The human population of a nation is the stock of resident persons, at any given time, of various ages, sexes, marital statuses, races, ethnic origins, religions, residences, and socioeconomic statuses (such as occupation, income, and wealth). These persons are spread across the area of the nation in distinctive patterns, living in central cities, suburban areas, or rural regions. The study of population is crucial because it is important to know such things as how legislative bodies are apportioned; what the stock of actual and potential workers in the labor force with various skills and education levels in different places is; what the base of income, property, sales, and excise taxpayers is; what the stock of citizens potentially eligible for military service is; how many children will need places in public and private schools; how many older citizens will be collecting retirement benefits and will be in need of medical care both now and in the future; and where the population is located (central cities, suburbs, or rural areas) so demand for public services (roads, schools, post offices, water and sewer systems, electricity, and telephone and cable services) can be predicted. The human resources of a nation are as important as anything it possesses. This essay describes the population of the United States since the date of the first federal census in 1790: the components of its growth; composition by age, sex, race, and ethnic group; and location (rural-urban residence, suburbanization, and regional shifts). The population prior to 1790 is treated in Chapter Eg, on colonial statistics. Although reference is made in this essay to the American Indian population, Chapter Ag is specifically devoted to American Indians. Finally, this essay includes a discussion on the Hispanic population.

Back To Top

A difficulty in the study of American historical demography is the lack of some types of data for the calculation of standard demographic measures. The principal source of population data is the decennial census of population, a house-by-house count made by the Census Bureau. In accordance with a constitutional provision for a decennial canvass of the population, the first census enumeration was made in 1790 (see Anderson 1988 for a recent history of the census). The primary reason for the census of population, as set forth in the Constitution, is to provide a basis for the apportionment of members of the House of Representatives among the several states. Until 1902, the census organization was temporary. It was assembled before each decennial census and disbanded after the work was finished. In 1902, the Census Bureau was established as a permanent agency of the government, charged with responsibility for the decennial census and for compiling statistics on other subjects as needed. Currently, this bureau provides population data based on surveys and estimates in addition to making the comprehensive decennial census enumerations.

In accordance with census practice dating back to 1790, each person is counted as an inhabitant of his or her usual place of residence or usual place of abode, that is, the place where he or she lives and sleeps most of the time. This place is not necessarily the same as his or her legal residence, voting residence, or domicile, although, in the vast majority of cases, the use of these different bases of classification would produce identical results. Indians living in Indian Territory or on reservations were not included in the population count until 1890, and in earlier censuses large tracts of unorganized and sparsely settled territory were not covered by enumerators. Alaska and Hawai'i were territories through 1950 and were first included in the United States in the 1960 Census.¹ The totals simply include the two new states for 1960 and later. Through 1930, the data presented are based on complete counts. Some of the data shown from Censuses of 1940 and later are based on sample tabulations (ranging from 3.33 percent to 25 percent), as indicated in the documentation to the tables.

Several tables present estimates from the Current Population Survey (CPS), a survey that is conducted monthly by the Bureau of the Census since 1947 (for example, Table Aa6–8 and Table Aa110–124). These data allow estimates of population size and composition on a more frequent basis without resorting to the costly and complex process of a full census count. Originally, the CPS covered a representative sample of approximately 21,000 interviewed households in areas throughout the United States. This sample size was increased to approximately 35,000 in May 1956 and to approximately 50,000 in January 1967. The sample covered 754 areas comprising 2,121 counties (of the approximately 3,300 counties in the United States), independent cities, and minor civil divisions with coverage in every state and the District of Columbia. Estimates of population characteristics based on the CPS will not agree with those from the census because of differences in procedures for collecting and processing data on subgroups of the populations (for example, racial and ethnic groups).

Exact agreement is not to be expected among the various samples, nor between them and the complete census count; however, the sample data may be used with confidence where large numbers are involved, and may be assumed to indicate patterns and relationships where small numbers are involved. Detailed statements regarding the sampling errors are given in the original sources.

Many errors appear in the census publications of 1790–1840. The data for these censuses were adjusted by county and race, and the revised figures were published in the 1840 and 1870 Censuses. Revised figures for the U.S. population by sex and race for 1790–1840 were published in the 1910 Census. Official revisions by age have not been made, and thus the 1790–1840 age data in this chapter for most race-sex groups add up to totals that differ slightly from the revised figures for race-sex groups. The data for individual states generally reflect the corrected 1840 results, except for 1790, for which corrected data given by Rossiter are used (U.S. Census Bureau 1909).

The Census Bureau has always been concerned about the degree of completeness of enumeration in the decennial censuses, although public interest in census coverage and statistical techniques for estimating coverage was quite limited prior to 1950. Discussions of coverage in earlier censuses were limited mostly to qualitative statements. The quantitative evaluation of census coverage can be done at the individual and aggregate levels. At the individual level, the approaches include re-interview (for example, postenumeration surveys) and record checks (for example, matching of census records and birth records). At the aggregate level, the approaches include demographic analysis (the use of data on births, deaths, and migration, and of life tables, expected sex ratios, etc.) and the use of aggregated data from administrative records (for example, comparing the enrollment in Medicare with the census count of the aged population).

In 1950, the postenumeration survey was thought to be a satisfactory method of determining net census underenumeration. The number of persons missed in the 1950 Census was estimated at about 2.1 million, or 1.4 percent, with corresponding estimates of 1.6 percent for 1940 and 0.7 percent for 1930. Demographers now generally believe, however, that postenumeration surveys tend to understate census omissions because persons missed in a census have an above-average probability of being missed in a postenumeration survey.

Evaluations of census coverage now rely heavily on demographic analysis. Analyses of coverage conducted by the Census Bureau and using demographic analysis show the following estimates and revisions of net census underenumeration (Edmonston and Schultze 1995, Table 2.1):

1940		7.0 million		5.4 percent
1950		6.3 million		4.1 percent
1960		5.6 million		3.1 percent
1970		5.5 million		2.7 percent
1980		2.8 million		1.2 percent
1990		4.7 million		1.8 percent

Studies back to 1880 show general improvement over time (Coale and Zelnik 1963; Coale and Rives 1973). There are also substantial differentials in undercount among groups and regions. For example, the postenumeration survey results for 1990 indicate the following differences in undercount by racial and ethnic groups: 0.7 percent for whites; 4.4 percent for blacks; 5.0 percent for Hispanics; and 4.5 percent for Amerindians. Analyses of census coverage are subject to revision on the basis of additional information and research.

A number of studies have been done on the U.S. census and on various systems that collected vital data in the nineteenth and twentieth centuries (Coale and Zelnik 1963; Condran and Crimmins 1979; Social Science History 1991). Overall, it seems that censuses in the mid-nineteenth century missed anywhere from 5 percent to 25 percent of the population. A careful analysis of the white population from 1880 to 1960 indicates overall underenumeration of 6.1% in 1880, declining to 5.7 percent by 1920 and 2.1% by 1960. Results varied by age and sex, with the very young and the elderly being least well enumerated. Blacks were more likely to be missed than whites. A summary of recent work on the mid-nineteenth-century federal census notes that those more likely to be counted were older, native-born heads of more complex households, with moderate wealth and better-paying occupations, in the political mainstream, and living in smaller communities or rural areas having slow economic and population growths. Those less likely to be enumerated were younger, native-born sons or foreign-born male boarders, living in smaller households, working in low-wage occupations in large, rapidly growing urban areas, and not in the political mainstream (Shryock, Siegel, and Associates 1971, p. 109; Parkerson 1991).

Another technique is the comparison of rates of change with respect to consistency and reasonablen. On this basis, it is believed that figures for the South showed unreasonably low rates of increase for the decade 1860–1870 and abnormally high rates of increase for 1870–1880. The differences are so great that it appears evident that the enumeration of 1870 in this area was seriously incomplete, undoubtedly as a result of the unsettled conditions of the Reconstruction period. For the portion of the United States outside the South, the rate of increase for 1860–1870 was about the same as that for 1870–1880. Therefore, the number initially enumerated in 1870 for the South was revised upward. ² This chapter generally reports the "official” published totals in the census volumes.

Nonetheless, many of these deficiencies do not affect overall results too dramatically. Calculation of rates involves canceling errors. The extent of the errors usually did not change too much from census to census or year to year. In addition, demographic estimates often involve some corrections to the data. Many of the tabular results presented here use uncorrected data; however, some of the estimates do make adjustments.

Although originally the census was intended to provide the basis for allocating seats in the U.S. House of Representatives, the published material grew from a modest one-volume compilation of spare aggregated statistics in 1790 to multiple-volume descriptions of the population, economy, and society by the late nineteenth and early twentieth centuries. Original manuscript returns exist for all dates except 1890, opening great analytical opportunities.³ The census has been the major source for the study of population growth, structure, and redistribution, as well as fertility prior to the twentieth century. Some states also took censuses, usually in years between the federal censuses. A number have been published and some also exist in manuscript form (Dubester 1948). Vital registration was left, however, to state and local governments and, in consequence, it was instituted unevenly. The entire United States was not covered until 1933.⁴

The federal census has consistently inquired about age, sex, and race, although the nonwhite population was not categorized by sex and age until 1820. The Censuses of 1790 through 1840 were taken by households using forms with spaces for counts by age, sex, and race categories. The first nominal census, in which each individual was enumerated on a separate line, was not taken until 1850 (for a description of the enumerator's schedules and instructions, see U.S. Census Bureau 1979 and 2002). Even then, the Censuses of 1850 and 1860 had special schedules for slaves, for whom less detailed information was taken than for whites and free blacks. The American Indian and Asian-origin populations were not separately designated until 1860. The 1850 Census was also the first to ask explicitly about the occupation of each person, although this was done only for males ages 15 and older.⁵ Females were asked about their occupations from 1860 onward, and children were asked from 1870 onward. Economic questions then began to multiply. Questions on unemployment began in 1880; on employment status, industry, and class of worker (employer, employee, working on own account) in 1910; on wage and salary income in 1940; and on personal income in 1950. Questions on real wealth were asked in 1850–1870 and on personal wealth in 1860–1870.

Interest in education dates back to the 1840 Census, with questions on schools, school attendance, and adult literacy. Questions on literacy continued up through the 1930 Census, after which the literacy question was replaced with a query on the number of years of formal schooling completed. Other personal characteristics were also included in the census over time. Marital status and relationship to the head of the household were first asked about in 1880; children ever born in 1890 (and children surviving in 1890–1910); duration of current marriage in 1910; number of times married in 1910; age at first marriage in 1940; and veteran status in 1910. Some questions were not asked in all censuses, and some have been dropped.

The U.S. census also provided, from 1850, information on a person's place of birth and, after 1870, on the nativity of each person's parents. This was either state of birth for the native-born or country of birth for the foreign-born. These data permit study of international migration (for example, the geographic distribution of the foreign-born) and also analysis of internal migration by providing cross-classification of the native-born by birth and current residence (from 1860 onward).⁶ Internal migration is a rather difficult issue because of lack of evidence on date of change of residence between the time of birth and current residence. For the foreign-born, questions on duration of residence in the United States were asked in the Censuses of 1890–1930 and again in 1970–2000, but a question about duration of current residence was not asked of all inhabitants until 1940 (when a question was asked concerning a person's place of residence five years prior to the census). Also of interest have been a person's mother tongue (1910–1940, 1960–1970) and language customarily spoken (1910, 1980–2000). An ambiguous question of self–identification of ancestry was introduced in 1980.

Demography, the study of human populations, depends heavily on measurement and estimation techniques. Most of the results presented in this chapter are simple tabulations or standard demographic rates. However, a number of the newer findings arise from rather sophisticated techniques (Haines 1998; Haines and Preston 1997). Estimation of better demographic information is of importance for research in social, demographic, and economic history. Basic demographic structures and events, reflected in birth and death rates, population size and structure, growth rates, the composition and growth of the labor force, marriage rates and patterns, household composition, the levels and nature of migration flows, causes of death, urbanization and spatial population distribution, and so forth, determine the human capital of society as producers and consumers and also how that human capital reproduces, relocates, and depreciates. Demographic events are important both as indicators of social and economic change and as integral components of modern economic growth.

Back To Top

Every modern, economically developed nation has undergone a demographic transition from high to low levels of fertility and mortality. This was certainly true for the United States, which has experienced a sustained fertility decline from at least about 1800 (see Table Ab1–10 and Table Ab40–51 and Table Ab52–117). Around that time, the typical American woman had about seven or eight live births during her reproductive years, and the average person had an expectation of life at birth of 39–40 years, an outcome greatly affected by high infant mortality. However, the American pattern was distinctive. First, the American fertility transition was under way from at least the beginning of the nineteenth century, and some evidence indicates that family size was declining in older settled areas from the late eighteenth century on. Other Western developed nations, with the exception of France, began their sustained, irreversible decline in birth rates only in the late nineteenth or early twentieth century (Coale and Watkins 1986). It is perhaps not coincidental that both France and the United States experienced important political revolutions in the late eighteenth century and were then characterized by small-scale, owner-occupier agriculture. Second, it appears that fertility in America was in sustained decline long before mortality. This is in contrast to the stylized view of the demographic transition, in which mortality decline precedes or occurs simultaneously with fertility decline. Mortality in the United States did not stabilize and begin a consistent decline until about the 1870s. Third, these demographic processes were influenced by both the large volume of international net in-migration and significant internal population redistribution to frontier areas and to cities, towns, and (later) suburbs.

Although the American case may be, in many respects, sui generis, it furnishes a long-term view of a completed demographic transition with accompanying urbanization. The United States was a demographic laboratory in which natives and migrants, different racial and ethnic groups, and varying occupational and socioeconomic strata experienced these significant behavioral changes in a fertile, land-abundant, resource-rich country.

Table Aa15–21 provides summary measures of population growth and its components by decades from 1790 to 2000.⁷ This table (as is Table Aa9–14) is organized around the demographic balancing equation, which states that the decade rate of population growth (RTI) equals the crude birth rate (CBR) minus the crude death rate (CDR) plus the rate of net migration (RNM). The difference between the birth rate and the death rate is the rate of natural increase (RNI) (that is, RTI = CBR - CDR + RNM, and CBR - CDR = RNI). Table Aa9–14 provides entirely new annual estimates of the population as of July 1 for the nineteenth century.

Several features of the American demographic transition can be discerned from Table Aa9–14, Table Aa15–21, Table Aa22–35 and Table Ab1–10 (see also Figure Aa-A). The United States experienced a truly remarkable population increase during its transition. From a modest 4 million inhabit- ants in 1790, the population grew to more than 281 million in 2000, an average annual growth rate of about 2 percent per year. In the early years of the republic, population growth rates were even higher, more than 3 percent per annum for the period 1790–1810 and again in the 1840s and 1850s. Such rapid growth is historically rather unusual and is comparable to the recent experience of some developing nations. Growth rates of this magnitude would lead to a doubling of the population in slightly more than two decades (approximately twenty-three years). The surge of growth in the 1840s and 1850s was particularly caused by a significant increase in migration from abroad - the now familiar story of Irish, Germans, and others from Western and Northern Europe fleeing the great potato famine, the "Hungry Forties,” and political upheaval and seeking better farming, business, and employment opportunities in the Western Hemisphere. Natural increase had been declining from the early 1800s, largely because of a decline in birth rates for both white and black populations. Some of the decline in natural increase in the 1840s and 1850s was also likely attributable to rising mortality in those decades. Table Ab1–10 indicates, however, that mortality did decline steadily from the 1870s onward.

Another feature notable in Table Aa9–14 and Table Aa15–21 is the dominant role played by natural increase in overall population growth. In the decades before 1840, less than one sixth or one seventh of total growth originated from net migration. With the surge in overseas migration after 1840, however, the share of net migration in total increase rose to between one fourth and one third. Notably, the share of labor force growth accounted for by migration was higher because migration was selective of persons of an age to participate in the labor force. Nonetheless, despite declining birth rates, the American population grew rapidly in the nineteenth century, principally from an excess of births over deaths, although it must be recognized that the births to the foreign-born and their descendants contributed importantly. If it could be assumed that no immigration occurred after 1790 and that the natural increase of the colonial stock population had been what it actually was (with no effect of immigration on the natural increase of the native-born), then the white population would have been about 52 million in 1920, or about 55 percent of what it actually was. The surge in migration after 1840 can also be recognized in the rise in the proportion of the population that was foreign-born from less than 10 percent in 1850 (the first census for which such data were available) to nearly 15 percent in 1890 and 1910 (see Table Aa1884–1895).

The effects of immigration restriction after World War I are apparent in the reduced rate of net migration after 1920. The Great Depression had a dramatic damping effect on both fertility and migration from abroad, resulting in the only decade of net out-migration since the first census in 1790. The post-World War II "baby boom” is apparent in the higher crude birth rates in the 1940s and 1950s. More recent changes in immigration regulations clearly affected the surge in net in-migrants, when 37 percent and 47 percent of population growth in the 1970s and 1990s, respectively, were attributable to this source. These proportions are larger than those in the decades preceding both the Civil War and World War I.

Table Aa22–35 provides some summary information about the changing composition of the American population by race and ethnicity, sex, and age since 1790 (for a discussion of the problems of defining race and ethnicity, see Haines 2003). The share of the white population in the total increased fairly steadily between 1790 and 1930 (from 80.7 percent to 89.8 percent), largely as a result of the substantial net immigration of white Europeans up through the 1920s. The restriction of migration after World War I, combined with the higher natural increase in the black population, led to a slow rise in the representation of the black population in the total. In recent decades, this has been accelerated by a rising net in-migration of individuals from sub-Saharan Africa, the Caribbean, and other areas with a large proportion of persons of African descent. Really dramatic, however, was the rising share of other ethnic groups in the late twentieth century. From less than 1 percent of the population in 1960, this share rose to 12.6 percent in the 2000 Census. This had its origins mostly in a heavy immigration of persons from East and South Asia; however, it also stems in part from the rapid increase in the American Indian population. The latter was the result of a large amount of ethnic self-reidentification. The political, social, and economic consequences of the changing racial and ethnic composition of the American population are already apparent and are likely to be profound in the future.

The median age of the population has been steadily rising since 1800, from about 16 in 1800 to about 23 in 1900 to more than 35 in 2000.⁸ This is mostly a function of the decline in birth rates (see Table Ab1–10). Birth rates (and not death rates) are the major determinants of the age structure of a population not subject to significant migration flows. Although the effects of migration on age structure are an issue for the United States, the fall in birth rates has dominated the aging of the American population in the last two centuries. A similar result holds for sex composition. Table Aa22–35 gives the sex ratio (males per 100 females) since 1790. Slightly more males than females are born to most human populations (about 5 percent more); however, male mortality usually tends to exceed female mortality at all ages (see Table Ab988–1047). For a high-fertility population with a young age structure, such as the United States in the early nineteenth century, this leads to a majority of the population being male. As birth rates fall and the population ages, however, the older age groups with a majority being women come to dominate. This was offset in the nineteenth and early twentieth centuries by immigration, in which migrant selectivity favored males.⁹ However, the sex ratio declined between 1910 and 1980 as the population aged. More recently, the effects of male selective migration have again been seen in the rising sex ratio.

Another notable feature in Table Aa22–35 is the share of the foreign-born in the population. As already mentioned, the Census of 1850 was the first to ask about the place of birth of all persons. Already in 1850, almost 10 percent of the American population was born abroad, in this case overwhelmingly in Europe (more than 90 percent; see Table Ad354–443). This rose to almost 15 percent in the Census of 1910, based on the huge immigration from the 1840s up through World War I. Immigration restriction from the 1920s onward, largely in the form of quotas by national origin, led to a reduction in net immigration (see Table Aa9–14) and a drop in the proportion of the foreign-born to 4.7 percent in 1970. However, recent changes in immigration legislation and a surge in immigration, now from Asia, Latin America, the Caribbean, and Africa, have once again raised the share of the foreign-born in the American population to more than 10 percent in the 2000 Census. We continue to be a nation of immigrants.

Table Aa22–35 also demonstrates that the United States experienced a transition from a predominantly rural society to a largely urban one. In 1790, only 5 percent of the population was living in officially classified urban areas. This rose to about 20 percent on the eve of the Civil War in 1860. It was not until 1920 that more than half of the population was enumerated as urban. Currently, more than three fourths of the American population is classified as urban; however, this understates the urban population because there are "rural” areas at the edges of metropolitan areas that will soon be placed in the urban category.

The urban population is currently defined as comprising all persons living in urbanized areas and in places of 2,500 inhabitants or more outside urbanized areas (for details, see the text for Table Aa684–698). In addition, the Census Bureau has, since 1950, provided information on metropolitan statistical areas (MSAs), consolidated metropolitan statistical areas, and primary metropolitan statistical areas. The general concept of a metropolitan area (MA) is that of a core area containing a large population nucleus, together with adjacent communities having a high degree of economic and social integration with that core. Currently defined MAs are based on application of 1999 standards (which were effective in 1996) to 1990 decennial census data. These MA definitions were announced by the Office of Management and Budget effective June 30, 1996. Table Aa824–831 and Table Aa832–1033 provide some information on MAs, using the different boundaries and definitions that have evolved since 1950. These data indicate that, at present, about one fourth of the population lives in areas that are classified as central cities, about one half in suburbs (non-central city portions of MSAs), and about one fourth in rural places.

The effects of regional differences in population growth are also apparent. The 4.5 million inhabitants in 1790 were clustered along the Atlantic coast, about evenly divided between North (New England and Middle Atlantic regions) and South (South Atlantic region). By 1860 only 51 percent of the 31 million Americans were still living in these regions, but this had fallen to 41 percent in 1920 and 38 percent by 1990. Regions of earliest settlement grew at an average rate of 1.6 percent per annum over the whole period 1790–1990, while the entire United States was growing at 2 percent. This regional disparity was driven, of course, by the relentless westward movement of population, agriculture, and industry. Much of the growth that did occur on the Atlantic coast was in yet another "frontier” - urban areas. In the regions of original European settlement, cities and towns grew from just 5 percent of the population in 1790 to 28 percent in 1860, 61 percent in 1920, and 74 percent in 1990, an annual growth rate of 2.9 percent per annum, while the growth of the rural population was merely 1 percent per annum. For the nation as a whole, this led to the large increase in the share of national urban population.

Not surprisingly, the demographic "center” of the nation has moved from the Atlantic coastal states (New England, Middle Atlantic, and South Atlantic regions) to the Midwest (East North Central and West North Central) and western South (East South Central and West South Central). By 1920, the Mountain and Pacific states were still relatively small demographically, comprising less than 10 percent of the total population (as opposed to 21 percent in 1990). Two migrations, thus, were driving the numbers - the movement from east to west and the movement from rural to urban areas. The average annual growth rate over that period was 3.4 percent for the urban population, in contrast to only 1.4 percent for rural dwellers. As we have every indication that the birth rates were lower and death rates higher in urban relative to rural areas, the more rapid growth of urban areas originated in population redistribution and not differences in natural increase. This rural-to-urban shift reflects, of course, labor market conditions as the economy changed its structure of opportunities from a rural population of smallholder agricultural proprietors to an urban, industrial, and service-based economy made up predominantly of employees. This is certainly exemplified by the increase in the nonfarm share of the labor force from 25.6 percent in 1800 to 44.2 percent in 1860, 74.1 percent in 1920, and 97.3 percent in 1990 (Weiss 1992, Table 1.1). A primary motive for migration in ordinary times is to take advantage of wage and income differences across space, which substitutes factor mobility for interregional trade in goods and services.

Urbanization did spread across regions, albeit unevenly. The Northeast was the urban-industrial center of the nation in the nineteenth century. By 1860, New England and the Middle Atlantic regions had 61 percent of the nation's urban inhabitants but only 33 percent of the overall population. Conversely, the South had 17 percent of the urban population but 36 percent of the overall total. Even in 1920, the Northeast still had 41 percent of urban dwellers, with the Midwest close behind at 33 percent. The South still had only 17 percent. However, by 1990 the Northeast had only 21.5 percent of America's urban dwellers, with the Midwest at 23 percent. The South now had 31.4 percent and the West 24 percent. The "Sunbelt” is not an illusion.

From 1850 onward we are able to examine migration by place of birth and current residence (also see Chapter Ac, on internal migration). The proportion of the native-born population residing outside the state of birth ("lifetime” migrants) was relatively stable from the middle of the nineteenth century - 23.3 percent of the white population in 1850, 23.5 percent in 1890, 23.9 percent in 1920, and 32 percent in 1990. The nonwhite population had lower rates of lifetime mobility in this period, about 15–20 percent until after 1920. In the nineteenth century, much of this interstate movement was on an east-west axis until the closing of the frontier at the end of the nineteenth century. For instance, in 1850, of those born in Pennsylvania but residing elsewhere, 67 percent could be found in Ohio, Indiana, or Illinois, whereas 77 percent of those born in South Carolina but residing outside that state were found in Georgia, Alabama, Mississippi, and Tennessee. A variety of explanations has been advanced for the migration along latitudes, but recently it has been shown that real and human capital invested in seed, livestock, implements, and farming techniques made movement along climatic bands highly rational. This also provides a partial explanation for the greater preference of the bulk of the nineteenth-century immigrants from Northern and Western Europe for the Northeast and the Midwest: their human capital matched that climatic band better. This was true for those going to rural areas at least. The remainder of the explanation was largely the greater opportunities in the more rapidly urbanizing and industrializing North, as well as the tendency of migration streams, once established, to grow along familiar paths. This has largely disappeared, as the varied opportunities of cities have drawn people in many directions, especially to the South and West regions.

Agrarian motives for migration diminished as the frontier closed in the late nineteenth century and as rural population growth slowed dramatically (to only 0.4 percent per year over the period 1890–1990). For most of the nineteenth century, migration flows westward were consistent with the land availability hypothesis discussed in connection with the fertility transition (see Chapter Ab, on vital statistics). Rural migrants moved west to secure cheaper, good-quality land. Frederick Jackson Turner's thesis that the frontier was a demographic "safety valve” in nineteenth-century America remains a durable view. Nevertheless, late in the century, rural-to-urban flows assumed the dominant role. However, much of the rural-urban migration was within regions or along an east-west axis because the bulk of urban and industrial growth from the Civil War to 1920 was in the Northeast or Midwest. Notably, the South failed to increase its share of urban population over this period. The major shift to a south-to-north movement began on a large scale only with the radical shifts in demand for labor accompanying World War I and the restriction, after 1921, of cheap immigrant labor. The shift to the Sunbelt came even later, largely after World War II. Changes in transportation technology, particularly electric street and underground railways and later the automobile and motorized bus, led to a movement out of central cities and into suburban communities. This process was under way in parts of the Northeast by the end of the nineteenth century, but really accelerated after World War I, and again after 1945. So, for instance, during the 1920s the rural part of metropolitan districts (as defined by the Census Bureau) increased by 55 percent, faster than any part of the metropolitan population except for small cities. This development was suburbanization.

The urbanization process was accompanied by a filling out in the city size hierarchy. Large cities did tend to grow most rapidly. In 1810 there were only two cities with populations of more than 50,000 (New York and Philadelphia), and together they made up 29 percent of the total urban population. By 1860, there were sixteen places with populations of more than 50,000, containing 50 percent of urban inhabitants. In 1920, in the first census when more than half of the American population was urban, 144 cities exceeded 50,000 persons (with 25 cities of more than 250,000 inhabitants), and they had 60 percent of city dwellers. The three largest cities with populations of more than one million each (New York, Chicago, Philadelphia) together had 19 percent of America's urbanites. In 1990, with three fourths of the population being urban, 555 places had populations of more than 50,000, 64 had more than 250,000, and 8 places had more than one million. The share of the urban population in those cities with populations of 50,000 and more was still only 62 percent of the urban population. However, the urban size hierarchy did not become distorted, as it has in some developing nations. That is, large cities did not grow such that medium and smaller urban places became unimportant. There were 213 places with populations of 5,000–50,000 in 1860, accounting for 41 percent of the urban inhabitants. In 1920 this number had risen to a total of 1,323 places, with 32 percent of the urban population. These places numbered 4,929 in 1990, with 41 percent of urban dwellers (see Table Aa684–698 and Table Aa699–715). And this urban growth had powerful economic linkages. Considerable industrial output of the period 1865–1920 was devoted to providing infrastructure and materials for housing, transportation, and public services for this massive population shift to towns and cities. Iron and steel for sewer and water pipes, bridges, rails, structural pieces, and nails; concrete, stone, brick, and asphalt for roads and structures; cut timber; transport equipment; glass; and so forth were demanded in huge quantities to build the cities.

Migration patterns, both internal and international, did affect regional population growth rates and shares. In 1790 the North and South had about 50 percent of the total population. But differential migration, and not differential natural increase, began to drive the share in the North upward as slower population growth in New England was balanced by more rapid growth in New York, Pennsylvania, and later the Midwest. The Northeast and Midwest together accounted for 56 percent of the nation's inhabitants in 1830 and 62 percent in 1860, compared to 35 percent for the South at the latter date. This demographic shift alone was instrumental in the political crisis leading up to the Civil War, as southern representation in the Congress slowly ebbed.

The regional preference of migrants from abroad, once they had landed in the United States, was strongly in favor of the Northeast and Midwest and not the South. For instance, in 1860 a mere 5.6 percent of the South's white population was foreign-born, whereas the proportion was 19.3 percent in the Northeast and 17.4 percent in the Midwest. For 1910, the proportion of foreign-born living in the Northeast had risen to 26.2 percent. It had fallen to 3.5 percent in the South and held at 17.4 percent in the Midwest. Further, at the latter date, only 6.1 percent of Southern whites had one or both parents foreign-born, whereas 30.1 percent of white residents of the Northeast were first-generation native-born. This had profound political implications in terms of regional growth both before and after the Civil War. Not only did it change the Congressional balance of power, but it also limited the labor supply in the South for industrial and agricultural development throughout the nineteenth and early twentieth centuries. The Southern share slipped even further, to 31 percent in 1920, whereas the Northeast and Midwest held about steady at 60 percent. These population realignments were both cause and effect of rapid industrial growth in the postbellum era, as many of the rural migrants and most of the later immigrants were destined for northern cities. By the late twentieth century, however, the South had managed to raise its share in the national total to 34 percent, whereas the Northeast and Midwest had slipped to 44 percent, as the movement to the Sunbelt became a dominant feature of the American migration story.

The census also makes it possible to study aspects such as marriage and household structure. Census-based measures of nuptiality, such as the median age at first marriage and the singulate mean age at first marriage, are among the most revealing indexes of marriage and changes in marriage age over time (see Table Ae481–488 and Table Ae489–506). Similarly, the census offers data on marriage incidence (for example, percentage single or married by age) and also information about widowhood and divorce. The census began to ask a question about the marital status of individuals in 1880 and began publishing tabulations of this in 1890.¹⁰ These data permit the study of long-term trends in marriage (see Table Aa614–683). From this and other information, we find that the age at marriage for both men and women was rising from the late nineteenth century (and likely from about 1800) until it reached a peak in 1900 (Haines 1996). Even at that point, however, the American age at marriage and the proportion of persons still single at ages 45–54 were low by the standards of Western and Northern Europe, the origins of much of the American white population (Coale and Watkins 1986). Thereafter, marriage age declined slowly until about 1970, after which there was a sharp rise in age at first marriage for both men and women, and whites and nonwhites. Similarly, divorce rates have been on the rise in recent decades (see Table Ae507–513).

Household composition and its change over time are also discernible from census documents. It is especially important to have the Integrated Public Use Microdata Series (IPUMS) because the published documents often gave very little detail on this topic.¹¹ Newer theoretical and methodological perspectives have changed our views on how household and family structure should be analyzed. This is dealt with in Chapter Ae, on family and household composition.

Back To Top

Estimating the size and composition of the Hispanic population or those of its subgroups prior to 1970 is difficult because the census rarely collected information other than birthplace variables that could allow such measurement.¹² In 1930, for example, the Census attempted to classify the Mexicans as a "race,” but this effort officially added nothing to previous information using birthplace variables: enumerators were to record as Mexican "all persons born in Mexico, or having parents born in Mexico, who are not definitely white, Negro.”  The race clause actually restricted Mexican descent because persons whose appearances met enumerators' standards for white, black, or other race categories were not to be recorded as Mexican, and variations in enumerator policies leave the meaning of published results unclear. Additional variables have sometimes offered the means for identifying Hispanics, especially the intermittent recording of language characteristics beginning with the 1910 Census.

In 1950 the Census first experimented with a more general Hispanic indicator, one that identified individuals beyond the second generation. The census recorded white persons with Spanish surnames in Arizona, California, Colorado, New Mexico, and Texas. Undoubtedly designed to focus on Mexican-origin Hispanics who lived in the Southwest, the surname variable did not provide any information outside the region, nor did it distinguish among subgroups. Although the Census Bureau knew that the Spanish surname lists omitted certain persons who were Hispanic, the opportunity to reach beyond the second generation encouraged the use of such indicators. More extensive lists were employed in the 1960, 1970, and 1980 Censuses, still limited to the five Southwestern states.

The 1970 Census marked an important change. In that year, two new efforts were made to classify the population. The first relied on a combination of variables that located "Spanish Americans,” including Spanish as a mother tongue, Puerto Rican family background in three Northeastern states, and Spanish surname in five Southwestern states. The second effort asked respondents to identify themselves in four "Spanish origin” categories (Mexican, Cuban, Puerto Rican, others). Although the 1970 origin question did not produce satisfactory results, self-identification of Hispanic origin became increasingly detailed and accurate in the 1980 and 1990 Censuses and has allowed estimates of the Hispanic population and its subgroups to be made with much greater confidence (Bean and Tienda 1987).

For years prior to 1980, the estimates provided in Table Aa2189–2215 make use of individual-level census sample data from the IPUMS project (Ruggles and Hall 1999).¹³ Hispanic origin is determined by place of birth, parents' place of birth, and Spanish surname; these data are available for all census years considered and produce consistent estimates; the methodology for these estimates can be found in Gratton and Gutmann (2000). Other variables, such as language spoken, are available in certain censuses and lead to more complete and accurate estimates; these are not, however, consistent in their coverage from year to year. For 1980 and 1990, the estimates reported use the results published in the census, based on the Hispanic identity question asked of all persons. The foreign-born estimates, however, are based on a question only asked of a sample of census respondents and the totals vary from the complete count data.

A notable feature of the Hispanic population of the United States has been its rapid growth, from high rates of both natural increase and net in-migration. In recent years there has also been some significant ethnic reidentification of the existing resident population toward Hispanic identity, which has further enhanced the apparent growth rate. The percentage share of the Hispanic population has grown from an estimated 0.6 percent in 1850 to 0.9 percent in 1910 to 4.4 percent in 1970 and 8.7 percent in 1990 (Gratton and Gutmann 2000, Table 2). In short, the Hispanic population has been growing faster than the American population as a whole over the past 150 years.¹⁴ The largest groups in the Hispanic population have been of Mexican origin (58.5 percent in 2000) and Puerto Rican origin (9.6 percent in 2000), although the shares have been shifting toward other national-origin groups. The Mexican-origin population has retained its dominant share, however, based on continued large net in-migration.

The sex ratio of the Hispanic population slightly favors males (103.8 males per 100 females), well above the overall sex ratio for the native-born non-Hispanic population (95.1 in 1990). Differential net immigration of males accounts for this, as it did for other foreign-born groups historically. The Hispanic population is overwhelmingly urban (92 percent in 1990), and a majority (63 percent in 1990) was born in the United States. Most Hispanics are concentrated in a few states in the West and Southwest (California, Texas, Arizona, New Mexico) and the Northeast (New York, New Jersey) and in Florida; however, a large number are now found in other areas (such as Illinois) and indeed they are distributing themselves throughout the nation where work can be found. A slight majority of Hispanics identify themselves as white (51.7 percent in 1990), but only 3.4 percent self-identify as black. Many (42.7%) classed themselves as "other race” in 1990, likely because of issues of mixed racial categories.

Back To Top

This essay has focused on the evolution of the American population over the period 1790 to the present. The discussion has perforce covered the nature of demographic data and measurement and the size, growth, and structure of the population. It has also dealt with urbanization with some treatment of both internal and international migration. The relatively rapid population growth over this period (averaging about 2 percent per year) was driven largely by high (though declining) birth rates and moderate levels of mortality; however, immigration was also significant. About three fourths of the overall growth was the result of natural increase and about one fourth was attributable to net in-migration. More than 34 million persons entered the United States between the 1790s and 1920 and an additional 29.4 million between 1921 and the present.

The population has experienced substantial changes over these two centuries. The population has aged significantly, moving from a median age of about 16 in 1800 to about 35 in 2000. Accompanying this has been a shift from a majority male to a majority female population. In 1800, about 51 percent of the population was male, but in 2000 it was only about 49 percent. American men and women now marry at the oldest ages in history, and an increasing number have experienced divorce.

Further, Americans have moved from predominantly rural and small-town life (5 percent of the population was urban in 1790) to become a nation of urban dwellers and suburbanites. At least three fourths of the population was living in officially designated urban places in 1990. Americans are extremely mobile, and migration has been a fact of life. In recent years, about 15–20 percent of the population has moved to a different house each year. Within the nation's boundaries, there has been significant movement from east to west, following the frontier (until the late nineteenth century); from rural to urban areas; and, later, from central cities to suburbs, from south to north, and ultimately to the Sunbelt. These developments have been responsible for changing the United States from a rural to an urban nation and then to a suburban nation.

The population of the United States shifted from the original areas of settlement on the Atlantic coast to the center of the nation and later to the Pacific and Mountain states. Migration from abroad, first from Western and Northern Europe and then, after about 1890, from Central, Eastern, and Southern Europe, occurred in waves in response to upswings in business cycles and the expansion of economic opportunities. This flood of immigrants both directly augmented population growth rates and indirectly acted to raise birth rates before immigration was severely restricted in the 1920s by legislation and subsequently by the Great Depression. However, it left an indelible stamp on the economy, society, and culture.

Back To Top

Table Aa22–35.

Back To Top

Anderson, Margo J. 1988. The American Census: A Social History. Yale University Press.

Bean, Frank D., and Marta Tienda. 1987. The Hispanic Population of the United States. National Committee for Research on the 1980 Census. Russell Sage.

Coale, Ansley J., and Norfleet W. Rives Jr. 1973. "A Statistical Reconstruction of the Black Population of the United States, 1880–1970: Estimates of True Numbers by Age and Sex.” Population Index 39 (1): 3–36.

Coale, Ansley J., and Susan Cotts Watkins, editors. 1986. The Decline of Fertility in Europe. Princeton University Press.

Coale, Ansley J., and Melvin Zelnik. 1963. New Estimates of Fertility and Population in the United States: A Study of Annual White Births from 1855 to 1960 and of Completeness of Enumeration in the Censuses from 1880 to 1960. Princeton University Press.

Condran, Gretchen A., and Eileen Crimmins. 1979. "A Description and Evalu- ation of Mortality Data in the Federal Census: 1850–1900.” Historical Methods 12 (1): 1–23.

Dubester, Henry J. 1948. State Censuses: An Annotated Bibliography of Censuses of Population Taken after the Year 1790 by States and Territories of the United States. U.S. Government Printing Office.

Edmonston, Barry, and Charles Schultze, editors. 1995. Modernizing the U.S. Census. National Academy Press.

Gratton, Brian, and Myron P. Gutmann. 2000. "Hispanics in the United States, 1850–1990: Estimates of Population Size and National Origin.” Historical Methods 33 (2): 137–53.

Haines, Michael R. 1996. "Long Term Marriage Patterns in the United States from Colonial Times to the Present.” The History of the Family: An International Quarterly 1 (1): 15–39.

Haines, Michael R. 1998. "Estimated Life Tables for the United States, 1850–1910.” Historical Methods 31 (4): 149–69.

Haines, Michael R. 2003. "Ethnic Differences in Demographic Behavior in the United States: Has There Been Convergence?” Historical Methods 36 (4): 155–92.

Haines, Michael R., and Samuel H. Preston. 1997. "The Use of the Census to Estimate Childhood Mortality: Comparisons from the 1900 and 1910 United States Census Public Use Samples.” Historical Methods 30 (2): 77–96.

Haines, Michael R., and Richard H. Steckel, editors. 2000. A Population History of North America. Cambridge University Press.

Parkerson, Donald H. 1991. "Comments on the Underenumeration of the U.S. Census, 1850–1880.” Social Science History 15 (4): 509–15.

Robinson, J. G., B. Ahmed, et al. 1993. "Estimation of Population Coverage in the 1990 U.S. Census Based on Demographic Analysis.” Journal of the American Statistical Association 88 (42): 1061–79.

Ruggles, Steven, and Patricia Kelly Hall, editors. 1999. "IPUMS: Integrated Public Use Microdata Series.” Historical Methods 32 (3): 102–58.

Shryock, Henry S., Jacob S. Siegel, and Associates. 1971. The Methods and Materials of Demography. U.S. Government Printing Office.

Siegel, Jacob S. 1974. "Estimates of Coverage of the Population by Sex, Race, and Age in the 1970 Census.” Demography 11 (1): 1–23.

Social Science History. 1991. 15 (4).

U.S. Census Bureau. 1909. A Century of Population Growth from the First Census of the United States to the Twelfth, 1790–1900. U.S. Government Printing Office.

U.S. Census Bureau. 1979. Twenty Censuses: Population and Housing Questions, 1790–1980. U.S. Government Printing Office.

U.S. Census Bureau. 2002. Measuring America: The Decennial Censuses from 1790 to 2000. U.S. Government Printing Office.

Weiss, Thomas. 1992. "U.S. Labor Force Estimates and Economic Growth, 1800–1860.”  In Robert E. Gallman and John Joseph Wallis, editors. American Economic Growth and Standards of Living before the Civil War. University of Chicago Press.

Back To Top