique data set ultimately to conªrm the hypothesis that complexion differences were as important a deter- minant of socioeconomic status in the rural Upper South as in the urban Lower South. Using information contained in the registers of more than 10,000 free African-Americans from twenty-three rural Virginia counties and drawing on recent advances in anthropometric history, this article shows that the advantages of complexion were undeniable. Light-skinned African-American children beneªted from nutritional advantages in adolescence that dark-skinned children did not realize. These advantages in adoles- cence translated into substantially taller light-skinned adults. Put simply, color mattered in the antebellum Upper South, too. His- torians may have failed to recognize this fact because a visible, outspoken mulatto elite failed to materialize in the rural Upper South. But the emergence of a mulatto elite and the primacy of color were not synonymous. blacks, whites, mulattoes, and virginia law and custom, 1620–1860 Africans ªrst arrived in the Virginia colony in 1619 or 1620 and almost immediately formed intimate relation- ships with whites. The ªrst reference to an African appeared in Virginia’s legislative record in 1630; it represented the opening salvo in a long battle against miscegenation. Virginia’s council or-

4 Tommy Bogger, Free Blacks in Norfolk, Virginia, 1790–1860:The Darker Side of Freedom (Charlottesville, 1997), 104; Mencke, Mulattoes and Race Mixture, 2; Leon F. Litwack, North of :The in the Free States, 1790–1860 (Chicago, 1961), 182. Johnson, Black Savannah, 16–17, makes a forceful case for the importance of complexion in the Lower South and how it was reinforced by white attitudes. Bodenhorn, “The Complexion Gap: The Economic Consequences of Color among Free in the Rural Antebellum South,” Ad- vances in Agricultural Economic History, II (forthcoming 2002). 24 | HOWARD BODENHORN dered Hugh Davis soundly whipped for “lying with a Negro,” an act he was required to acknowledge publicly on the Sabbath. A decade later, Robert Sweet was forced to do penance in church for impregnating a blackwoman. The woman was whipped. 5 Concerns about miscegenation ultimately provoked a signiªcant colonial departure from English law. English tradition held that a child’s status followed the father’s. In miscegenation Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 cases, identiªcation of the father was often problematic; it was simpler to inhere the mother’s status to the child. In 1662, Virginia law made the mulatto child of a slave woman a slave. Until 1691, the mulatto child of a white woman was free, but an act of that year imposed a penalty of ªve years’ forced servitude on the white mother of a mulatto child and thirty years of servitude on the child itself.6 Although they labeled it servitude, most colonials viewed the children’s punishment as slavery, if only for a term of years. Many masters, however, ignored the term limitation and kept their ser- vants in lifetime bondage. Others released them only when the courts forced them to do so. The case of Ann Redman may have been typical. Ann, a mulatto daughter of an English woman, was freed only after a court ordered her “freed from slavery and dis- charged from the service of Thomas Lloyd” of Richmond County, Virginia, in 1697. Courts were forced to intercede in such instances because masters often represented their mulatto term servants as slaves and sold them to unsuspecting buyers who held them past their thirty-ªrst birthday. The practice became so egregious that in 1765 the legislature levied substantial punish- ments for misrepresenting and selling term servants as slaves.7 Interracial affairs, once discovered, carried a stigma in the United States not seen in other slave societies. In the British West Indies, gender imbalances among whites led to widespread misce- genation. British planters in the West Indies were probably as squeamish about interracial mixing at ªrst as their North Ameri- can counterparts, but demographic forces overrode social mores. Horowitz argues that by the mid-eighteenth century, miscegena-

5 Jane Purcell Guild, Black Laws of Virginia (New York, 1969), 21. 6 Ibid., 23–24. 7 James Hugo Johnston, Race Relations in Virginia and Miscegenation in the South, 1776–1860 (Amherst, 1970), 178. Johnston reports several other cases besides that of Ann Redman. Guild, Black Laws, 58. THE MULATTO ADVANTAGE | 25 tion between white planters and blackslaves was widespread, mu - latto progeny commonplace, and manumission the rule.8 Similar practices never evolved in . Despite more balanced white gender ratios, white men (single, married, and widowered) carried on illicit affairs with slave women and produced mulatto children. By contemporary North American standards, miscegenation was reprehensible, but manumitting the Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 child was a singularly dangerous, antisocial act. Better the half- white offspring suffer a life in bondage than society be overrun with free African-Americans. After a brief post-Revolutionary ex- periment with liberalized manumission, Virginia law discouraged it, ªrst by requiring planters to post bonds guaranteeing that man- umitted slaves would not become wards of the state and then in 1806 by requiring manumitted slaves to emigrate within twelve months of emancipation. Manumitted slaves who failed to emi- grate faced sale backinto slavery. Few masters would free a fa- vored slave only to see her sold into the service of another.9 Not unexpectedly, legislative attempts to thwart miscegena- tion and manumission failed. By 1860, there were about 24,000 free mulattoes (40 percent of the free African-American popula- tion) and about 70,000 mulatto slaves (about 15 percent of the slave population). A thorny political question revolved around how to deªne the mixed-race population. Davis argued that the “one-drop rule” (a single drop of blackblood made people black) that became the standard under Jim Crow had colonial anteced- ents. He provided evidence of the one-drop rule in the early eigh- teenth century, but recognized that the mulatto’s exact legal status remained uncertain as late as the 1750s. Degler, too, cited cases de- cided by southern courts in the eighteenth and nineteenth centu- ries to support his contention that blacks and mulattoes were, as far as southern law and southern society was concerned, one and the same.10 Degler’s interpretation seems excessive given that, in 1785, Virginia legally deªned a person as Negro if he or she had one

8 Donald L. Horowitz, “Color Differentiation in the American Systems of Slavery,” Journal of Interdisciplinary History, III (1973), 509–541. 9 Carl N. Degler, Neither Black nor White:Slavery and Race Relations in and the United States (New York, 1971), 194–195. 10 F. James Davis, Who is Black? One Nation’s Deªnition (University Park, Pa., 1991), 31–34; Degler, Neither Black nor White, 241–243. See also Mencke, Mulattoes, ix–x. 26 | HOWARD BODENHORN blackgrandparent. For many Virginians, however, this deªnition of whiteness was too generous. Some contemporaries complained that the law created a distinct class of people “who were signiªcantly black, visibly black, and known to be black, but by the law of the land and the rulings of the court had the privileges of whites.” That the law recognized visibly African-American people as white unsettled whites who preferred sharp and un- Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 equivocal deªnitions. Historians insist that Upper South whites may have been forced to recognize the distinction de jure but re- fused to do so de facto. Instead, Upper South racial interactions aimed to make all African-Americans black even while Lower South and West Indies whites elevated mulattoes to a distinct in- termediate class.11 The different attitudes of Upper and Lower South whites are often attributed to regional differences in mulatto ancestry. As noted previously, West Indian and Lower South mulattoes tended to be the offspring of wealthy white men and blackwomen, either slave or free. Thus, mulattoes were recognized as the offspring of the elite and provided with many of the advantages that followed from wealth. Nearly all were manumitted, most were educated, and many even inherited from their father’s estate. Upper South mulattoes, however, tended to be the offspring of poor disfran- chised whites and even poorer unfranchised slaves. These children were born poor and were thought to carry an indelible stain of poverty throughout their lives. A fair complexion signiªed gentry in New Orleans; it signiªed peasantry in Richmond.12 Upper South whites are portrayed as unremitting in their ef- forts to disavow the mulatto’s whiteness, and historians are unre- mitting in their efforts to portray whites in this way. In summarizing his thesis, Degler states that there “are two qualities in the United States racial pattern: white and black. A person is one or the other; there is no intermediate position.” Evidence presented below contradicts this interpretation. Both the African- American and white communities drew sophisticated color lines. Both recognized subtle gradations rather than sharp distinctions. Why else would contemporary whites describe individual Afri- can-Americans as brown, reddish brown, copper, ginger, chest-

11 Guild, Black Laws, 29; Williamson, New People, 13. 12 Mencke, Mulattoes, and Horowitz, “Color Differentiation,” provide variations on this argument about different attitudes of Upper and Lower South whites. THE MULATTO ADVANTAGE | 27 nut, or nutmeg? Seeing such differences and acting on them are very different things, but the anthropometric evidence implies that both communities treated fair-complected peoples preferen- tially.13 anthropometric history and the virginia registrations of free african-americans During the past two decades, a Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 growing number of economic historians have turned to anthropometry to gain insights into the biological environment in which various peoples lived and to infer something about their standards of living. Various anthropometric measures, such as height at age, weight at age, age at adolescent growth spurt, and terminal adult height, among others, reveal something about the well-being of human populations. Steckel shows that average adult height is highly (though nonlinearly) correlated with per ca- pita income among modern populations. The strength of this rela- tionship for historical populations is less certain, but the connection between height and health and economic well-being should hold.14 Those unfamiliar with anthropometry are often skeptical of the claim that height at age or average terminal adult height can stand as a proxy for socioeconomic status and other environmental conditions under which people mature. They tend to regard height as an inherited trait, which to some degree, it is. Genetics deªnes an upper bound; we do not produce many giants, no mat- ter how wealthy or well-fed the child. Yet the average adult height of a well-deªned population is strongly inºuenced by its typical environmental conditions during the growing years. Aver- age height reºects cumulative nutritional intakes tempered by the hygienic, infectious, and psychological and social stresses experi- enced during childhood and adolescence. Tanner ably summa- rized the hypothesis for those conversant with statistical terminology. “In the language of analysis of variance,” he wrote, “most of the within-group variation is due to heredity, and most of the between-group variation is due to childhood environ-

13 Degler, Neither Black nor White, 102. 14 Two valuable anthropometric collections are John Komlos and Timothy Cuff (eds.), Classics in Anthropometric History (Katherinen, Germany, 1998); Richard Steckel and Roderick Floud (eds.), Health and Welfare during Industrialization (Chicago, 1997). Steckel, “Height and Per Capita Income,” Historical Methods, XVI (1983), 1–7. 28 | HOWARD BODENHORN ment.” Armed with this fundamental insight, economic historians are exploiting long-ignored data sets to draw inferences about comparative standards of living that complement more conven- tional measures like gdp, wages, etc.15 Despite the difªculties inherent in sorting out the causes of stature, much is understood about the relationship between envi- ronment and growth. Short periods of acute nutritional deªciency Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 or prolonged periods of moderate deªciency have few negative consequences for terminal adult height. They simply delay the age at which the adolescent growth spurt commences and increases the age at which terminal height is reached. Modest or short-lived deªciencies rarely have noticeable effects on attained adult height because moderately undernourished people continue to grow past the age at which adequately nourished populations cease growing. Males in a well-fed, healthy population tend to stop growing by age eighteen. In a moderately undernourished population, growth continues into the early twenties; at age twenty-three or twenty- four, the two groups attain comparable ªnal heights. Severe or prolonged periods of inadequate nutrition during the critical de- velopmental growth years (birth to four years and the adolescent growth spurt), however, produce shorter adults. Stunting occurs because prolonged malnutrition shunts children onto a lower growth path. In other words, “malnourished and unhealthy popu- lations are short, while well-fed and healthy populations are tall,” relative to their genetic potential.16 Diet and disease are not the only proximate determinants of attained adult height. Height is a ªne-scale gauge in that it is a measure of net nutrition, reºecting both the intake of nourishment as well as claims made on the body by work, disease and infection, psychological/social stresses, and the efªciency with which the body transforms inputs (food) into outputs (basal metabolism, workeffort, disease resistance, and growth). A given quantity and quality of food that in one place and time produced adults of a particular average height, may produce adults of another size un- der different circumstances. Children who workhard during their growing years become shorter adults than children who expend

15 James M. Tanner, “Growth in Height as a Mirror of the Standard of Living,” in Komlos (ed.), Stature, Living Standards, and Economic Development (Chicago, 1994). 16 Phyllis Eveleth and Tanner, Worldwide Variation in Human Growth (Cambridge, 1990), 176; Floud, “Heights of Europeans since 1750,” in Komlos (ed.), Stature, 14. THE MULATTO ADVANTAGE | 29 less workeffort during the same years and eat the same amount of calories. Anthropometric data about antebellum Virginia’s free Afri- can-Americans were gathered from registers kept by county clerks in twenty-three rural counties. Beginning in 1793, Virginia law required that every free-born or manumitted African-American register. Clerks kept a ledger and provided registrants with a copy, Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 which served as their freedom papers. Most registrations provided detailed descriptions of individuals, recording the registrants’ names along with any known aliases, age, gender, height, com- plexion, any identifying scars or other notable physical attributes, sometimes an occupation, rarely the county of birth, and whether the individual had been free-born or manumitted. Registers of free-born individuals often identiªed the mother. Registers of manumitted slaves included the name of former owners and, sometimes, the date that the deed of manumission was recorded by the clerk(which may or may not have been the date of manu- mission itself ). Because of the detail that they provide, these regis- ters represent a valuable source for examining the connection between complexion and childhood development and terminal adult height.17 One point of concern with the registration data is that only a fraction of those legally required to register actually did so. In Campbell County, for example, the clerk’s ledger dated 1801 through 1850 included information on only 287 African-Ameri- cans. The 1850 census enumerated 846. Clearly, many failed to register. The issue is whether nonuniversal registration biases the sample. Even with lax compliance, the registers should fairly rep- resent Virginia’s free African-American population. Registration was unlikely to have been functionally related to stature. It was more likely a function of proximity to the courthouse and the clerk’s ofªce. It was also likely to have occurred just prior to or just after obtaining employment outside the home. The fact that Virginia law required free African-Americans to provide a copy of

17 Slaveholders often recorded a deed of manumission years before a slave was freed, condi- tioned on the good behavior of the slave. Many slaveholders believed that the promise of freedom prompted greater workeffort and better behavior. Given this practice, it was impos - sible to determine if there was a lag between the date of manumission and the date of the reg- istration. In some instances, the newly freed slave registered on the day that he or she was freed. In other instances, many months or years may have passed. 30 | HOWARD BODENHORN their registration to employers probably explains the large propor- tion of registrants between ªfteen and twenty-ªve, ages at which full-time employment outside the home typically commenced. Yet, less-than-universal registration may introduce a bias into the statistics derived from the registrations. Registration com- monly occurred either at manumission or at the start of employ- ment outside the home. If lighter-skinned people were more Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 likely to be manumitted and receive better treatment while en- slaved, they were more likely to be measured and were likely taller. Similarly, if light-skinned African-Americans had better job prospects, then relatively more lighter complected African-Ameri- cans would have registered. This differential propensity to register, if it existed, may not bias estimates of the differences in heights be- tween mulattoes and blacks, but it may introduce an upward bias in the estimated average height of African-Americans as a whole and distort comparisons of height differences between free Afri- can-Americans and whites or slaves. Although the potential bias is small, it should be taken into account, especially if the statistics re- ported herein are used comparatively. growth during childhood and adolescence Table 1 treats the sample of 3,400 registered African-American children. Parents were more likely to register male than female children, reºecting a greater likelihood for boys to work outside the home or serve as apprentices earlier in life. That most were born be- tween 1800 and 1839, even manumitted children, is surprising given that Virginia law required slaves manumitted after 1803 to emigrate. The continuation, even acceleration, in the number of manumissions suggests that the law was not well enforced. The Piedmont is slightly overrepresented at the expense of Tidewater counties, but not so much as to engender a signiªcant bias. Ac- cording to the 1860 census, the Tidewater region was home to 47.8 percent of free African-Americans, the Piedmont about 42.2 percent, and the Blue Ridge 10.0 percent; the register sample is broadly representative of the geographical distribution of free peoples.18

18 This article uses the modern boundaries of Virginia. The counties now included in West Virginia contained few free African-Americans or slaves. Moreover, Brunswick, Chesterªeld, and Fairfax counties, which lay on the border between the Tidewater and the Piedmont, THE MULATTO ADVANTAGE | 31 The complexion panel of Table 1 highlights the complexion attributes of the children and youths sample. The proportions in each division for free-born males and females are similar, though female children and youths were less apt to be described as black or darkand more apt to be described as “bright.” Among manu - mitted slaves, clerks showed a similar propensity to describe males as darker than females. It is unclear whether slaveholders manu- Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 mitted darker males and lighter females, or if clerks saw male chil- dren as darker than female children even if they were not. The latter seems improbable, given that clerks recorded similar propor- tions of mulatto, brown, and yellow complexions across all four categories. So long as clerks consistently deªned complexions within genders, the results reported below appear to be secure. Table 2 reports height-at-age (at last birthday) estimates for blacks and fair-complected mulattoes, as well as the number of observa- tions at each age. Because of the small number of observations at some ages, heights were estimated using ordinary least squares re- gressions (ols) (results not reported). Several features of the height-at-age proªles deserve comment. First, darkand light indi- viduals were of largely comparable stature until about age sixteen for males and age thirteen for females, or approximately the ado- lescent growth spurt. During and after the growth spurt, however, dark- and light-complected individuals followed divergent growth trajectories. From nearly equal stature at age sixteen, light males attained an average terminal stature of 68.5 inches compared to a terminal stature of just 67.1 inches for darkmales. The light- complexion advantage during the adolescent spurt was not likely to have resulted from genetic differences between darkand fair in - dividuals. Human biologists have long noted that people of West- ern European and western-African origins raised in similar environments (namely, North America) attain similar adult stat- ure. Nor was miscegenation likely to have produced hardier indi- viduals more resistant to environmental insults. A more convincing explanation is that light-complected mulatto children were raised in more salubrious environments. Mulatto heads of household rose higher on the occupational hierarchy sooner and accumulated signiªcantly more wealth than blackheads of house - were classiªed as Piedmont. If these populations were distributed across the two regions, the proportions would more closely reºect the census proportions. 32 | HOWARD BODENHORN

Table 1 Characteristics of the Virginia Free African-American Register Sample—Males One to Twenty-One Years, Females One to Eighteen Years free-born manumitted free-born manumitted males (%) males (%) females (%) females (%) birth cohort 1770s 0.2 1.7 0.0 0.0 Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 1780s 1.5 6.9 0.8 4.0 1790s 5.0 10.7 1.0 7.5 1800s 10.4 7.6 5.9 5.7 1810s 22.2 20.3 22.0 9.8 1820s 25.3 19.9 21.4 21.3 1830s 27.1 21.3 33.1 25.3 1840s 8.2 9.6 16.2 21.1 1850s 0.2 2.1 0.8 5.2 region of registration Tidewater 39.3 21.2 29.9 16.1 Piedmont 49.7 60.7 62.9 69.5 Blue Ridge 11.0 18.2 7.2 14.4 complexion Black20.2 38.8 12.7 19.5 DarkMulatto 21.3 16.8 17.1 21.8 Mulatto 11.7 6.2 16.0 11.5 Brown 14.4 9.6 15.1 16.1 Bright Mulatto 24.6 22.0 32.7 24.7 Yellow 7.8 6.6 6.4 6.4 N 1,892 292 1,049 174 holds. Mulatto parents tended to be more well-to-do; they had greater access to food, health care, housing, and clothing. In short, mulatto children matured in a better net-nutritional environment, and the advantages became apparent in adolescence.19 A second notable feature is that light-complected females, like their light male siblings, grew signiªcantly taller than dark

19 Separate ols regressions were computed for light males, darkmales, light females, and darkfemales. Regressions with height at age as dependent variable included the following in - dependent variables: a constant, a dummy variable for age (one through twenty for males, one through seventeen for females), decade of birth (1770s through 1850s), region of registration (Blue Ridge Piedmont, and Tidewater), and a dummy variable if the individual was manu- mitted. Adjusted R-squares exceeded 0.83 for all regressions, all F-statistics of joint signiªcance exceeded 33.3, and most t-statistics were signiªcant. Detailed results are available from the author on request. Eveleth and Tanner, Worldwide Variation; Bodenhorn, “Com- plexion Gap.” THE MULATTO ADVANTAGE | 33 notes Tidewater counties include Charles City, Essex, Lancaster, Middlesex, Northampton, Surry, and Westmoreland. Piedmont counties include Arlington, Bedford, Brunswick, Campbell, Chesterªeld, Cumberland, Fairfax, Fauquier, Fluvanna, and Orange. Blue Ridge counties include Alleghany, Augusta, Botetourt, Montgomery, Roanoke, and Rockingham. Black complexion includes individuals described as “very” black. Brown in- cludes copper, chestnut, and reddish brown. Yellow includes tawny, olive, and nearly white. sources Alleghany County, Register of Free Negroes and Mulattoes in Alleghany County, 1855– lov

1856, Library of Virginia (hereinafter ); Arlington County, Register of Free Negroes, 1797– Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 1861, lov; Bedford County, Register of Free Negroes, 1803–1820, lov; Dorothy A. Boyd-Rush (ed.), Register of Free Blacks, Rockingham County, Virginia, 1807–1859 (Bowie, Md., 1992); idem (ed.), Free Negroes Registered in the Clerk’s Ofªce Botetourt County, Virginia, 1802–1836 (Athens, Ga., 1993); Katherine G. Bushman (ed.), The Registers of Free Blacks 1810–1864, Augusta County, Virginia and Staunton, Virginia (Verona, Va., 1989); Campbell County, Register of Free Negroes, 1801–1850, lov; Charles City County, Register of Free Negroes, 1839–1864, lov; Ches- terªeld County, Register of Free Negroes, 1804–1854, lov; Cumberland County, Register of Free Negroes, 1821–1863, lov; Richard B. Dickenson (ed.), Entitled!:Free Papers in Appalachia Con - cerning Antebellum Freeborn Negroes and Emancipated Blacks of Montgomery County, Virginia (Washington, D.C., 1981); Essex County, Register of Free Negroes, 1843–1861, lov; Fluvanna County, Register of Free Negroes, 1851–1862, lov; Karen King Ibrahim, Karen Hughes White, and Courtney Gaskins (eds.), Fauquier County, Virginia Register of Free Negroes, 1817–1865 (1993); Dennis Hudgins (ed.), Surry County Register of Free Negroes (Richmond, Va., 1995); Lancaster County, Register of Free Negroes, 1803–1860, lov; Frances Bibbins Latimer (ed.), The Register of Free Negroes, Northampton County, Virginia, 1853 to 1861 (Bowie, Md., 1992); Middlesex County, Register of Free Negroes, 1800–1860, lov; Orange County, Papers, 1837–1850, lov; Roanoke County, Free Negro Register, 1838–1864, lov; Donald Sweig (ed.), Registrations of Free Negroes Commencing September Court 1822, Book No. 2 and Register of Free Blacks 1835 Book 3 (Fairfax, Va., 1977); Westmoreland County, Register of Free Negroes, etc., 1828–1849, lov; Frances Holloway Wynne (ed.), Register of Free Negroes and Also of Dower Slaves, Brunswick County, Virginia, 1803–1850 (Fairfax, Va., 1983). girls. As with males, light and darkgirls demonstrated comparable growth trajectories through childhood and up to adolescence, at which time light-complected girls raced past darkgirls. At adult - hood, light-complected women were, on average, more than 2 inches taller than dark-complected women. Due to the small number of observations at some ages and the presence of outliers that skew the results at others, the height-at- age proªles reported in Table 2 were smoothed using the nonlin- ear methods suggested by Preece and Baines. From those smoothed heights, centiles of modern stature were calculated us- ing the method and benchmarks provided by Steckel; they are presented in ªgures 1 and 2. Centiles provide a useful benchmark for comparison in that the modern ªftieth centile is that attained by an average child at each age. If the average free African-Ameri- can growing up in antebellum Virginia had been as healthy and well fed as the average American in the late twentieth century, the 34 | HOWARD BODENHORN

Table 2 Estimated Height at Age, Light- and Dark-Complected Free Afri- can-Americans—Males One to Twenty-One Years, Females One to Eighteen Years light males dark males light females dark females age (inches) (inches) (inches) (inches) 1 28.7 (4) 27.2 (1) 27.4 (7) — (0) 2 32.4 (7) — (0) 32.5 (4) 30.5 (2) Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 3 35.0 (12) 34.3 (2) 36.3 (6) — (0) 4 38.3 (10) 38.6 (5) 35.4 (11) 37.1 (2) 5 40.8 (6) 40.4 (4) 41.0 (9) 42.5 (1) 6 42.4 (7) 41.7 (5) 43.9 (11) 40.8 (3) 7 45.6 (10) 45.7 (2) 45.3 (11) 47.8 (1) 8 46.9 (10) 49.7 (8) 47.2 (8) 50.2 (6) 9 51.2 (12) 50.9 (13) 49.9 (16) 50.3 (3) 10 52.8 (14) 51.9 (3) 53.0 (8) 50.7 (5) 11 54.7 (11) 53.5 (8) 55.1 (14) 55.7 (5) 12 57.6 (14) 55.0 (9) 55.2 (23) 56.6 (6) 13 57.8 (18) 59.1 (9) 59.9 (18) 59.7 (10) 14 60.6 (21) 59.8 (15) 60.7 (34) 60.6 (12) 15 62.6 (22) 63.5 (13) 61.5 (43) 60.7 (12) 16 65.2 (19) 64.5 (23) 62.3 (52) 62.3 (29) 17 65.5 (33) 64.7 (25) 62.5 (60) 60.8 (27) 18 67.0 (39) 66.8 (36) 62.9 (129) 60.8 (38) 19 68.3 (51) 67.7 (31) 20 68.3 (81) 67.5 (54) 21 68.5 (300) 67.1 (222) N 700 488 464 162 notes Heights estimated by ordinary least squares regressions (not reported) that include a constant, a manumission dummy, age dummies (one through twenty for males, one through seventeen for females), region of registration, and birth cohort. Adjusted R-squares all ex- ceeded 0.85, and all F-tests failed to reject joint signiªcance. See Table 1 for cohort and re- gion deªnitions. Blankcells imply no observations. Numbers in parentheses are number of observations at each age. sources See Table 1.

ªgures would display straight lines at the ªftieth centile. That is, the average child at each age would have been as tall as, or taller than, 50 percent of the children in his or her age cohort (and, hence, shorter than the other 50 percent). Thus, the use of centiles facilitates comparisons between groups and periods.20

20 M. A. Preece and M. J. Baines, “A New Family of Mathematical Models Describing the Human Growth Curve,” Annals of Human Biology, V (1978), 1–24. The curves were ªtted us- ing Model II. Steckel, “Percentiles of Modern Height Standards for Use in Historical Re- search,” Historical Methods, xxix (1996), 157–166. Fig. 1. Centiles of Modern Stature Males, One to Twenty-one Years Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021

Fig. 2 Centiles of Modern Stature Females, One to Eighteen Years 36 | HOWARD BODENHORN Both ªgures suggest that, compared to modern populations, the adolescent growth spurt occurred relatively late among Vir- ginia’s African-American males. Both groups demonstrate signi- ªcantly accelerated growth beginning at age sixteen and a period of signiªcant catch-up (relative to modern standards). Only among light-complected males, however, did the catch-up growth persist long enough for light-skinned men to reach the Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 thirty-seventh centile of modern stature. Dark-skinned men at- tained only the twentieth centile. In Figure 2, females display sim- ilar, if less pronounced, adolescent growth trajectories. Light- and dark-complected girls attained similar heights at age up to about age twelve or thirteen (approximately the same age at which mod- ern female populations experience the adolescent growth spurt), but their experiences diverge markedly thereafter. The typical light-skinned women eventually attained the twenty-fourth centile of modern stature; the typical dark-skinned woman at- tained just the ninth. Even though light-skinned men attained the thirty-seventh centile of modern stature, and both light and darkwomen entered the adolescent growth spurt at approximately the same age as modern populations, the evidence does not support a sanguine in- terpretation of early African-American health and nutrition. Hu- man biologists identify two critical periods in human growth, birth to four years of age and adolescence. The typical antebellum African-American infant—male or female, light or dark—at less than the tenth centile, and certainly those below the ªrst centile up to age three, would provoke consternation in a modern pedia- trician’s ofªce. By modern standards, antebellum African-Ameri- can infants were severely stunted. Dark-complected adults of both sexes, but especially women, were nearly so. At only the ninth centile, dark-complected women were as short as, or shorter than, 91 percent of modern U.S. women. Combined with evidence on a delayed and short-circuited adolescent growth spurt for many youths and evidence that house- holds with blackheads of family were poorer than households with mulattoes at their head, growth in infancy implies that all groups, but especially blacks, were born into, matured, and labored under conditions of moderate to acute and prolonged nutritional deprivation. Compared to well-fed, healthy North American populations of the present day, African-American chil- THE MULATTO ADVANTAGE | 37 dren in the nineteenth century faced an inhospitable environment that stunted growth in infancy and blunted the recovery or catch- up growth in late adolescence that is evident in some undernour- ished populations. Yet, even in an environment that was generally harsh, blackchildren faced especially severe conditions. Seen in concert with information from the 1860 census about the poverty of blacks relative to mulattoes, many black children Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 were probably undernourished from conception. Whether undernutrition in utero and infancy leads to permanent stunting remains unresolved, but studies indicate that “early disturbance lasts a lifetime.” Individuals experiencing pronounced dietary deªciencies or other environmental insults between birth and four years are usually shorter adults. Blacks clearly lived in austere and unforgiving environments that stunted their growth in infancy and adolescence. The adolescent growth spurt also demands signiªcantly increased nutritional intake. Nicholas and Steckel note that even a single-year delay in the onset of the adolescent growth spurt may imply some measure of early childhood malnu- trition, retarded food intake, increased work loads at younger ages, or a host of other environmental insults during critical growth periods. Again, the evidence shows that poverty afºicted blacks more than light-complected mulattoes. The adolescent growth spurt among blackchildren began later and was shorter in duration than among mulattoes.21 The growth patterns implied by the percentiles in ªgures 1 and 2 show a signiªcant growth recovery in adolescence, a pattern reminiscent of that found by Steckel for U.S. and slaves. Male slaves typically failed to achieve even the ªfth centile by their seventeenth birthday, but they experienced a short burst of recovery growth in their late teens, achieving the twenty- seventh centile, on average, in adulthood. Female slaves were also pathologically short up to their ªfteenth birthday, after which they experienced a brief period of recovery growth.22

21 A. Theodore Steegman, Jr., “18th Century British Military Stature: Growth Cessation, Selective Recruiting, Secular Trends, Nutrition at Birth, Cold and Occupation,” in Komlos and Cuff (eds.), Classics in Anthropometric History, 168; Stephen Nicholas and Steckel, “Heights and Living Standards of English Workers during the Early Years of Industrialization, 1770– 1815,” in ibid., 187. 22 Steckel, “A Peculiar Population: The Nutrition, Health, and Mortality of American Slaves from Childhood to Maturity,” Journal of Economic History, XLVI (1986), 721–741; idem, 38 | HOWARD BODENHORN The growth patterns found herein accord with the believed poor nutritional and health status of enslaved children. In the reg- istrant sample, nearly 42 percent of darkmale registrants between one and ten years were born into slavery; 60 percent of darkfe - male registrants between one and ten were born enslaved. Yet, only 15 percent of light males were born enslaved and about 17 percent of light females. The observed recovery growth in pre- Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 adolescence may reºect improving conditions for children after emancipation, but without details on the interval between manu- mission and measurement, such a conclusion is unwarranted. The observed pattern of recovery growth may be attributable to a declining percentage of manumitted children at later ages. The percentage of manumitted individuals declines sharply after age ten. Although 60 percent of darkgirls aged one through ten were manumitted, only about 14 percent of girls registering be- tween ages eleven and seventeen were ex-slaves. Among light girls, only 9 percent of the older group were born into slavery. Similarly, the percentage of manumitted darkmales declines after age ten; twenty-ªve percent of ªfteen- through twenty-year-olds were manumitted. Less than 9 percent of light males aged ªfteen through twenty years were manumitted. The registration data is consistent with Steckel’s ªnding that slave children were malnour- ished, particularly during infancy and youth. Blackand mulatto children and youth in antebellum Virginia were undernourished and grew under innumerable health hazards compared to modern Americans, but the question of whether they suffered inordinately compared to contemporary whites is still un- answered. No deªnitive study of growth of antebellum white children seems to have been done, but Komlos shows that white youths admitted to West Point were short and thin compared to modern populations. The average sixteen- to eighteen-year-old white cadet was just 0.5 inches taller than a light mulatto; the typi- cal nineteen- to twenty-one-year-old light mulatto was about 0.5 inches taller than white cadets. A late adolescent growth spurt may have left light-complected African-Americans temporarily behind contemporary whites, but they attained similar terminal heights. Moreover, the pattern displayed by all four groups in ªgures 1

“Growth Depression and Recovery: The Remarkable Case of American Slaves,” Annals of Human Biology, XIV (1987), 111–132. THE MULATTO ADVANTAGE | 39 and 2 are consistent with the centile-at-age pattern displayed by modern populations of African descent living in the New World. Urban African-American youths in 1970s Philadelphia and Ja- maica follow a similar pattern (a double-humped path with rela- tively low centiles in infancy and declining centiles just prior to the adolescent growth spurt), though at higher centiles at every age. This pattern is consistent with ªndings reported in Garn and Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 Clark, as well as Schutte, that, despite attaining similar terminal adult heights, modern African-Americans follow different growth paths than whites.23 Only additional evidence will reveal whether light-com- plected mulatto children suffered relative to contemporary whites, but the anthropometric evidence concerning Virginia’s African- American children demonstrates that blackchildren received sub - standard care as infants and adolescents, even by contemporary standards. Blackparents, whether free or in slave quarters, pro- vided their progeny with few nutritional advantages. Blackhouse- holds had fewer resources and inferior health services, and black mothers may have worked until later in pregnancy, translating into a compromised nutritional legacy for their children that per- sisted into adulthood. adult height and complexion Table 3 describes the charac- teristics of the sample of more than 7,300 adult African-Americans registered in the twenty-three sample counties. The distributions by region of birth, birth cohort, and complexion are similar to those of children reported in Table 1. Men and women were about equally likely to register. The Piedmont is overrepresented, particularly for manumitted slaves, where nearly three-quarters of all registering former slaves resided. Among free-born men, the proportions of black, dark mulatto, and bright mulatto were ap- proximately equal, but those described as darkgenerally outnum - bered those described as light or bright by a ratio of about four to three. Among manumitted slaves, darkindividuals outnumbered light by about three to one. Free-born women were more likely

23 Komlos, “The Height and Weight of West Point Cadets: Dietary Change in Antebellum America,” Journal of Economic History, XLVII (1987), 897–927; Stanley M. Garn and Diane C. Clark, “Problems in the Nutritional Assessment of Black Individuals,” American Journal of Pub- lic Health, LXVI (1976), 262–267; James E. Schutte, “Growth Differences between Lower and Middle Income BlackAdolescents,” Human Biology, LII (1980), 193–204. 40 | HOWARD BODENHORN

Table 3 Characteristics of the Virginia Free African-American Adult Sample—Males Twenty-Two to Fifty Years, Females Nineteen to Fifty Years free-born manumitted free-born manumitted males (%) males (%) females (%) females (%) birth cohort

1760s 0.9% 7.2% 0.6% 4.1% Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 1770s 2.4 12.7 1.8 10.6 1780s 5.5 13.4 5.2 15.1 1790s 11.4 17.2 8.3 16.0 1800s 23.5 17.5 20.6 20.0 1810s 23.4 20.0 23.3 20.4 1820s 22.9 9.7 25.0 9.6 1830s 10.0 2.3 13.8 4.2 1840s — — 1.4 — region of registration Tidewater 32.5 11.5 35.4 16.8 Piedmont 54.5 70.9 53.7 73.0 Blue Ridge 13.0 17.6 10.9 10.2 complexion Black24.1 41.4† 15.3 27.6† DarkMulatto 20.8 23.0 19.8 25.4† Mulatto 12.8† 8.5 11.9† 8.4 Brown 12.6† 6.2 16.7† 10.5 Bright Mulatto 21.2† 15.6 27.0† 22.0 Yellow 8.5† 5.3 9.2† 6.1 N 2,763 988 2,923 927 notes For variable deªnitions, see text and Table 1; a dagger (†) signiªes rejection of the null hypothesis of equal population proportions by gender; z-values for all signiªcant propor- tion differences exceed 3.0 (5% two-tailed critical value of 1.96). sources See Table 1. than men to be described as bright or light; the numbers of dark and bright were divided almost equally; and among manumitted females, darkindividuals outnumbered light by about ªve to three. Manumitted slaves, both male and female, were darker than free-born African-Americans. This ªnding runs counter to the notion that slaveowners tended to manumit the mixed-race off- spring of their illicit, exploitative, and relatively common relation- ships with female slaves. The evidence in the complexion panel of Table 3 suggests otherwise. Free-born African-American were signiªcantly lighter than those born enslaved. Indeed, nearly 65 THE MULATTO ADVANTAGE | 41 percent of all manumitted men were recorded as darkor black. Similarly, more than one-half of all manumitted women were considered darkor black.Statistical tests also show signiªcant dif - ferences in population proportions. The free-born were signi- ªcantly lighter than manumitted individuals. Although not deªnitive evidence, the complexion proportions suggest that free- born African-Americans were more likely to engage in miscege- Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 nation and produce mixed-race progeny than slaves were. This evidence suggests, though it does not prove, that mixed-race Afri- can-Americans were more likely to have resulted from consensual relations between free individuals than from forced, nonconsen- sual relations in the slave quarters. Indeed, absent extensive con- sensual interracial relations, Virginians would not have seen ªt to legislate against them and threaten to punish them severely. Although the registrations indicate that blacks were darker than darkmulattoes, who were darkerthan mulattoes, who were darker than bright mulattoes, who were darker than those labeled as yellow, court clerks did not use a common ªne-scale gauge of complexion. Sometimes the modiªer “very” was added to the blackand bright descriptors, implying darkerand lighter shades of blackand bright, respectively. Similarly, several descriptions of yellow individuals included such descriptors as “almost white” or “three-quarters white,” whereas such terms were only rarely at- tributed to bright mulattoes. It is impossible to know whether dif- ferent clerks would have assigned the same complexion desig- nation to the same person, just as it is impossible to determine whether those considered brown were lighter or darker than mu- lattoes. An informed conjecture might warrant that brown indi- viduals were darker and had fewer Caucasian features than mulattoes. The brown description was often used to describe indi- viduals also identiªed as “Negro,” as in “a brown Negro,” or “a Negro of reddish-brown hue”; the terms Negro and mulatto were rarely used together in the same registration. Ordinary least squares regression results reported in Table 4 demonstrate two notable features of the adult population. First, African-Americans registering in Blue Ridge counties were gen- erally taller than individuals registering in the Piedmont and, among men, individuals registering in Tidewater or coastal coun- ties were signiªcantly shorter than Piedmont men. Second, and consistent with the evidence on African-American youths in the Table 4 Factors Inºuencing the Terminal Height of Virginia’s African- Americans free-born manumitted free-born manumitted males males females females constant 67.26*** 68.90*** 62.29*** 61.47*** (303.28) (105.39) (364.72) (135.41) region Coast Ϫ0.39 1.39 Ϫ0.88 Ϫ0.14 *** *** Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 (3.18) (4.48) (0.82) (0.57) Blue Ridge 0.51*** 0.51** 0.64*** 0.32 (3.05) (2.16) (3.96) (1.16) complexion Black0.09 Ϫ0.52 Ϫ0.13 0.25 (0.46) (1.36) (0.76) (0.82) Dark Ϫ0.01 Ϫ0.43 0.20 0.31 (0.07) (1.06) (1.24) (1.02) Mulatto 0.36* 0.79* 0.20 0.70* (1.63) (1.70) (1.05) (1.85) Ϫ Bright 0.58*** 0.12 0.52*** 0.68** (3.00) (0.28) (3.41) (2.19) Ϫ Yellow 0.52** 0.10 0.47*** 0.48 (2.12) (0.20) (2.34) (1.15) birth cohort Ϫ 1760s 0.71 0.84 1.24** 0.09 (1.18) (1.25) (1.97) (0.15) Ϫ 1770s 0.26 1.60** 0.54 0.95* (0.66) (2.25) (1.40) (1.94) Ϫ Ϫ 1780s 0.12 1.15* 0.60** 0.57 (0.41) (1.83) (2.41) (1.24) Ϫ Ϫ 1790s 0.04 1.11* 0.58** 1.23*** (0.16) (1.80) (2.70) (2.69) Ϫ 1800s 0.17 1.10* 0.37** 1.20*** (0.80) (1.79) (2.21) (2.70) Ϫ 1810s 0.09 1.22** 0.51*** 1.13** (0.45) (2.01) (3.10) (2.54) Ϫ 1820s 0.35* 0.87 0.42*** 1.23*** (1.70) (1.36) (2.59) (2.57) 1840s — — 0.44 — (1.03) Adj R2 0.01 0.05 0.02 0.02 F-stat 3.90*** 4.53*** 4.13*** 2.12*** *** implies signiªcance at 1% conªdence. ** at 5%. * at 10%. notes Absolute value of Student’s t-statistics in parentheses. Constant reºects average height of individuals born in the 1830s, registering in the Piedmont region, and described as brown, copper, chestnut, or “Negro” by the county clerk. All regressions estimated using or- dinary least squares. sources See Table 1. THE MULATTO ADVANTAGE | 43 previous section, lighter-complected African-Americans were typically taller than darker ones. Light-skinned people, in fact, av- eraged about one-half to three-quarters of an inch taller than black or darkpeople. Regional differences in height may reºect the conºuence of a number of contributing factors. Tidewater culture was reportedly

more racist than other regions in Virginia, and it became increas- Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 ingly so after slave rebellions erupted in 1800 (Gabriel’s rebellion) and 1831 (Nat Turner’s rebellion). Moreover, Russell argues that free African-Americans residing in the Tidewater region were more conªned to low-wage, low-status jobs than African-Ameri- cans elsewhere in the state. Similarly, Tidewater slavery is consid- ered to have been harsher, more demanding, and more demeaning than Piedmont or Blue Ridge slavery. More virulent racism and greater labor effort in the Tidewater may explain part of the ob- served differences in height, but neither is fully convincing alone or in concert.24 Steegman noted that people raised in colder climates are taller than people raised in warmer climates, a regularity observed in many populations. Regional differences in average annual temper- ature undoubtedly serve as a proxy for other causative factors and, in Virginia, the hot, humid Tidewater and cooler, less humid Blue Ridge represent distinct topographical, geological, and pathologi- cal environments. Because the sandy soils in the Tidewater were less productive, raising a given crop implied greater exertion and expense than in the Piedmont or Blue Ridge regions. Moreover, the hot, humid climate and marshy conditions in the Tidewater provided an amenable climate for mosquitoes and geohelminths (parasites). Tidewater residents labored under a greater incidence of malaria and hookworm, as well as a host of other communica- ble diseases that thrive in warm, wet climates, all of which took their toll on human growth.25 The greater incidence of infection and disease, combined with relatively low agricultural productivity, had synergistic effects

24 John H. Russell, The Free Negro in Virginia, 1619–1865 (Baltimore, 1913), 173. 25 Steegman, “18th Century British Military Stature,” 169. The 1860 manuscript mortality census reports greater mortality from remittent fever (often thought to be malaria) in the Tidewater region. Doctors working for the Rockefeller Sanitary Commission in the 1910s uncovered a greater incidence of hookworm in Tidewater counties. Relative regional inci- dences in the nineteenth century were probably similar. See Rockefeller Sanitary Commis- sion, Fifth Annual Report (Washington, D.C., 1915). 44 | HOWARD BODENHORN on nutrition, health, and, ultimately, height. Poor nutrition made people more susceptible to disease, which inhibited their ability to convert food into energy. Substandard diets and lower energy out- put decreased labor productivity, which led to lower incomes, which diminished nutrition, made people more susceptible to dis- ease, and created a vicious cycle of misfortune. After the demands of workand battling infection were satisªed, fewer nutrients were Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 available for growth among Tidewater African-Americans. Be- cause the infections were chronic and recurred throughout in- fancy, childhood, and youth, residents of the Tidewater attained lesser stature. The second notable feature of the regressions is the pro- nounced complexion effect, which corroborates the ªndings re- ported in the previous section. Among free-born men and free- born women, individuals recorded as bright mulatto and yellow were about one-half inch taller than individuals listed as brown (the excluded group). Free-born mulatto men, manumitted mu- latto men, and manumitted mulatto women were also signiªcantly taller than those designated as brown. Free-born mulatto men were one-third of an inch taller; manumitted mulatto men and women about three-quarters of an inch taller. Light-skinned people appear to have received better net nu- trition (fewer disease-exertion claims on a given diet or, perhaps, more nutrients for a given disease-exertion climate) than dark- skinned African-Americans. Because mulattoes tended to be wealthier than blacks, they had greater access to food, clothing, housing, and medical care, advantages that manifested themselves in greater stature. The signiªcantly greater stature among light-skinned slave women may well indicate a social custom among planters and plantresses. Planters reportedly preferred light-skinned slave girls to attend their wives and daughters in the main house. Life in a planter’s house entailed less strenuous labor, provided greater ac- cess to nutrients, and released slave girls from the virulent disease environment of the slave quarters. A more amenable environment resulted in greater stature. Light-complected males, however, were not brought into the household and received no discernible nutritional advantage on the plantation. As the fair-complected girl insinuated herself into the planter’s household, the family might gradually have recognized her humanity and released her THE MULATTO ADVANTAGE | 45 from servitude. If so, this scenario might explain why manumitted women were lighter, on average, than manumitted men (see Ta- ble 3). The advantages of light-skinned slave women relative to slave men may also betray the sinister side of slavery. Virginia planters may have treated light-complected women preferentially for eco- nomic or sexual reasons. As so-called “fancy” girls, fair-com- Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 plected African-American slave girls were taught to read, sew, cook, and converse about literature and current events. They were given all the desirable attributes of a mistress. An active mar- ket for African-American mistresses existed in Lower South cities, and Virginia and Kentucky were thought to be important sources of mulatto women. Though much discussed, Fogel and Engerman’s study shows that this trade was not large enough to account for the mulatto advantage. Had proªts in this trade been substantial, Virginia planters would have manumitted few light- complected slave women. They would have sold them South.26

The purpose of this article is not to explain the source of the mu- latto advantage in the rural Upper South, only to show its exis- tence. Although rural antebellum Virginia did not witness the emergence of an economically privileged, socially inºuential, po- litically connected mulatto elite like Charleston and New Orleans did, the anthropometric evidence shows a marked mulatto advan- tage. Rural antebellum Virginia’s culture was not monochro- matic. Important distinctions between blackand white were drawn and exploited. Undoubtedly, African heritage worked against an individuals’ ability to reach their innate potential, but a fairer complexion opened at least some doors closed to dark-hued men and women. Light-complected children were about the same height as darkchildren, but light adults were taller than darkones. DarkAf - rican-Americans failed to experience a long enough period of catch-up growth in late adolescence to overcome the nutritional deprivations and other environmental insults that they experi- enced as infants, children, and youths. Thus, the historiography that the Upper South had only two classes of people is mistaken. It

26 Williamson, New People, 23; Robert W. Fogel and Stanley L. Engerman, Time on the Cross:The Economics of American Negro Slavery (Boston, 1974), 2v. 46 | HOWARD BODENHORN should not be surprising that antebellum Virginians drew com- plexion-based distinctions with obvious anthropometric conse- quences. Although Upper South culture differed in many respects from northern and Lower South cultures, it was neither so isolated nor so isolationist that it developed a unique attitude toward mu- lattoes. Northerners recognized complexion differences and acted on them. So too did whites in the Lower South. That Upper Downloaded from http://direct.mit.edu/jinh/article-pdf/33/1/21/1695226/00221950260029002.pdf by guest on 25 September 2021 South whites did as well is hardly remarkable. What is remarkable is that historians have failed for so long to see it.27

27 James O. Horton, “Shades of Color: The Mulatto in Three Antebellum Northern Com- munities,” in idem (ed.), :Inside the African American Community (Washing- ton, D.C., 1993), 122–144; Horton and Horton, In Hope of Liberty.