Equality of Conditions: Simple Foraging Communities

According to existing archaeological material, humans lived in hunter-gatherer communities until the Mesolithic era. Relatively unsophisticated production technologies and strong norms of sharing, linked to the need to sustain risk-sharing strategies, led to interpersonal equality of food intake in simple foraging communities (Hawkes Reference Hawkes and Diehl2000). Population density was low (averaging 0.05 people per square mile; Keeley Reference Keeley1988), and communities were dispersed across vast spaces. Within those communities, the inability to store food without it rotting and the need to stay mobile in pursuit of herds undercut asset accumulation and the intergenerational transmission of resources (Smith et al. Reference Smith, Mulder, Bowles, Gurven, Hertz and Shenk2010). Stable political hierarchies were absent or minimal.Footnote ²

Rather than form vertical hierarchies to manage conflict, hunter-gatherer societies tend to split into smaller groups, some of which then move away to exploit other territories (Hirschman Reference Hirschman1981). Low-dispersion measures of height and health in foraging societies would constitute corroborating evidence of the equalizing properties of the hunter-gathering political economy.

The Breakdown of Equality: Complex Foraging Communities and Agrarian Societies

The relative economic and political equality of simple foraging societies broke down because two interrelated channels. The first one was strictly economic. The rate of technological innovation differed across individuals and communities, and even when the new production technologies were available to everyone, their application varied across geographical areas. For example, the invention of large boats and fishing equipment such as weirs and nets—resulting in higher productivity rates than simple foragers, higher population densities (around 3.5 persons per mile), some social stratification, and considerable stickiness in the intergenerational transmission of wealth—could only be exploited along a river or a maritime coast.Footnote ³ The domestication of plants and animals (jointly with a growing storage capacity) that began around 9000 BC produced even sharper patterns of territorial differentiation, economic specialization, and social stratification across the globe: According to the Ethnographic Atlas, among agricultural societies four out of five have inheritance rules on real property and five out of six have social classes (Boix n.d.). Smith et al. (Reference Smith, Mulder, Bowles, Gurven, Hertz and Shenk2010) report a Gini index around 0.4–0.5 for agrarian and pastoral societies versus less than 0.2 among hunter-gatherers for the 21 societies they examined.Footnote ⁴

A political channel increases inequality as well. As individuals and territories diverge in output and income, the potential for sustained interpersonal and intercommunity conflict rises, particularly when good land becomes scarce.Footnote ⁵ Those producers—fisher folk and farmers who have benefited the most from technological innovation—face the possibility that other, less well-off producers as well as “professional” bandits may systematically raid them and rob them of their crops.Footnote ⁶

To overcome this permanent threat of conflict (and the attending dissipation of growth), those communities of producers will seek to form state institutions capable of controlling violence and imposing order. On the one hand, they may attempt to defend themselves directly against any would-be predators by spending some fraction of their time setting up a defensive structure, making weapons to deter any potential plunderers, and fighting them. Producers double as rulers. On the other hand, in a scenario modeled by Olson (Reference Olson1993, Reference Olson2000), individuals with the incentive and capacity to loot the producers in the short run decide instead to protect them against other predators. In exchange for providing protection, these “stationary bandits” extract some permanent rents from producers. In other words, the Neolithic shift to food production and storage increased the transparency of output, making it easier for predators to identify and target producers (Mayshar, Moav, and Neeman Reference Mayshar, Moav and Neeman2011). Increased vulnerability to predation created the conditions for protection rackets of various kinds ranging from imperial kingdoms to the fragmented feudal structures of early medieval Europe.

Each of these institutional solutions has different distributional effects. When producers govern themselves, state institutions remain relatively horizontal, there are few transfers of resources across individuals through political mechanisms, and the distribution of income matches the underlying structure of the economy. Of course this does not mean that there is perfect equality of conditions. Differences in the fertility of soil will lead to income heterogeneity.Footnote ⁷ Economies of scale in certain crops such as sugar or cotton, for example, may result in the formation of a system of property that is highly concentrated in a few hands (Engerman and Sokoloff Reference Engerman and Sokoloff2002), something we explore in the section “The Impact of Factor Endowment and Production Regimes on Height.”Footnote ⁸ By contrast, monarchical regimes have clear tools and incentives for distributing resources to the military class, independently of the type of agricultural system in place. Rent extraction by monarchs and their allies—and therefore inequality—will increase with the power of rulers over their subjects.Footnote ⁹

Political outcomes depend on the available military technology and its impact on the power ratio between producers and Olsonian bandits.Footnote ¹⁰ Broadly speaking, the power ratio is more favorable to producers and self-defense is more feasible when military technologies are simple. As long as the production of violence is labor intensive and does not rely on sophisticated weaponry such as chariots, horses, and heavy armor, plunderers have little comparative advantage over producers and can subjugate very few peasants or farmers at a time to extract rents from them.Footnote ¹¹ In addition, producers have little incentive to subject themselves to a knightly elite because the opportunity costs of fighting relative to the production of goods and services are still low.

However, once military technologies become more complex, the comparative advantage of would-be predators in the production of violence tends to increase. For example, the domestication of the horse and the invention of flexible two-wheel chariots benefited the pastoralist peoples of the steppes more than the richer agricultural populations of the Near East and resulted in one of the most significant military and population expansions in history: Babylon fell under the attack of the Hurrites, the Aryans conquered Northern India, the Achaeans settled in Greece, and the Hyksos invaded Egypt (Anthony Reference Anthony2010). The introduction of the iron stirrup, which stabilized the horseman and allowed him to engage in mounted shock combat, may have contributed to the creation of a feudal class in Western Europe (Oman Reference Oman1936; White Reference White1962; for emphasis on cavalry tactics see Bachrach Reference Bachrach2001, 196) and in Byzantium (Haldon Reference Haldon1999; Kantorowicz Reference Kantorowicz and Coulborn1956). In addition, more complex technologies increase the opportunity costs of producers and as a result reduce their incentives to wage war directly. At some point, producers may prefer to pay some third party to defend them while devoting themselves fully to production, even if they retain some absolute advantage in war-making. However, because it is difficult for producers to control the “stationary bandit” they accept as a defender, higher opportunity costs generally result in the spread of monarchical or dictatorial institutions.Footnote ¹²

The Industrial Revolution and Democratization

Commerce, the rise of cities, and industrialization gave states new stores of wealth to tax while undermining the military monopoly of the landed elite. Industrialization accelerated from the eighteenth century, resulting in higher incomes overall and, eventually, a more equal distribution of wealth in most cases. In the short run, the higher returns on new production processes increased income disparities. However, as the higher returns in manufacturing industries spread to larger segments of the population and increased the returns to education, the wage gap between skilled and unskilled workers that had initially widened at the beginning of the industrial revolution narrowed again (Kuznets Reference Kuznets1955; Williamson Reference Williamson1991). This progressive equalization, at least in opportunities, facilitated a political shift toward mass democracy that in turn reinforced the trend toward equality as liberal institutions invested more in human capital formation and welfare states.

Income, Nutrition, and Height

Individual height is determined by both genetic dispositions and environmental factors (intrauterine conditions and the nutritional status and disease environment during childhood).Footnote ¹³ Fogel et al. (Reference Fogel, Engerman, Floud, Friedman, Margo and Sokoloff1983) show that the mean heights of boys aged 14 to 16 from families of London laborers between 1770 and 1865 were well correlated with the real wages of London artisans. Both Floud, Wachter, and Gregory (Reference Floud, Wachter and Gregory1990) and Steckel (Reference Steckel1995) provide extensive surveys from the twentieth century that confirm the same association. Based on a sample of the U.S. population with European ancestors in the 1970s and 1980s, Steckel (Reference Steckel1995: 1915) shows that income and height are positively correlated, with decreasing marginal returns to income. Individuals with an income of $12,000 (in 1985 US$) are about 10 cm higher than those with an income of $1,000. Similar results have been found for cross-country correlations (Baten and Blum Reference Baten and Blum2012). Cumulative evidence seems to warrant using heights as a proxy for economic development.Footnote ¹⁴

Likewise, variance in the heights of individuals within a given society may be used as a proxy for income distribution if we assume that access to food, shelter, and health increases with income and social position. That seems to be particularly true for preindustrial societies, which often lived on the Malthusian frontier. In their examination of the diets of rich and poor households in England from the fifteenth century to the middle of the eighteenth century, Drummond and Wilbraham (Reference Drummond and Wilbraham1991) find that wealthy individuals had diets rich in meat and fish. Poor households systematically relied on black bread, cheese, and pulses and occasionally on bacon. Fogel (Reference Fogel1994) has calculated calorie consumption by decile in France and England by 1790. The French and English top decile consumed 3,672 and 4,329 kcal per day, twice the median intake and three times the consumption of the lowest decile. The bottom 10% of the French labor force lacked the energy for regular work, and the next 10% had enough energy for less than three hours of light work daily. More directly related to this article's empirical strategy, several recent studies—Baten (Reference Baten2000) for Bavarian districts in 1798–1837; Sunder (Reference Sunder2003) for Norwegian regions in 1937–97; Baten and Fraunholz (Reference Baten and Fraunholz2004) for contemporary Latin America; Moradi and Baten (Reference Moradi and Baten2005) for contemporary sub-Saharan Africa; and Guntupalli and Baten (Reference Guntupalli and Baten2006) for India in 1915–44—have reported a statistically significant cross-country or cross-regional correlation between the coefficient of variation in individual heights and direct measures of income inequality.Footnote ¹⁵

The use of height variance as a proxy for income inequality faces three potential problems: reverse causality, selection, and genetic sorting. We discuss them here sequentially. First, height may result in higher earnings rather than the other way around. The positive effect of height has been attributed to its impact on self-esteem, social recognition, or discrimination (Hensley Reference Hensley1993; Magnusson, Rasmussen, and Gyllensten Reference Magnusson, Rasmussen and Gyllensten2006; Wilson Reference Wilson1968) or to its correlation with greater strength and better health (Strauss and Thomas Reference Strauss and Thomas1998) or with cognitive ability (Case and Paxson Reference Case and Paxson2008). Our results do not depend, however, on showing that individual heights and incomes are related causally in either direction. What matters for our purposes is that we observe higher variance in heights in those societies that have economic, military, and institutional factors that are more conducive to inequality. A correlation between height dispersion and these factors predicting inequality indicates that either a causal relationship exists or that, regardless of the direction of causality, these factors make inequality possible—that is, they allow the marginal effects of height to be positive. The role that contemporary wage-bargaining institutions play in wage formation helps clarify our point. Variance in skills and height distributions is similar across the most advanced economies. Within-country wage compression is positively associated with how centralized wage-setting institutions are (Wallerstein Reference Wallerstein1999), independently of whether particular wage-setting procedures generate equality or whether wage equality makes centralized bargaining institutions sustainable (Beramendi and Rueda Reference Beramendi and Rueda2012).

Height distributions are vulnerable to survivor bias because only survivors into adulthood are included in our samples. A disproportionate number of deaths of poor children, for example, might chop off the bottom end of the height distribution. We are not able to correct in any systematic way for that distortion, but note only that Gini measures of income inequality also measure survivors and therefore suffer from the same problem (see Moradi and Baten Reference Moradi and Baten2005; van Zanden et al. Reference van Zanden, Baten, Földvari and van Leeuwen2013). It is thus necessary to interpret any distribution of surviving populations with caution.

Finally, using height variance as a proxy for economic inequality could be suspect if heights are determined overwhelmingly by genetics and if genetically distinct groups in a society coincide perfectly with their economic status. This is not a great concern, at least for populations of similar ethnic background. For example, better nutrition and health care were behind the process of growth and regional convergence in heights in Norway where the genetic pool has remained unchanged in the last two centuries (Sunder Reference Sunder2003). Later in this article we present data on other populations that did not change genetically but whose average height and height variance changed with economic and political conditions: Japan, medieval and modern Poland (especially high-status males and peasants), and African Americans (before and after the Civil War). Over the course of history many populations of warriors have invaded and subjugated different ethnic groups. However, invasions and population movements pose an inference problem in the use of heights data only if neither genetic nor political-economic mixing occurs. Although this situation strikes us as improbable, we remain attentive to this possibility while examining our empirical evidence.Footnote ¹⁶

Evidence

Two types of evidence can help us gauge the height of human populations: (1) skeletal remains from human settlements as reported by existing archeological and anthropological research and (2) direct measurement of heights in living individuals, mostly from census data and military records. Data from skeletal remains provide a rare glimpse into the size of prehistoric populations, but suffer from two problems. First, archaeologists and anthropologists rarely find complete skeletons, let alone in the numbers necessary to draw statistically valid inferences about the populations under study. Second, methods to estimate heights from skeletal remains, mainly from the bones of the extremities, are in some cases specific to particular ethnic groups (having been derived from large statistical samples of living people). We have well-tested formulas for some groups (modern whites, blacks, Japanese, and Mesoamericans), but heights—particularly the variance in the distribution—are sensitive to which height estimation formulas are used (Krogman and Isçan Reference Krogman and İşcan1986). Full height measurements are accurate as far as they go, but are available only beginning in the late eighteenth century (and, exceptionally, in the early eighteenth century for France), and until censuses appeared, they were primarily for military recruits and prison populations.

In our analysis, we use both skeletal and measured height data but, to be conservative, we rely on Franz Boas's data collected for living Native Americans from various tribes in the nineteenth century that approximate the conditions of precontemporary or even premodern societies. We supplement the findings of those data with inferences derived from skeletally estimated height data.

Data on heights from different populations come to us in two formats: (1) complete datasets that report heights for all its individuals and from which we can estimate summary statistics of interest and calculate density distributions and (2) information only of height means and standard deviation of a given population (as a whole or for its different social strata). We discuss the representativeness of our samples when presenting the data. Note, however, that the question of sample representativeness is a less pressing concern in those cases for which we have separate measurements for each economic or social strata.

For the sake of comparison across samples, we focus on two types of measures: primarily the coefficient of variation of each society because it controls for the effect of differential height means and, when this is not feasible, the difference in heights across social strata. In the few cases where we have both measures, the coefficients of variation and the height gap between lower and upper classes are positively correlated. Where possible, we also present and discuss the kernel density distributions of different populations to assess the impact of political and economic factors on inequality.

The Evolution of Height

Figure 1 recapitulates succinctly what is known about the evolution of average human height over time. Both men and women were taller in pre-agrarian societies than in agrarian societies. In Paleolithic and Mesolithic sites, femur lengths seem to indicate a height around 175 cm for men and 165 cm for women. These values are in the range of average heights today. Pre-agrarian individuals were tall as a result of their abundant and diverse diet based on scores of plants and animal species, naturally related the use of birth control strategies and low human population densities (Bogin Reference Bogin2001; Gould Reference Gould1981; Hayden Reference Hayden1981; Hill and Hurtado Reference Hill and Magdalena Hurtado1989; Lee Reference Lee1984). By contrast, agriculturalists tended to rely on a single cereal stable—rice in Asia, wheat in temperate Asia and Europe, and maize in the Americas—supplemented with vegetables and, particularly among wealthy strata, some fish and meat (Scott and Duncan Reference Scott and Duncan2000).Footnote ¹⁷

FIGURE 1.

Heights in Western Europe and Eastern Mediterranean

The economic transformation spurred by the industrial revolution had two consequences. In the short run, it led to a decline in average heights as rural dwellers moved to densely crowded, unhealthy urban centers. Transitionally, the opening of markets permitting sales of food for other goods might also have raised the price of food, reducing consumption. In the long run, however, growing per capita incomes and the improvement of public sanitation resulted in better nutrition and health conditions and, eventually, increased heights to the genetic potential—until increased income had no additional effect (Komlos Reference Komlos1989; Steckel Reference Steckel1995).

The Zuni Pueblo

The Zunis, a sedentary people in northwestern New Mexico, represent a simple, relatively nonstratified agrarian political economy. Although Spain tried to conquer and evangelize them in the seventeenth century, an Indian uprising eventually expelled the Spanish from the region in the 1680s. Until the middle of the nineteenth century the Zunis had little contact with external societies and remained politically and economically self-sufficient. Systematic trade and political relationships with the United States only started in the 1880s. This led to a devastating smallpox epidemic in 1892–3 and a move toward livestock cultivation as their main source of livelihood at the turn of the century. In the 1880s the Zuni Pueblos were still governed by a “committee” filled with former warriors who had taken scalps from their enemies. This governing council regulated religious ceremonies, repressed witchcraft activities, coordinated responses to big crises, and resolved inter-town conflicts. All other social relations were left in the hands of families and clans (Ladd Reference Ladd and Ortiz1979).

Panel A of Table 2 displays the mean, standard deviation, and CV of male and female heights in the four periods for which we have data, coming from skeletal remains spanning from AD 600 to AD 1680 and the Boas dataset. The data suggest relative equality. The CV of male height, of 3 or less, is almost a third lower than its value in modern societies. Female heights CVs fluctuate around 3 except in one case.

TABLE 2.

Height Distribution in Agrarian Societies with Simple War Technologies

^a From Steckel and Rose (Reference Steckel and Rose2002), ch.16.

^b From Boas project.

^c Own estimates based on Steckel and Rose (Reference Steckel and Rose2002).

Mayan Cities

The characterization of precolonial Mayan states has been controversial. Traditionally, Mayan city-states were thought to be highly stratified, theocratic polities (Haviland Reference Haviland1967). More recently, however, several scholars have suggested that the Mayan lived in only mildly stratified cities, with a small priestly and warrior elite, in the middle of a very poor and equal rural economy. From the point of view of our inquiry, the most relevant variable is the kind of military and war-making technologies of Mayan societies. Although the Mayan civilization was not conflict free as some scholars have claimed, its weaponry was clearly underdeveloped and primarily based on chipped-stone spear and dart points (Aoyama Reference Aoyama2005; Webster Reference Webster2000). Mayans had no bronze metallurgy, no wheels, and no domesticated animals, such as the horse, that could be used to engage in the type of large-scale warfare conducted in the Middle East or China.

Table 2, Panel B reports data from a number of Mayan sites during its classical period: Individuals were rather short, even though their statures later dropped even more—by 2 and 5 cm under the Spaniards and by another 2 to 3 cm after Mexican independence. Height variance was quite low: The CV is 2 or less among men (the lowest in all our data) and 2.5 among females in our largest sample (in urban Copan). By area, individuals were shorter and the dispersion was minimal in rural areas—the skeletal remains come from areas with “modest dwellings of perishable materials dispersed in rural areas away from the urban core” and probably belonged to “agricultural laborers and part-time craftspeople” (Steckel and Rose Reference Steckel and Rose2002, 289). Urban settlements show more height dispersion, probably reflecting the mix of people—wealthy individuals and their servants and perhaps a class of craftsmen—who lived there. Still, Mayan cities were small. The population of Copan peaked at 27,000 in the ninth century and then fell to a third of that number by AD 1000. Compared with the large polities of Europe that emerged after the military revolution of the Renaissance period, stratification could not have been very extreme, and according to Table 2, it was not.

Mycenae and Ancient Egypt

Although agriculture took root in the Middle East around 9000 BC, the formation of states with a clear hierarchical political structure, a division between a military and religious elite and a mass of subordinate laborers only took place around the late fourth millennium, coinciding with the metal revolution and the use of copper to build short-range weapons, such as maces, axes, swords, helmets, and shields (Yadin Reference Yadin1963). A thousand years later, the introduction of bronze, which was both lighter and sturdier than copper, opened the way to the first big monarchical states in the Middle East (Keegan Reference Keegan2004; McNeil Reference McNeil1981).

We report data for Egypt under the New Kingdom and the Middle Bronze Mycenae as representative of ancient agrarian states and complex war-making machines. For Egypt we have direct measurements of mummies for two separate populations: 31 royal mummies (pharaohs and spouses) from the eighteenth to twentieth dynasties of the New Kingdom (1550–1137 BC), x-rayed in the early twentieth century (Smith Reference Smith1912), and 260 complete skeletons buried in the upper Nile Valley (Asiut, Gebelen, and Aswan) for the Dynastic period (Masali Reference Masali1972).Footnote ²⁵

The data are reported in Panel A of Table 3. Height differences were substantial in the highly stratified Egyptian polity. Even using a conservative estimate according to which royal mummies shrank 8 cm, pharaohs averaged 166 cm and were about 9 cm taller than common males. Women belonging to the royal family averaged 156 cm or about 8 cm taller than the average common female.Footnote ²⁶ Because we do not have data on the shares of aristocratic and nonaristocratic population and given that Masali does not report the standard deviation for commoners, we cannot estimate the true CV for all the Egyptian population. Even assuming, however, a standard deviation of 6 for commoners (a number on the upper range of all the data we have), the median pharaoh would be in the 95th percentile of the commoners’ population.

TABLE 3.

Height Distribution in Agrarian Societies with Complex War Technologies

^a Smith (Reference Smith1912).

^b Masali (Reference Masali1972).

^c Angel (Reference Angel, Cohen and Armelagos1984).

^d Komlos (Reference Komlos1989).

Mycenae, which had a powerful militarized monarchy with a Homeric aristocracy in the Middle Bronze period (around 1800–1700 BC), reveals a similar pattern. Differences in height were remarkable. Whereas the estimated stature of kings and males buried in kings’ tombs was 172.5 cm, commoners averaged 166.1 cm or 6.4 cm less (Angel Reference Angel, Cohen and Armelagos1984). Among women, the upper strata averaged 160.1 cm and the commoners 153.5 (Table 3, Section B). As Angel (Reference Angel, Cohen and Armelagos1984) points out, these differences cannot be attributed to ethnic origin because the royalty “show[s] the same striking diversity of morphology (and implied origins) as the general population, then in the process of absorbing Indo-Europeans and other settlers, including some from Africa.” Instead, “the 4 percent increase in stature and in pelvic depth and the 30 percent increase in skull base height in the royalty, their thicker and more rounded long bones, and the five-fold improvement in their dental health all show nutritional improvement that must involve more meat protein than the average citizen got” (Angel Reference Angel, Cohen and Armelagos1984, 66).

Medieval and Modern Europe

Medieval European societies offer us a picture of preindustrial societies characterized by a rural economy (about 90% of the population lived in the countryside) and feudal institutions with a thin rural nobility, an overwhelming population of farmers and laborers, and the clergy itself reproducing the hierarchically constructed structure of nobles and peasants (Duby Reference Duby1968). With the expansion of commerce and the invention of the cannon, Europe moved from the feudal order to a system of sovereign states with strong royal dynasties that often co-opted the old nobility within absolutist institutions. In the West, feudalism gave way to a more modern agrarian system, based on the recognition of the Roman concept of absolute property rights, money-based transactions, and the emergence of a free class of farmers. In Eastern Europe, despite or perhaps because of greater land/labor ratios, the nobility successfully engineered a second feudal revolution and reestablished serfdom in the sixteenth and seventeenth centuries (Anderson Reference Anderson1974; Domar Reference Domar1970).

Extensive data from Poland—gathered from 53 cemeteries, encompassing 3,000 individuals, and spanning the tenth to the nineteenth centuries—can help us assess the evolution of inequality over time in feudal countries with some detail. Figures 4A and 4B show the evolution of male and female heights in Poland from the central centuries of the Middle Ages (tenth to fourteenth centuries) to the Baroque period (mid-seventeenth century). For each period they distinguish between high-status, rural, urban, and Jewish cemeteries. Until the population crisis of the fourteenth century, we observe some systematic but small differences between social segments. High-status males and urban dwellers men were 3 cm taller than rural populations. Among women there were no significant differences; in fact, and against our expectations, rural women were 1.5 cm taller than high-status females up until the fourteenth century. After the Black Plague and especially the reintroduction of serfdom in the sixteenth century, height differences across classes increased dramatically. High-status men were 5 cm taller and high-status women were 2 cm taller than peasant folk. The difference with Jewish populations is even larger—the latter were 7 and 5 cm shorter, respectively—probably reflecting their extreme poverty (even though we cannot exclude genetically determined differences at work there). Differences in height due to income and nutrition were essentially interclass. Rural communities, for which we have individualized data, appeared to be relatively homogeneous: The CV of male heights in the cemetery of Ostrow Lednicki (fourteenth to the seventeenth centuries) is 3.3.

FIGURE 4A.

Male Heights in Poland, 11th to 18th Centuries

FIGURE 4B.

Female Heights in Poland, 11th to 18th Centuries

Late-eighteenth-century Germany seems to tell a similar story (Table 3, Panel C). Based on conscription data, Komlos reports that from the 1760s to the 1790s German aristocrats at age 17 averaged 164.3 cm, compared to 159.9 for middle-class recruits and 151.1 for German peasants. Although this gap probably closed as the peasants experienced some delayed growth (Komlos Reference Komlos1989: 94), research shows that stunting during adolescence is never fully overcome afterward (Case and Paxson Reference Case and Paxson2008).Footnote ²⁷

Figures 5A and 5B summarize our discussion by comparing the kernel density distributions of male and female heights in two simple-warfare societies (the Zunis and the Maya) and two complex-warfare cases (Egypt and Poland).Footnote ²⁸ Particularly for males, the variance in heights is much higher in the latter two and seems to corroborate our theory about the differential impact of politics and warfare.

FIGURE 5A.

Distribution of Male Heights in Agrarian Economies

FIGURE 5B.

Distribution of Female Heights in Agrarian Economies

The differences in height inequality match the direct—although very scarce—data we have on income inequality in both feudal and absolutist or despotic monarchies. Goldsmith (Reference Goldsmith1987) reports that, at the death of Augustus (14 AD), the top 1/10,000 households of the Roman Empire received 1% percent of all income. (For comparison, the British royal family received 0.2% of all income in 1979–80.) In Mughal India around 1600 AD, they received 5% of all income. The annual income of the Indian emperor was the equivalent of the wage of about 650,000 unskilled workers. Milanovic, Lindert, and Williamson (Reference Milanovic, Lindert and Williamson2011) examine the level of inequality in 28 preindustrial societies and, by comparing it with the theoretically maximum feasible inequality, compute the actual level of rent extraction by the elite. According to their estimations, income transfers to elites were extremely high in monarchical societies: Elites captured more than two-thirds of all the resources available once we exclude the total sum of the minimum subsistence wage for all the population. By contrast, the level of “rent extraction” is estimated to be from one-third to two-fifths in today's advanced economies.

Japan

Table 4 displays height data for people living on the Japanese islands from 850 BC to the nineteenth century, calculated from bone measurements, using the Fujii method of calculation that is derived from Japanese populations.Footnote ²⁹ The early hunter-gatherer population (the Jomon people, who lived in Japan from at least 4000 BC and probably earlier) were more evenly sized (with a male CV of 2.70) than the agricultural Yayoi who supplanted them in the fourth century BC (with a male CV of 3.56; Matsushita Reference Matsushita1994; Yamada Reference Yamada2003). Because some of the Yayoi may have been of different genetic stock, one may take this as suggestive but not conclusive evidence for the effects of sedentary agriculture on intra-male inequality (Yamada Reference Yamada1997).

TABLE 4.

Height in Japan

Note: Formula employed to calculate height: Fujii method.

Sources: See text.

Intra-male height dispersion appears to have grown in later centuries as agriculture became more labor intensive and as society became disrupted by warfare, although our data do not allow us to weight these factors. The Kofun period (third century BC to the seventh century AD) is named after the grave tumuli (kofun) in which the upper classes were buried. For that era we have only height data estimated from people buried in these mounds, without commoners as a baseline for comparison. Yet it is noteworthy, given that this is a high-status group, that the Kofun people are the tallest of the Japanese sample and have the lowest intra-male dispersion after the Jomon.

Commoner and low-ranking warrior families from the war-wracked medieval period (thirteenth to sixteenth centuries) had greater height dispersion than in most other periods in Japanese history, although not as great as in medieval Europe (Hiramoto Reference Hiramoto1972, 222; Nakahashi and Nagai Reference Nakahashi and Nagai1985). Unlike the stable and stratified Tokugawa social order that followed it, medieval Japan was plagued by endemic civil war, political upheaval, and upward social mobility whereby many rural families became part of the warrior class.

We expected more evidence of nutritional variation between the Tokugawa shogun and Edo commoners (seventeenth to nineteenth centuries), but find that the shogun for whom we have available data were surprisingly short. Three of the four shogun for whom there are height data—Tokugawa Iyenobu (1662–1712; 162 cm), Iyeshige (1711–61; 153 cm), Iyeyoshi (1793–1853; 153 cm), and Iyemochi (1846–66; 154 cm)—were shorter than the average height of 157 cm for commoners in the Edo period. Either commoners were adequately nourished, or as some scholars have suggested, the Buddhist strictures of the Tokugawa family prevented the shoguns from consuming sufficient quantities of protein (Matsushita Reference Matsushita1994; Suzuki Reference Suzuki1985). It is also likely that the shoguns strengthened their genetic “tether” by assortive mating.

Midwest Farming Economies

Table 5A presents average heights for Ohio guardsmen by their occupations and decade of enrollment—estimated using the data gathered by Steckel and Haurin (Reference Steckel, Haurin and Komlos1994). Although some cohort CVs are high, what is remarkable is that height differences across occupations were low: The largest gap in average heights was 2.2 cm. The small occupational differences observed in the heights of the Ohio National Guard are in line with results for American data in other time periods. Union troops in the Civil War who were farmers were only 1.4 cm taller than laborers.

TABLE 5.

Production Regimes And Height Inequality in Mid-Nineteenth Century United States

Sources: Ohio: Steckel and Haurin (Reference Steckel, Haurin and Komlos1994). Panels B & C: Steckle and Rose (2002).

Table 5B reports height statistics for two North American communities in the Great Lakes in the 1840s: the middle-class city of Belleville and the skeletal remains of the cemetery of the poorhouse of Highland Park (taken from Steckel and Rose Reference Steckel and Rose2002). Differences in mean heights are very small—a maximum of 1.7 cm for men. The CV of the two populations together is 3.36 for men and 3.64 for women, which are lower than the U.S. CV in 1977.

Slave Economies

Table 5C reports height statistics for white elites and black slaves in the antebellum American South and for free African Americans after the Civil War (also taken from Steckel and Rose Reference Steckel and Rose2002). Male slaves were 4 cm shorter than southern white men (and 6 cm shorter than Ohio national guards). Enslaved women were 6.5 cm shorter than white women. This wider gap in heights for females was likely related to investment decisions made by their owners as a function of their specialization.Footnote ³⁰ Male free blacks (measured in 1888) caught up very quickly with whites. This catch-up did not occur, however, among black women.

Figures 6A and 6B represent the kernel density distributions of the male and female populations of the Midwest (Belleville and the poor house together and, for males only, Ohio National Guard individuals) and of the American South (whites and slaves before 1860 and whites and free blacks after 1880). Midwest distributions approach the pattern of a normal distribution (with a higher mean for Ohio men). The distribution of South Carolina men is clearly bimodal in the antebellum era, but normalizes after the war.Footnote ³¹

FIGURE 6A.

Distribution of Male Heights in 19th-Century United States

FIGURE 6B.

Distribution of Female Heights in 19th-Century United States