A Weighty Issue: Diminished Net Nutrition Among the U.S. Working Class in the Nineteenth Century

Carson, Scott Alan

doi:10.1007/s13524-015-0384-3

Skip Nav Destination

Research Article| May 12 2015

A Weighty Issue: Diminished Net Nutrition Among the U.S. Working Class in the Nineteenth Century

Scott Alan Carson

University of Texas, Permian Basin, 4901 East University, Odessa, TX 79762, USA

University of Munich and CESifo, Shackstrasse 4, 80539 Munich, Germany

Carson_S@utpb.edu

Search for other works by this author on:

This Site

Google

Demography (2015) 52 (3): 945–966.

https://doi.org/10.1007/s13524-015-0384-3

Abstract

Much has been written about the modern obesity epidemic, and historical BMIs are low compared with their modern counterparts. However, interpreting BMI variation is difficult because BMIs increase when weight increases or when stature decreases, and the two have different implications for human health. An alternative measure for net current nutritional conditions is body weight. After controlling for height, I find that African American and white weights decreased throughout the late nineteenth and early twentieth centuries. Farmers had greater average weights than workers in other occupations. Individuals from the South had taller statures, greater BMIs, and heavier weights than workers in other U.S. regions, indicating that even though the South had higher disease rates in the nineteenth century, it had better net nutritional conditions.

Anthropometrics, Nineteenth century U.S. weights, Net nutrition, Health

Introduction

It is now well accepted that a population’s average stature measures the net cumulative difference between nutrition and calories consumed for work and for fending off disease (Case and Paxson 2008a, b; Deaton 2008). Similarly, the body mass index (BMI) is interpreted to represent the net current difference between the same variables (Fogel 1994). However, as the ratio of weight to height, BMI does not fully isolate net current nutrition: weight in the numerator represents net current nutrition, while height in the denominator reflects net cumulative conditions, and this complicates the interpretation of BMI variation (Fogel 1994; Komlos and Brabec 2010, 2011; Stauss and Thomas 1998:773).^¹ Moreover, recent studies have called into question the meaning of BMI decline in the late nineteenth and early twentieth century because U.S. BMIs may have increased at the same time that average statures decreased. Statures and BMIs are also sensitive to early-life conditions. For example, as a function of weight and height, individuals receiving insufficient nutrition in childhood may have shorter statures, smaller frames, and low basal metabolic needs in later life. If their nutrition improves as they get older without a corresponding increase in physical activity, their BMIs are more likely to be high because their frames have less surface area to distribute weight (Baum and Ruhm 2009; Herbert et al. 1993:1438; Ravelli et al. 1976). Moreover, because it does not distinguish from fat-free mass—such as muscle, sinew, and bone—or the distribution of fat, BMI remains only a coarse measure for obesity. Recent studies have also called into question the meaning of stature decline in the late nineteenth and early twentieth century because U.S. BMIs may have increased at the same time that average statures and weight declined (Carson 2009a; Costa and Steckel 1997:50–55; Komlos 1987; Komlos and Brabec 2010, 2011). An alternative measurement for net current nutritional conditions is body weight, after controlling for height. Total weight increased in the nineteenth century when net nutrition improved and more calories, proteins, and fats were available, and taller heights can accommodate greater weight; however, BMIs are inversely related with statures because taller statures allow weight to be distributed over greater physical dimensions (Carson 2009a; Herbert et al. 1993:1438).^² In the results that follow, weight is measured in observation period and by current residence, while stature is measured by birth year and by nativity. Because weight varies more than stature with present conditions, weight—after controlling for height—provides important insight into how net current nutrition varies over time, by ethnic group, and by socioeconomic status. This seeming paradox is resolved by examining weight and height variation in isolation.

Health and net nutrition were compromised during nineteenth century U.S. industrialization and urbanization when food production was separated from food consumption (Carson 2008a; Floud et al. 2011:311–317; Haines 2004; Komlos 1985, 1987), and early food processing and canning techniques became common in late nineteenth century urban centers (Comer 2000:1314). Despite its importance relative to other physical measurements, weight has received little attention in historical health studies, mostly because of a lack of data. Two studies that addressed U.S. weights in the nineteenth century are Komlos (1987) and Coclanis and Komlos (1995). Using nineteenth century West Point cadet records, Komlos (1987) showed a general decline in net nutrition that was geographically widespread and affected blue-collar worker and farmer weights. Socioeconomic status and nutrition are related, and middle-class cadets were not as vulnerable to nutrition declines until the Civil War. Nutritional status may have also declined because meat and per capita animal protein production decreased when population growth, urbanization, and industrialization increased the demand for food. Reinforcing this trend, average weight for students at The Citadel decreased between 1880 and 1900 (Coclanis and Komlos 1995:104), which indicates that net current nutrition declined throughout the late nineteenth century. Studies of weight in the nineteenth century have otherwise gone without notice.

It is against this backdrop that this study considers three paths of inquiry into the relationship between historical U.S. weights, demographics, and socioeconomic characteristics during the early stages of economic growth. First, how did weight and height vary by race over the late nineteenth and early twentieth centuries? Weight and height for blacks and whites decreased throughout the century, and there is no sign that average weights or obesity increased among the working-class prior to the modern obesity epidemic. Second, how did weight vary with respect to occupations and socioeconomic status? Consistent with stature and BMI studies, farmers and unskilled workers had heavier weights and better net nutrition than white-collar and skilled workers. Third, what was the relationship between weight and residence? Diets and physical activity varied regionally, and after controlling for height, I find that individuals in the West and Deep South had greater body weight than individuals from the Middle Atlantic, Northeast, and Upper South.

U.S. Prisons and Weight Variation in the Nineteenth Century

Separating variation in weight and height over the late nineteenth and early twentieth centuries requires a reliable data set, and the two most common sources for historical weight and height data are military and prison records. One common shortfall of military records that may be related to weights is a stature requirement for military service, and arbitrarily truncating shorter statures may underestimate weights because shorter people are more likely to have heavier weights; however, only taller individuals with lower BMIs remain in military records (Carson 2009a; Herbert et al. 1993:1438). Fortunately, prison records do not suffer from such a constraint and the subsequent truncation bias observed in military samples. Nevertheless, because they may have selected many of the materially poorest individuals, prison records are not above reproach, and it is not clear which portion of the socioeconomic strata prisoners represent. For example, skilled and white-collar workers were imprisoned for fraud and embezzlement, which represents conditions among the upper class, while unskilled workers were imprisoned for assault and theft crimes, which represent conditions among the working class. It is also not clear whom the prison records represent because policing and law enforcement may have selected many of the materially poorest individuals who resorted to crime to survive. Alternatively, policing and law enforcement may have selected taller, more physically fit individuals involved in assault crimes who, because they were larger than other physical assault participants, were targeted for police arrest. Judicial action may have also varied with economic and social conditions. Because the majority of prisoners were incarcerated for assault and theft crimes, prison records represent conditions among the working class—the segment of society that is most sensitive to economic change. There is also little evidence of a consistent relationship between weight, height, and the type of crime for which an individual was incarcerated (Carson 2005:414; Carson 2007:44). Stature measurements in military and prison records may also be more common than weight measurements because height was easier to measure than weight, and accurate scales and weight measurement systems were developed and adapted later for both military and prison units.

The data used here are part of a large prison sample from the nineteenth century. Most blacks in the sample were imprisoned in Border States and the Deep South, in states such as Kentucky, Missouri, and Texas. Most whites in the sample were imprisoned in Pennsylvania and Texas, but whites were also in Missouri and the Far West (Table 1). Between 1840 and 1920, prison officials regularly recorded the dates inmates were received, pre-incarceration occupation, weight, height, nativity, complexion, and age. Physical descriptions were recorded by prison enumerators at the time of incarceration as a means of identification and therefore reflect pre-incarceration conditions. All records with complete age, weight, height, occupations, and nativity are used for this study. Because accurate recordings had legal implications for identification in case inmates escaped and were later recaptured, prison enumerators took care in recording inmate heights and weights. Arrests and prosecutions across states may represent various selection biases that affect the results of this analysis. However, variations in stature of blacks and whites across U.S. prisons are consistent with other historical height studies (Margo and Steckel 1982, 1992; Steckel 1979). Because the sample contains too few females and the focus of this study is U.S. male health conditions, females and immigrants are excluded from this analysis.^³

Enumerators recorded a diverse set of occupations but defined them narrowly, recording over 200 different occupations, which are classified here into four categories. First, laborers and miners are classified as unskilled workers. Unfortunately, prison enumerators did not always distinguish between farm and common laborers. Because common laborers probably encountered less favorable net nutritional conditions, this may overestimate the net nutritional benefits of being a common laborer and underestimate the advantages of being a farm laborer (Carson 2013b:62–63). Rounding out the other three categories, workers in the agricultural sector are classified as farmers; light manufacturing, craft workers, and carpenters are classified as skilled workers; and merchants and high-skilled workers are classified as white-collar workers.

Inmate enumerators were thorough when recording inmate complexions. For example, enumerators recorded inmates’ race in a complexion category and recorded white complexions as light, fair, medium, and dark. This white inmate complexion classification is also supported by European immigrant complexions that were always recorded with fair complexions as light, medium, and dark. African Americans were recorded as black, light-black, dark-black, and assorted shades of mulatto. Although mulatto inmates shared genetic traits with inmates of both European and African ancestry, they were treated as blacks in the nineteenth century. Thus, when comparing whites and blacks, I group mulattos with blacks (Carson 2009b).

Table 2 presents black and white inmates’ weight and height by age, observation period, residence, and occupations. Although average weights and heights are included, they are not reliable because of possible compositional effects, which are accounted for in the regression models in a later section. Blacks represented a smaller portion of the prison population than whites; 48 % of the U.S. prison population was black. However, blacks were overrepresented in the prison sample compared with the general population. Age percentages demonstrate that black inmates were incarcerated at younger ages, and whites were incarcerated at older ages (Hirshchi and Gottfredson 1983). During the early nineteenth century, slaves were less likely to be incarcerated because Southern planters lost money when their slaves were imprisoned (Wahl 1996, 1997). The result was that Southern law evolved to allow planters to punish their slaves who broke the law on their plantations, where slaves worked for their masters while paying the social costs of their crimes. Blacks were more likely to be incarcerated after the passage of the 13th Amendment, which abolished U.S. slavery and involuntary servitude. Whites were more likely to be from Northern and Far Western states; blacks were more likely to be from the South. Whites were more likely to be white-collar workers, skilled workers, or farmers. Blacks were more likely to be unskilled workers.

Changing Body Mass, Obesity, and Health

To demonstrate the magnitude of the variation in weight during the nineteenth century, BMIs illustrate how net current nutritional status may have varied throughout the late nineteenth and early twentieth centuries. The BMI coding system of the World Health Organization (WHO) provides thresholds for weight and obesity classifications: BMIs less than 18.5 indicate underweight; BMIs between 18.5 and 24.9 are normal; BMIs between 24.9 and 29.9 are overweight; and BMIs greater than 29.9 are obese.^⁴

Figure 1 demonstrates that the majority of nineteenth century U.S. black and white BMIs were in the normal category, and neither obesity nor starvation was common (Carson 2009a, 2012a). There was a sizable increase in BMIs with age, which is a standard part of the adolescent growth spurt (Bogin 2001:238–240). Average black youth and adult BMIs were 23.03 and 24.07, respectively; average white youth and adult BMIs were 21.96 and 22.78. Also noteworthy is that proportionally, few blacks and whites fell into the underweight and obese categories, indicating that nineteenth century black and white net nutritional conditions were not at privation (Fogel 1994, 2004:11). Costa and Steckel (1997:50–55) found that BMIs likely increased during the second half of the nineteenth century. However, Carson (2009a, 2012a) found that black and white BMIs decreased throughout the nineteenth century, which, like decreases in stature, indicates that net nutrition decreased with the separation of food production from food consumption.

During the late nineteenth and early twentieth centuries, few individuals were either underweight or obese. Height was inversely related with obesity because taller heights allow weight to be distributed over a greater surface area (Carson 2009a, 2012a). Compared with whites, blacks were more likely to be obese and less likely to be underweight because blacks have greater percentage muscle mass, and muscle is heavier than fat. Close proximity to nutrition in the nineteenth century decreased the relative price of nutrition for farmers, and farmer’s greater physical activity increased their BMIs and basal metabolic rates (Carson 2014). Moreover, Southern rural diets were associated with higher BMIs, and urban Northeastern diets were associated with lower BMIs (Carson 2009a, 2012a). In sum, black and white BMIs were in normal categories, and net dietary stress increased among the working class throughout the late nineteenth and early twentieth centuries.

Measuring weight and height separately illustrates how observable characteristics were related to changing weight versus changing stature. Moreover, weight studies address lingering questions in studies of stature and BMI: a decrease in weight, after accounting for height, during the nineteenth century indicates that net current nutrition decreased with economic development.

The distribution of weight in a population provides insight into its health. If weight is positively skewed, individuals in a population are underweight; if the weight distribution is negatively skewed, a population is overweight or obese. Figure 2 presents kernel density estimates for weight among blacks and whites and illustrates they were approximately symmetric. In addition, men in the late nineteenth and early twentieth century were neither starved nor living in caloric excess. Average black weight was 150.74 pounds, and average black height was 66.96 inches. Average white weight was 146.26 pounds, and average white height was 67.47 inches.^⁵ However, weights for blacks in the nineteenth century are not necessarily a sign of superior net nutrition because blacks were short and heavy, while whites were tall and thin (Carson 2009a, 2012a; Herbert et al. 1993; Margo and Steckel 1982). Average statures for blacks and whites can reach comparable adult terminal statures when brought to maturity under similar net nutritional conditions (Carson 2009b; Tanner 1977:341–342), but blacks likely have heavier weights because of greater bone mineral density and greater muscle mass (Wagner and Heyward 2000).

The Relative Effects of Race, Age, Residence, and Socioeconomic Conditions With Weight

The timing and extent of weight variation reflects the net current relationship between diets, disease, socioeconomic status, and residence. Because weight distributions are frequently skewed, a conditional median quantile model accounts for a possibly skewed weight distribution. The median quantile is

Weigh t_{i}^{.50} = Q_{y} (p| x) = θ^{.50} x + ε^{.50},

which is the median weight quantile, given x. The θ coefficient vector represents how independent variables were related to weight at the median. For example, the coefficient associated with height indicates that “average” weight changed by 1 pound when average height increased by 1 inch.

I test which of these variables were associated with weight variation in the late nineteenth and early twentieth century. To start, the weight of the ith individual is assumed to be a function of height, race, demographics, observation period, residence, and occupation.

\begin{array}{l} Weigh t_{i}^{.50} = α + β_{i}^{.50} Inche s_{i} + \sum_{r = 1}^{2} β_{r}^{.50} R a c e_{i} + \sum_{a = 1}^{14} β_{a}^{.50} A g e_{i} + \sum_{b = 1}^{8} β_{b}^{.50} Decade R e ceive d_{i} \\ + \sum_{p = 1}^{10} β_{p}^{.50} R e sidenc e_{i} + \sum_{j = 1}^{4} β_{j}^{.50} Occupatio n_{i} + ε_{i}^{.50} . \end{array}

A continuous height variable is included to account for the relationship between weight and height. Black and mulatto race dummy variables are included to measure the relationship between weight and skin complexion. Dummy variables are included for ages 14 to 22; dummy variables for adult age decade are included for ages 30 through 70. Dummy variables for decade received are included to account for how net current conditions varied by observation period. Residence dummy variables account for the relationship between weight and the physical environment. Occupation dummy variables are included to account for how weight was related to socioeconomic status.

To estimate the relationship between weight, height, and observable characteristics, Models 1 and 2 of Table 4 include all black and white observations. Models 3 and 4 present estimates for black weight and height, and Models 5 and 6 do the same for whites.

Three general patterns emerge in analyses of weight among blacks and whites in the nineteenth century. First, much has been written on the modern obesity epidemic (Cawley 2011; Flegal et al. 2009, 2010, 2013; Grossman and Macon 2011; Ogden et al. 2012); however, variation in weight over the nineteenth century is less clear. Like stature and BMI, weight among blacks and whites decreased throughout the late nineteenth and early twentieth centuries, and the decrease in weight was greater than the decrease in height.^⁶ Between 1840 and 1900, the average black stature decreased by less than 0.3 %, and the average black weight decreased by 5.0 % (Fig. 3). Over the same period, average white stature was constant, while average white weight decreased by 8.1 %, indicating that the decrease in weight relative to height over time was greater for whites than for blacks. Much of the decline was related to nutrition and the price of nutrition, and any understanding of weight variation in the nineteenth century must be made against the backdrop of nutritional adequacy. Throughout the late nineteenth and early twentieth centuries, net calorie allocations worsened for blacks and whites (Fig. 3; see also Floud et al. 2011:311–320; Komlos 1987:909; Putnam 2000), and much of the decline was related to industrialization (Carson 2008a:367–368; Komlos 1985, 1987). Technological change was associated with nutritional variation in the nineteenth century, and several agricultural innovations were related to available nutrition and physical activity (Cochrane 1979:189–202). The first two of these agricultural innovations in the late 1830s were Cyrus McCormack’s reaper and John Deere’s plow. Other nineteenth century agricultural innovations include harrowers, corn planters, disks, and hay-making equipment. However, because moving plows and heavy farm equipment was demanding on draft animals, mechanization was required for agriculture to fully develop. After 1865, the Civil War created an even greater need for labor-saving technology in U.S. agriculture, and many agricultural improvements came more from adopting existing labor-saving technologies than new agricultural innovations. Ransom and Sutch (1977:151–156) demonstrated that Southern agricultural production decreased throughout the nineteenth century, with accompanying increases in the relative price of nutrition. For whites, their disproportionate weight decrease compared with blacks also reflects that with the end of slavery, working-class whites were, for the first time, exposed to competition with black labor (Carson 2009c; Woodward 1951:134). The nineteenth century increase in stature was sustained, and statures increased mildly throughout the mid-twentieth century (Carson 2009a; Komlos 1987; Margo and Steckel 1982). In sum, throughout the late nineteenth and early twentieth centuries, weight decreased among both blacks and whites, with the decrease for whites exceeding that for blacks (Fig. 1; see also Carson 2009a, 2012a).

Second, as with height, weight for U.S. blacks and whites varied by occupational category, with farmers and unskilled workers having greater weights than workers in other occupations (Carson 2008a, b, 2012b). After controls for height are included, greater weight among farmers indicates that rural net nutrition was healthy compared with urban, industrialized net nutrition (Green 2012:88; Komlos 1987; Zehetmayer 2011). Black farmers’ average stature between 1840 and 1920 was 0.85 % taller than stature among all black workers, and the average weight for black farmers was 1.45 % greater than for all black workers. Stature and weight among white farmers was 0.92 % taller and 1.78 % greater, respectively, than among average white workers, indicating that the net nutritional advantage was greater for white farmers than for black farmers and workers in other occupations. The weight advantage for white farmers may have been partly due to their access to land and opportunity (Atack and Bateman 1987:93, table 6.3; Ransom and Sutch 1977:80–83). Greater weight among farmers and laborers was also partly related to rural diets, which gave farmers access to good nutrition before it was compromised when it was transported to market and food production was separated from consumption (Fletcher 1955:165, 195–202; Carson 2008a; Komlos 1985). For example, nineteenth century Middle Atlantic dairy production had greater nutritional value in rural locations because shipping milk to markets in steel containers led to spoilage (Carson 2008a:349–350; Fletcher 1955:165). In 1840, most U.S. agriculture was from owner-operated farms that primarily produced foodstuffs for household consumption. By 1900, U.S. agriculture was transformed into a highly organized industry that supplied food for urban communities (Atack and Bateman 1987; Fletcher 1955:165). Greater weight among farmers and laborers was also related to physical activity: occupations that require greater physical activity decrease fat and increased muscle, and for the same tissue volume, muscle is heavier than fat (Fogel 1997:448; World Health Organization 1985). By contrast, white-collar and skilled workers were typically urban residents and physically less active relative to the calories they consumed, and they did not put on additional weight.

Third, weight is related to residence, and diets and weights in the nineteenth century varied regionally. Abundant food supplies, rural environments, and diets high in animal protein were associated with heavier weights in the Deep South. For example, blacks from the Southwest were 1.4 % taller than average black stature, and weight among blacks in Tennessee was 1.6 % greater than the average black weight. Whites from the Southwest were 1.7 % taller than average for whites, and weight among whites in Tennessee was 2.1 % greater than the average weight for whites; these statistics indicate that net nutritional advantages were greater for Southern whites than for Southern blacks. Subsequently, a nineteenth century net nutrition premium was associated with Southern nativity and residence. Much of the regional weight variation was related to diet. Southern whites consumed more diverse and calorie-abundant diets than other U.S. whites, and white diets included pork, beef, corn, and Irish potatoes. Before the Civil War, the South was self-sufficient in food production and was a net food exporter (Ransom and Sutch 1977:150), and heavier weights in the South demonstrate that the benefits of close proximity to rural diets offset the calorie claims placed on diets by Southern disease climates (Crimmens and Condran 1983). Alternatively, the Upper South, Middle Atlantic, and Northeast were agriculturally less productive, more urbanized, and had diets that were less protein-based, with a greater share of calories coming from breads and other carbohydrates.

Northern food prices were the lowest in states that produced surpluses in select crops (Atack and Bateman 1987:228–246). For example, Illinois, Iowa, and Kansas produced wheat surpluses, and wheat prices were lowest in these states. The Middle Atlantic and Northeast were also more urban, which increased the relative price of nutrition (Komlos 1987). Shergold (1982:185–195) found that Northeastern diets were high in grains, breads, and dairy products (Floud et al. 2011:313; U.S. Census 1975:1175). Weights were heavier in the West than elsewhere within the United States, but statures were shorter, indicating that net cumulative health for Far Western settlers was substandard prior to migration compared with other U.S. regions; after native-born U.S. internal migrants arrived in the West, however, their diets and net nutrition improved, and they put on weight (Green 2012:88). America’s Far West had only recently been settled, promoting greater access to protein from domesticated animals and feral game. Because of recent settlement of the Far West, its residents may have also been more physically active and faced less-virulent disease environments than workers in other locations.

Other patterns are consistent with expectations. Height was positively related with weight but inversely related with BMI, indicating that although taller individuals had larger physical frames and dimensions to distribute weight, their taller statures allowed weight to be distributed over greater surface area (Table 3; Carson 2009a, 2012a). After controlling for height, I find that black and mulatto weights were, respectively, 5 % and 3 % greater than the average white weight. A considerable amount of work has demonstrated that mulattos were taller than darker-complexioned blacks; however, blacks had greater BMIs than mulattos, who, in turn, had greater BMIs than whites. Blacks have a greater percentage of muscle mass than whites, and muscle is heavier than fat (Evans et al. 2006; Flegal et al. 2009:507; Flegal et al. 2010:240).

Modern health studies also illustrate the relationship between weight change and age. For example, even among the most physically active modern males, weight gain increases with age and may occur because diets contain greater fat content or because physical activity decreases with age (Sherwood et al. 2000:398–401; Williams and Wood 2006:545). However, plausible explanations for weight gain among blacks and whites at older ages for both historical and modern populations relate to testosterone, human growth hormone, and metabolism, which are important hormonal weight regulators; each of these decreases with age (Rudman et al. 1990:1 and 5). With decreased hormones, metabolisms decline and central adiposity increases (Rosmund and Björntorp 1998). This historical weight-age finding is important because it illustrates that weight gain at older ages existed before the rise of the modern obesity epidemic and cannot be attributed to the same factors that are associated with the modern obesity epidemic. As a result, like height, weight in the nineteenth century was related with a complex set of relationships between observation period, socioeconomic status as measured by occupation, residence, and ethnicity. Further, height and weight measurements support each other to show that net nutritional conditions declined for blacks and whites with industrialization in the United States in the nineteenth century.

Explaining the Black and White Weight Difference in the Nineteenth Century

To more fully account for the source of the black-white weight difference, I construct a Blinder-Oaxaca decomposition for weight. A Blinder-Oaxaca decomposition is a statistical technique used to separate sample dependent variable differences that are due to average characteristics and returns to characteristics (Oaxaca 1973). If differences are due to average characteristics, differences in characteristics between the two populations account for differences between populations. However, if differences are due to returns to characteristics, market and institutional differences between populations account for the differences between the black and white subsamples.

Let W_b and W_w represent black and white weights, respectively; α_b and α_w are the autonomous weight components associated with black and white weights; and β_b and β_w are the black and white weight returns associated with specific weight-enhancing characteristics, such as age and occupation. X_b and X_w are matrices of characteristic of blacks and whites, and black weights are assumed to be the base structure. Weight models by race are constructed by regressing black and white weights on demographic, residential, socioeconomic status, and observation period:

\begin{array}{l} Black weight function : W_{b} = α_{b} + β_{b} X_{b} \\ White weight function : W_{w} = α_{w} + β_{w} X_{w} . \end{array}

The black and white weight gap is the difference between their average characteristics and returns to characteristics.

Δ W = W_{b} - W_{w} = α_{b} + β_{b} X_{b} - α_{w} - β_{w} X_{w} .

Adding and subtracting β_wX_b to the right side of the equation and collecting like terms leads to

Δ W = W_{b} - W_{w} = (α_{b} - α_{w}) + (β_{b} - β_{w}) X_{b} + β_{w} (X_{b} - X_{w}) .

The first right-side element, (α_b – α_w), is the part of the weight gap that is due to non-identifiable sources, such as the greater percentage muscle mass and greater bone mineral density that favored blacks (Wagner and Heyward 2000). Differences in the intercepts also exist because blacks and whites had different access to diets and disease environments. The second right-side element, (β_b – β_w)X_b, is the component of the weight gap that is due to returns to characteristics. The third right-side element, β_w(X_b – X_w), is the part of the weight gap that is due to differences in average characteristics; this element is unknown because whites probably had characteristics that were associated with greater weights, but blacks were shorter.

Using coefficients from the weight regressions (Tables 2 and 3), the weight decomposition indicates that the majority of heavier black weights were from non-identifiable characteristics, such as greater bone mineral density and lean muscle mass (Barondess et al. 1997; Flegal et al. 2010:240; Flegal et al. 2009:507; Evans et al. 2006; Wagner and Heyward 2000). However, the majority of the weight gap that was due to observable characteristics was associated with stature: whites in the nineteenth century had greater weight returns associated with height than blacks, indicating that white stature returns extended to weights. Whites also had greater weight returns than blacks for age, residence, and occupations. Other observable characteristics did not contribute to the black-white weight gap. In sum, the greatest share of the black-white weight gap was due to non-identifiable characteristics, such as differences in lean muscle mass and greater bone mineral density, and the white stature advantage also extended to white weights (Barondess et al. 1997:967–971).

Conclusion

When traditional methods to measure wealth are unavailable, the use of height and BMIs are now well-accepted measures of economic welfare. Even if wealth and other traditional measures for well-being, such as income and wealth, are available, weight and height add useful perspectives on human welfare. Stature measures a population’s net cumulative nutrition, whereas BMI is frequently interpreted as representing a population’s net current nutrition. However, as the ratio of weight to height, BMI differences over time and across residence are problematic because BMIs increase with weight and decrease with stature. Indeed, stature and weight in the nineteenth century varied both over time and by residence. As a result, weight may be a better measure for net current nutrition than BMIs because weight is not constructed as the ratio of net cumulative to net current nutrition variables and may be more sensitive to immediate conditions than BMI. Nevertheless, BMIs are important measures for obesity because they allow reasonable classifications for weight relative to height and obesity.

Like height and BMI, weight among blacks and whites decreased throughout the nineteenth century, and there is no sign that average weight among the working class increased prior to the modern obesity epidemic. Weight among both blacks and whites decreased throughout the late nineteenth and early twentieth centuries, and declines in weight relative to height were greater for whites than for blacks. Thus, net current nutrition declines may have been greater for whites than blacks because free white labor was integrated with the market, and with emancipation, lower-class whites had to compete with working-class blacks. Farmers and unskilled workers were taller and had greater BMIs than white-collar and skilled workers, and when height is controlled for, farmers and unskilled workers had greater average weight than workers in other occupations. Weight was also related to residence in the nineteenth century. Working-class men from the Northeast and Middle Atlantic had lower weights and shorter statures than individuals from other U.S. regions. Southerners had taller statures and greater weight than individuals with other nativities, indicating that close proximity to agricultural production and rural diets were associated with taller statures and heavier weights.

A long-standing debate centers on the net interaction of diets versus disease in the nutritional well-being of nineteenth century Americans. The regional weight comparison with heights and BMIs indicates that Southern rural net current and cumulative nutritional conditions had a greater influence on weight and height than disease: although the South had more-virulent disease environments, they had taller statures, greater BMIs, and heavier weights than elsewhere within the United States, indicating that net nutritional conditions were better in the South. Results reported here also have important implications for the modern obesity epidemic. Weights increased with age for historical populations, as they do for modern populations, indicating that male weights increase with age and do not depend on factors associated with the modern obesity epidemic.

Acknowledgments

I appreciate comments from John Komlos, Gary Taubes, Doug Henderson, Joe Beane, Paul Hodges, Meekam Okeke, and two anonymous reviewers.

Notes

B M I = \frac{Weight (kg)}{{(Height (m))}^{2}} .

Individuals who receive sufficient net nutrition during their youth reach taller statures, and their BMIs may be lower as they age because their frames have more surface area to distribute weight. In sum, average stature is an easier measure to interpret than BMI because stature measures the net cumulative difference between calories consumed and expended for work and for fending off disease but may not isolate how net current nutritional conditions vary.

All state prison repositories were contacted, and available records were acquired and entered into a master data set. These prison records include Arizona, California, Colorado, Idaho, Illinois, Kansas, Kentucky, Missouri, Montana, Nebraska, New Mexico, Ohio, Oregon, Pennsylvania, Texas, and Washington (Table 1). I examined female statures in prior studies (Carson 2011, 2013a).

⁴

Henderson (2005) demonstrated that these BMI thresholds may have shifted to the right between the nineteenth and early twentieth centuries.

⁵

Visual inspection reveals that there may have been greater weight rounding for blacks than for whites, which suggests that prison enumerators may have been less diligent when recording weight for black prisoners.

⁶

\begin{array}{l} B M I = \frac{w (K)}{h {(M)}^{2}} = w h^{- 2} . \Rightarrow ln B M I = ln w - 2 ln h . \\ ε_{B M I, w} = \frac{% Δ BMI}{% Δ w} = \frac{d ln B M I}{d ln w} = 1; ε_{B M I, h} = \frac{% Δ BMI}{% Δ h} = \frac{d ln B M I}{d ln h} = - 2 . \end{array}

References

Atack, J., & Bateman, F. (

1987

To their own soil: Agriculture in the antebellum North

Ames

Iowa State University Press

	Black		White
Location	N	%	N	%
Arizona	194	0.29	2,156	2.93
Colorado	483	0.71	3,502	4.76
Idaho	36	0.05	575	0.78
Kentucky	6,167	9.09	6,602	8.97
Missouri	4,292	6.33	7,984	10.85
New Mexico	344	0.51	1,993	2.71
Oregon	45	0.07	1,683	2.29
Pennsylvania	2,685	3.96	11,214	15.24
Philadelphia	5,481	8.08	11,410	15.51
Tennessee	20,940	30.88	10,384	14.11
Texas	27,154	40.04	16,083	21.85
Total	67,821	100.00	73,586	100.00

	Black				White
	N	%	Weight	Height	N	%	Weight	Height
Ages
Teens	14,044	20.71	140.08	65.99	10,035	13.64	137.75	66.77
20s	36,128	53.27	152.89	67.24	36,607	49.75	146.59	67.64
30s	11,074	16.33	154.95	67.30	16,191	22.00	148.55	67.58
40s	4,216	6.22	155.07	67.07	6,841	9.30	149.71	67.43
50s	1,678	2.47	154.44	66.87	2,841	3.86	149.90	67.34
60s	557	0.82	152.81	66.70	896	1.22	147.91	67.18
70s	124	0.18	146.89	66.18	175	0.24	148.31	66.86
Decade Received
1840s	20	0.03	163.6	69.21	165	0.22	158.97	69.04
1850s	55	0.08	155.07	67.35	839	1.14	149.13	68.22
1860s	980	1.44	150.10	66.37	1,307	1.78	148.95	67.75
1870s	7,615	11.23	152.61	66.94	8,748	11.89	144.36	67.37
1880s	12,508	18.44	151.99	67.26	10,888	14.80	146.32	67.47
1890s	14,285	21.06	150.99	66.93	14,114	19.18	147.51	67.56
1900s	16,319	24.06	149.52	66.76	17,782	24.16	145.68	67.23
1910s	15,090	22.25	149.87	67.01	18,533	25.19	146.23	67.61
1920s	949	1.40	150.34	66.86	1,210	1.64	147.05	67.62
Residence
Arizona	194	0.29	151.76	67.56	2,156	2.93	147.36	67.41
Colorado	483	0.71	154.43	67.15	3,502	4.76	150.26	67.40
Idaho	36	0.05	152.63	67.04	575	0.78	150.06	68.06
Kentucky	6,167	9.09	147.55	66.65	6,602	8.97	145.79	67.75
Missouri	4,292	6.33	145.63	66.72	7,984	10.85	142.66	67.50
New Mexico	344	0.51	154.46	67.55	1,993	2.71	150.27	67.89
Oregon	45	0.07	155.57	66.63	1,683	2.29	151.82	67.23
Pennsylvania	2,685	3.96	147.80	66.32	11,214	15.24	145.10	66.67
Philadelphia	5,481	8.08	146.42	66.22	11,410	15.51	140.83	66.57
Tennessee	20,940	30.88	150.32	66.56	10,384	14.11	148.36	67.61
Texas	27,154	40.04	153.62	67.58	16,083	21.86	149.30	68.42
Occupation
White collar	1,747	2.58	149.06	66.81	7,024	9.55	145.90	67.36
Skilled	5,147	7.59	151.03	66.97	16,395	22.28	145.96	67.85
Farmers	6,411	9.45	154.38	67.54	7,307	9.93	150.08	68.20
Unskilled	38,551	56.84	150.87	67.08	32,289	43.88	146.30	67.50
No occupation	15,965	23.54	149.04	66.45	10,571	14.37	144.17	67.25
Total	67,821	100.00			73,586	100.00

	Total		Black		White
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	Weight	Height	Weight	Height	Weight	Height
Intercept	–95.68**	44.84**	–85.27**	49.39**	–99.05**	45.67**
Height (inches)	3.57**		3.54**		3.59**
Race
White	Reference	Reference
Black	7.51**	–0.941**	Reference	Reference
Mulatto	5.15**	–0.672**	–2.28**	0.250**
Ages
14	–19.40**	–4.58**	–21.05**	–4.38**	–13.55**	–4.86**
15	–17.25**	–3.09**	–18.73**	–3.01**	–14.44**	–3.05**
16	–13.23**	–2.08**	–14.76**	–2.00**	–10.95**	–2.06**
17	–9.58**	–1.23**	–10.62**	–1.25**	–8.36**	–1.13**
18	–7.27**	–0.847**	–8.13**	–0.868**	–6.26**	–0.776**
19	–4.59**	–0.479**	–5.44**	–0.500**	–3.85**	–0.367**
20	–2.91**	–0.145**	–3.77**	–0.124**	–2.19**	–0.146**
21	–1.75**	–0.033	–2.25**	2.04^–13	–1.37**	–5.24^–5
22	–0.968**	–0.025	–0.886**	–0.091**	–1.15**	0.017
23–29	Reference	Reference	Reference	Reference	Reference	Reference
30s	1.12**	–0.055**	1.49**	0.008	.863**	–0.117**
40s	1.77**	–0.277**	1.46**	–0.250**	1.91**	–0.314**
50s	2.02**	–0.664**	1.80**	–0.657**	2.32**	–0.625**
60s	–0.215	–1.01**	–0.342	–0.932**	–0.042	–1.01**
70s	–1.31	–1.47**	–4.29**	–1.29**	1.41	–1.15**
Birth Decade
1800		1.05**		0.789^†		1.15**
1810		0.941**		0.632**		1.11**
1820		0.725**		0.289^†		0.954**
1830		0.232**		0.091		0.267**
1840		0.134**		0.091		0.176**
1850		0.052^†		0.091		0.008
1860		Reference		Reference		Reference
1870		–0.162**		–0.250**		–0.117**
1880		–0.357**		–0.376**		–0.341**
1890		–0.256**		–0.250**		–0.270**
1900		–0.177^†		–0.132		–0.305*
Decade Received
1840	10.52**		8.14**		12.56**
1850	4.54**		2.98		4.38**
1860	4.33**		3.43**		5.00**
1870	1.13**		2.14**		0.208
1880	–0.148		–0.228		–0.249
1890	0.319**		0.342		0.208
1900	Reference		Reference		Reference
1910	–0.558**		–1.17**		0.054
1920	–2.43**		–1.91**		–2.78**
Nativity
Northeast		0.090		0.308		0.067
Middle Atlantic		–0.142**		–0.059		–0.133**
Great Lakes		Reference		Reference		Reference
Plains		0.043		–0.065		0.149*
Southeast		0.313**		0.215**		0.473**
Southwest		1.17**		0.960**		1.15**
Far West		0.294**		0.336		0.422**
Occupation
White collar	–0.442	0.132**	–1.32**	0.124*	0.299	0.135*
Skilled	0.950**	0.082**	–0.329	0.159*	1.65**	0.053
Farmer	2.14**	0.493**	2.14**	0.500**	2.25**	0.500**
Unskilled	1.88**	0.223**	1.17**	0.250^†	2.55**	0.178**
No occupation	Reference	Reference	Reference	Reference	Reference	Reference
Residence
Arizona	0.848**		–2.33^†		1.53**
Colorado	3.04**		1.18		3.79**
Idaho	1.58*		0.810		1.98**
Kentucky	–2.94**		–3.41**		–2.25**
Missouri	–4.22**		–4.54**		–3.77**
New Mexico	1.72**		-1.05		3.01**
Oregon	6.43**		3.71^†*		7.52**
Pennsylvania	0.325		-2.39***		1.89**
Philadelphia	–2.34**		-3.52***		–0.997**
Tennessee	2.72**		1.94***		3.62**
Texas	Reference	Reference	Reference	Reference	Reference	Reference
Environment
Insolation		10.48**		7.78**		10.18**
Insolation, squared		–1.20**		–0.864**		–1.18**
N	141,407	141,407	67,821	67,821	73,586	73,586
R²	.2217	.0514	.2371	.0459	.1915	.0483

	$(β_{b} - β_{w}) {\bar{X}}_{b}$	$({\bar{X}}_{b} - {\bar{X}}_{w}) β_{w}$	$(β_{b} - β_{w}) {\bar{X}}_{w}$	$({\bar{X}}_{b} - {\bar{X}}_{w}) β_{b}$
Levels	Returns to Characteristics	Mean Characteristics	Returns to Characteristics	Mean Characteristics
Sum	7.71	–2.50	6.85	–1.64
Total		5.21		5.21
Proportion
Intercept	2.65		2.65
Height	–0.643	–0.352	–0.648	–0.348
Age	–0.115	–0.159	–0.066	–0.209
Decade	–0.007	–0.017	–0.017	–0.007
Residence	–0.211	0.037	–0.367	0.193
Occupation	–0.190	0.011	–0.233	0.054
Sum	1.48	–0.481	1.32	–0.316
Total		1.000		1.000

A Weighty Issue: Diminished Net Nutrition Among the U.S. Working Class in the Nineteenth Century

Abstract

Introduction

U.S. Prisons and Weight Variation in the Nineteenth Century

Changing Body Mass, Obesity, and Health

The Relative Effects of Race, Age, Residence, and Socioeconomic Conditions With Weight

Explaining the Black and White Weight Difference in the Nineteenth Century

Conclusion

Acknowledgments

Notes

References

Data & Figures

Contents

Supplements

References

Related

Citing articles via

Email alerts

Related Articles

Related Topics

Related Book Chapters

This Feature Is Available To Subscribers Only