obtained for many of the items-so much more that, beyond the basic counts, which were released promptly, publication of these data was not completed until nearly 1890.

The census act of 1880 replaced the marshals and their assistants with specially appointed agents (experts assigned to collect technical data, such as on manufacturing processes), supervisors, and enumerators, every one of whom was forbidden to disclose census information. Maintaining the confidentiality of the data was a result of what some people regarded as the census' invasion of privacy, especially since prior to the 1880 census, there was no law limiting the extent to which the public could use or see the information on any schedule. (Subsequent demographic and economic censuses, as well as most surveys, have been carried out according to statutes that make compliance mandatory, with penalties for refusal; and responses confidential, with penalties for disclosure. Congress codified these laws in 1954 as Title 13, U.S. Code.) For the first time, enumerators were given detailed maps to follow, so they could account for every street or road and not stray beyond their assigned boundaries. (The National Archives' Cartographic and Architectural Branch maintains this map collection.4) Again, in 1890, there was an extension of the decennial census's scope, and some subjects were covered in even greater detail than in 1880. Data were collected in supplemental surveys on farm and home mortgages and private corporations' and individuals' indebtedness. The 1890 census also used, for the first time in history, a separate schedule for each family.

Herman Hollerith, who had been a special agent for the 1880 census, developed punch cards and electric tabulating machines to process the 1890 census returns, considerably reducing the time needed to complete the clerical work. (Hollerith's venture became part of what is now the IBM Corporation.) Both the cards and the machines were improved progressively over the next 50 years.5

The 1890 census was historic in another way. In the first volume of the results, the Superintendent of the Census

wrote

"Up to and including 1880, the country had a frontier of settlement, but at present the unsettled area has been so broken into by isolated bodies of settlement that there can hardly be said to be a frontier line. In

the discussion of its extent, its westward movement,

etc., it can not, therefore, any longer have a place in the census reports.6

Commenting on this statement, historian Frederick Jackson Turner wrote in 1893 that, "up to our own day, American history has been in a large degree the history of the colonization of the Great West. The existence of an area of free land, its continuous recession, and the advance of American settlement westward, explain American development."7 The censuses that followed 1890 reflected the filling in, rather than the expansion of, the colonized areas; symbolizing a turning point in America's development as a Nation.

Censuses of 1900 to 2000

Although the censuses in the early 1900s did not witness the expansion of inquiries as had been witnessed in the late nineteenth century, geographic coverage of the census reflected the Nation's growing status as a political and military power. As a result of the country's expanding global influence, the following areas saw their first censuses administered by the United States in the early 1900s:

■ Following its annexation in 1898, Hawaii (where the local government took a census every 6 years from 1866 though 1896) was included in the 1900 census, which also had the first count of the U.S. population abroad (Armed Forces and Federal civilian employees, and their households).

■ The War Department carried out an enumeration in Puerto Rico in 1899 following that island's acquisition from Spain in 1898 (there were periodic censuses from 1765 to 1887 under Spanish rule), and there have been decennial censuses conducted in Puerto Rico from 1910 to the present.

The U.S. Census Bureau compiled and published one census of the Philippine Islands following their accession by the United States in 1898; this census was taken under the direction of the Philippine Commission in 1903. (Under Spanish rule, there had been censuses in 1818 and 1876. The Philippine legislature directed in 1918, and the Commonwealth's statistical office began periodic enumerations in 1939. The Philippines became an independent republic in 1946).

■ The Isthmian Canal Commission ordered a general census of the Panama Canal Zone when the United States took control of the area in 1904; there was another general census in 1912 and several special censuses at various

*The National Archives Cartographic and Architectural Branch, 8601 Adelphi Road, College Park, MD 20740-6001, 301-713-7040. 5For more information, see “100 Years of Data Processing: The Punch card Century." U.S. Department of Commerce, Bureau of the Census, January 1991.

126

"U.S. Census Office, Compendium of the Eleventh Census: 1890. Part 1.-Population. Washing, DC: Government Printing Office, 1892, p. xlviii.

The Frontier in American History. New York: H. Holt & Company, 1958, p. 1.

Measuring America

times. The Canal Zone was included in the U.S. censuses from 1920 to 1970. (Sovereignty over the Zone was transferred to the Republic of Panama in 1979.)

Following the United States' occupation of Guam in 1899, the local governor conducted a census there in 1901. The island has been included in U.S. censuses from 1920 onward.

The governors of American Samoa conducted censuses at various times after the United States acquired the islands in 1900, and the population was enumerated in U.S. censuses from 1920 onward.

Prior to the acquisition of the Danish Virgin Islands by the United States in 1917, the Danish government took periodic censuses between 1835 and 1911. The U.S. census was conducted in 1917 and the islands appeared in the 1930 and subsequent U.S. censuses.

■ A census of Cuba was conducted under a provisional U.S. administration in 1907. There were earlier censuses under Spanish rule (which ended in 1898). The U.S. War Department conducted an enumeration in 1899, and subsequent censuses were overseen by the Republic (established in 1901) beginning in 1919.

There had been quinquennial Japanese censuses from 1920 to 1940 for the islands that became the U.S. Trust Territory of the Pacific Islands. The U.S. Navy conducted a census in 1950, and the U.S. High Commissioner carried out a census in 1958, the results of which appeared in the 1960 U.S. census reports. The U.S. Census Bureau conducted the 1970 and 1980 censuses8; in 1980, 1990, and 2000, there was a separate census of the Northern Mariana Islands, which had been part of the Trust Territory.

A number of the censuses noted above collected data on agriculture, housing, and economic subjects and included enumerations on isolated islands, such as Truk and Yap, mainly in the Pacific.

Stateside Developments

From the 1840 through the 1900 censuses, a temporary census office had been established before each decennial enumeration and disbanded as soon as the results were compiled and published. Congress established a permanent Bureau of the Census in 1902, in the Department of the Interior, so there would be an ongoing organization capable of carrying out censuses throughout the decades instead of concentrating all the work in those years ending in "0." The Census Bureau moved to the new Department of Commerce

In 1986, compacts of free association were implemented between the Federated States of Micronesia and the Marshall Islands, and the United States. Under the terms of Title 13, U.S. Code, the United States was no longer authorized to take the decennial censuses in those areas that were formerly part of the Trust Territory.

Measuring America

and Labor in 1903, and remained within the Department of Commerce when the Department of Labor was split off in 1913.

The 1910 census was the first for which prospective census employees took open competitive examinations throughout the country (since 1880, appointees had been given noncompetitve tests). The way in which results were published also was changed, with those statistics that were ready first-especially those in greatest demand (such as the total population of individual cities and states, and of the United States as a whole)-issued first as press releases, then in greater detail as bulletins and abstracts, the latter appearing 6 months to 1 year before the final reports were issued. In 1920 and 1930, there were minor changes in scope. A census of unemployment accompanied the 1930 census -data were collected for each person reported to have a gainful occupation, but who was not at work on the day preceding the enumerator's visit.

Sampling. In many ways, 1940 saw the first modern census. One of its major innovations was the use of statistical techniques, such as probability sampling, that had only been experimented with previously, such as in 1920s crop sampling, a Civil Works Administration trial census and surveys of retail stores conducted in the 1930s, and an official sample survey of unemployment in 1940 that covered about 20,000 households. Sampling in the 1940 census allowed the addition of several questions for just 5 percent of the persons enumerated, without unduly increasing the overall burden on respondents and on data processing. Sampling also made it possible to publish preliminary returns 8 months ahead of the complete tabulations. The Census Bureau was able to increase the number of detailed tables published and review of the quality of the data processing was more efficient.

Most population and housing inquiries included in the 1940 census were repeated in later years, and a few were added, including

■ Place of work and means of transportation to work (1960).

Occupation 5 years before the census (1970 and 1980 only).

Housing costs (1980).

Inquiries relating to childcare by grandparents (2000).

In 1940 and 1950, the sample population questions were asked only for those persons whose names fell on the schedules' sample lines. Sampling was extended to the housing schedule in 1950, with a few questions asked on a cyclic basis: One pair of questions for household 1, another pair for household 2, etc., until household 6, when the cycle was started again with the first pair of questions.

In the 1960 census, the sampling pattern was changed for population and housing questions alike: If a housing unit

127

was in the sample, all of the household members were in the sample, too. This scheme yielded sufficient data for accurate estimates of population and housing characteristics for areas as small as a census tract (an average of 4,000 people). The only population questions asked on a 100-percent basis (name and address, age, sex, race, and since 1980, Hispanic origin, marital status, and relationship to householder), were those necessary to identify the population and avoid duplication.

The sampling pattern changed in later censuses. For 1970, some sample questions were asked of either a 15-percent or a 5-percent sample of households, but some were asked for both, thus constituting a 20-percent sample. There was no "split sample" for 1980, but it was used at every other household (50 percent) in places with fewer than 2,500 inhabitants and at every sixth household (17 percent) elsewhere. For Census 2000, the overall sampling rate was 1-in-6 households, but the actual sampling rate for any given geographic entity depended on its estimated population density and included four rates: 1-in-2, 1-in-4, 1-in-6, and 1-in-8 households.

Reflecting the concerns of the Depression years, the 1940 census asked several questions to measure employment and unemployment, internal migration, and income. It was also the first to include a census of housing that obtained a variety of facts on the general condition of the Nation's housing and the need for public housing programs. (Prior to this, the housing data collected as part of the population censuses were generally limited to one or two items.) At the time of the 1950 census, a survey of residential financing was conducted as a related, but separate, operation with information collected on a sample basis from owners of owner-occupied and rental properties and mortgage lenders. Similar surveys accompanied subsequent censuses. There also were surveys of components of housing change with the 1960, 1970, and 1980 censuses (but not 1990, when the survey was scheduled for 1989 and 1991). These measured the quantitative and qualitative impact of basic changes that occurred in the Nation's housing stock during the previous decade. The survey also offered a measure of "same" units, i.e., the preponderant part of the housing inventory that was not affected by the basic changes. (The first survey of this type had been a key part of the National Housing Inventory in 1956.)

Processing. The major innovation of the 1950 census was the use of an electronic computer-the Universal Automatic Computer I (UNIVAC I)-the first of a series delivered in 1951 to help tabulate some of the census data. Nearly all of the data processing was done by computer in the 1960 census. Beginning in 1960, census data were tabulated with the aid of the film optical sensing device for input to computers (FOSDIC)-an electronic device for "reading" the data on questionnaires. Special questionnaires were designed on which the answers could be indicated by marking small

128

circles. The completed questionnaires were photographed onto microfilm with automatic cameras. The FOSDIC then "read" the blackened dots (which appeared as clear holes on the negative film) and transferred the data they represented to magnetic tape to be processed by computer at speeds that ranged from 3,000 items a minute in 1960 to 70,000 items per minute by the time the 1990 census data were tabulated.9

Collecting the data. The 1960 census was the first in which population and housing data were extensively collected by mail. The field canvass was preceded by the delivery to every occupied housing unit of a questionnaire that contained the 100-percent questions (those asked for all persons and housing units). Householders were asked to complete the questionnaire and hold it until an enumerator visited.

The sample items were on a different questionnaire. In urban areas, containing about 80 percent of the Nation's population, the enumerators carried questionnaires containing the sample population and housing questions for every fourth housing unit. If the units were occupied, the householders were asked to fill out the sample questionnaires themselves and mail them to the census district office. The enumerators completed the questionnaires for vacant units. When these questionnaires were received in the district offices, the responses were transcribed to the special FOSDIC schedules. In rural areas, the enumerators obtained the sample information during their visits, and they recorded it directly on FOSDIC schedules.

The 1970 census marked the use nearly everywhere of separate, FOSDIC-readable household questionnaires-approximately 70 million of them-rather than the large schedules that contained information for four or more households. Thus, respondents could mark the appropriate answer circles on their questionnaires, which then could be processed directly without transcription.

Subsequent censuses were taken principally by mail-approximately 60 percent of the population in 1970, 90 percent in 1980, and 94 percent or more in 1990 and 2000. The questionnaires contained the 100-percent and, where appropriate, sample questions. In areas where the mailout/mailback procedure was used, enumerators contacted, either by telephone or personal visit, only those households that had not returned questionnaires or had given incomplete or inconsistent answers. For the remainder of the population, most of which was located in rural areas or small towns, postal carriers left a census form containing the 100-percent questions at each residential housing unit on their routes. An enumerator visited each of these households to collect the completed questionnaires

"The FOSDIC was replaced for Census 2000 with mark and optical character recognition (OMR and OCR) equipment.

Measuring America

and ask the additional questions for any household or housing unit in the sample. These procedures were continued, with modifications, for 1990 and 2000. In many rural areas, the enumerators, rather than the postal carriers, delivered the questionnaires and asked that they be completed and mailed back. In some inner-city areas, the enumerators took address lists with them, checked for additional units, and enumerated any persons they found living there.

Publishing. For 1970, extensive discussions with census data users led to a major increase in the amount of statistics tabulated, especially for small areas. As part of the 1970 census program, the Census Bureau published 100percent (but not sample) data for each of 1.5 million census blocks (including all blocks in urbanized areas), as compared with 1960, when block data were provided for 750,000 blocks within the city limits of places with 50,000 or more inhabitants. For 1980, there were data for 1.8 million blocks, with the population limit lowered to include incorporated places with populations of 10,000 or more; several states were "blocked" in their entirety. For 1990 and 2000, the block statistics program was expanded to cover the entire country, or approximately 7.5 million blocks.

The 1970, 1980, and 1990 population and housing census data appear in series of printed reports-either on paper or microfiche, or both-similar to those issued after the 1960 census, with accompanying maps, where appropriate. In addition, the U.S. Census Bureau issued public-use microdata tapes, usually containing much more detail than the printed reports, for users with computer access. After 1980, some data were made available on diskettes, online, through commercial computer networks, and by 1990, on CD-ROMs. Following Census 2000, data were released via the Internet, on CD-ROMs, computer-assisted "print-ondemand," and for some publications, the traditional printed volumes.

INDIVIDUAL HISTORIES OF THE UNITED STATES CENSUSES

The First Census: 1790

The first enumeration began on Monday, August 2, 1790, little more than a year after the inauguration of President Washington and shortly before the second session of the first Congress ended. The Congress assigned responsibility for the 1790 census to the marshals of the U.S. judicial districts under an act that, with minor modifications and extensions, governed census-taking through 1840. The law required that every household be visited and that completed census schedules be posted in "two of the most public places within [each jurisdiction], there to remain for the inspection of all concerned . . ." and that "the aggregate amount of each description of persons" for every district be transmitted to the President. The six inquiries in 1790 called for the name of the head of the family and the number of persons in each household of the following descriptions: Free White males of 16 years and upward (to assess

Measuring America

the country's industrial and military potential), free White males under 16 years, free White females, all other free persons (by sex and color), and slaves.

It is presumed that the Secretary of State (Thomas Jefferson), acting under the authority of the President, sent the marshals within each state, copies of the census act, and the required inquiries 10. The marshals then incorporated these inquiries into "schedules" of their own design.

The Second Census: 1800

A February 28, 1800, act provided for the taking of the second census of the United States, which included the states and territories northwest of the Ohio River and Mississippi Territory. The guidelines for the 1800 enumeration followed those of the first enumeration, except that the work was to be carried on under the direction of the Secretary of State. The enumeration was to begin, as in 1790, on the first Monday in August, and conclude in 9 calendar months. The marshals and secretaries were required to deposit the returns of their assistants, which were to be transmitted to the Secretary of State (not the President as in 1790), on or before September 1, 1801.

The Third Census: 1810

The third census, taken by the terms of an act of March 26, 1810, stipulated that the census was to be "an actual inquiry at every dwelling house, or of the head of every family within each district, and not otherwise" and commenced on the first Monday of August.

The results of the 1810 census were published in a 180page volume. Data for the population were presented by counties and towns in the northern sections of the country (except New York, which was by counties only), and in Ohio, Kentucky, and Georgia. The returns for the southern states were limited to counties. Territories were generally returned by counties and townships.

No additional details concerning the population were collected by the census; however, an act of May 1, 1810, required marshals, secretaries, and assistants to take (under the Secretary of the Treasury), "an account of the several manufacturing establishments and manufactures within their several districts, territories, and divisions." The marshals collected and transmitted these data to the Secretary of the Treasury at the same time as the results of the population enumeration were transmitted to the Secretary of State. No schedule was prescribed for the collection of industrial data and the nature of the inquiries were at the discretion of the Secretary of the Treasury.

An act of May 16, 1812, provided for the publication of a digest of manufactures containing data on the kind, quality,

10Carrol D. Wright and William C. Hunt, The History and Growth of the United States Census. Government Printing Office: Washington, DC, 1900. p.17.

129

and value of goods manufactured, the number of establishments, and the number of machines of various kinds used in certain classes of manufactures. The report, containing incomplete returns covering these items for more than 200 kinds of goods and included several items that were principally agricultural, was published in 1813.

The Fourth Census: 1820

The fourth census was taken under the provisions of an act of March 14, 1820. The enumeration began on the first Monday of August, and was scheduled to conclude within 6 calendar months; however, the time prescribed for completing the enumeration was extended to September 1, 1821. The 1820 census act required that enumeration should be by an actual inquiry at every dwelling house, or of the head of every family within each district.

As in 1810, the 1820 census attempted to collect industrial statistics. Data relating to manufactures were collected by the assistants, sent to the marshals, and then transmitted to the Secretary of State at the same time as the population returns. The report on manufactures presented the data for manufacturing establishments by counties, but the results were not summarized for each district and an aggregate statement was compiled as a result of incomplete returns. (The poor quality of manufacturing data was blamed partly on insufficient compensation for the collection of the data and the refusal of manufacturers to supply it).

The Fifth Census: 1830

Prior to the passage of the census act authorizing the fifth census in 1830, President Adams, in his fourth address to the U.S. Congress on December 28, 1828, suggested the census commence earlier in the year than August 1. He also proposed that the collection of age data should be extended from infancy, in intervals of 10 years, to the "utmost boundaries of life". These changes were incorporated into the census act of March 23, 1830. As in the previous census, the enumeration was made by an actual inquiry by the marshals or assistants at every dwelling house, or, as the law stated, by "personal" inquiry of the head of every family, and began on June 1 (instead of the first Monday of August as in previous censuses.) The assistants were required to transmit their returns to the marshals of their respective districts by December 1, 1830. Marshals filed these returns and the aggregate counts for their respective districts to the Secretary of State, by February 1, 1831. However, because of delays in the compilation of the census returns, the filing date was extended to August 1, 1831.

The 1830 census concerned the population only. No attempt was made to collect additional data on the Nation's manufactures and industry.

The Sixth Census: 1840

The sixth census was governed by the same general provisions of law as in 1830. Under the provisions of an act of March 3, 1839 (and amended by an act of February 26,

130

1840), the enumeration began on June 1, 1840. Marshals were to receive two copies of the census receipts from enumerators by November 1, 1840, one of which was to be sent to the Secretary of State by December 1, 1840. Again, as a result of delays, the deadlines for assistants and marshals were extended to May 1 and June 1, 1841, respectively. (The January 14, 1841 act extending these deadlines also provided for the re-enumeration of Montgomery County, Maryland, [due to discrepencies in the reports], to begin on June 1, 1841, and to be completed, with receipts. returned, by October 1, 1841.)

No population questionnaire was prescribed by the Congress-the design of the questionnaire was left to the discretion of the Secretary of State, and closely followed that used in 1830. The law did specify the inquiries to be made of each household.

The Seventh Census: 1850

In March 1849, Congress enacted a bill establishing a census board, whose membership consisted of the Secretary of State, the Attorney General, and the Postmaster General. This board was "to prepare and cause to be printed such forms and schedules as may be necessary for the full enumeration of the inhabitants of the United States; and also proper forms and schedules for collecting in statistical tables, under proper heads, such information as to mines, agriculture, commerce, manufactures, education, and other topics as will exhibit a full view of the pursuits, industry, education, and resources of the country."

The Congress also authorized the creation of the Department of the Interior in March 1849, and part of the enabling act provided that the Secretary of the Interior should "exercise all the supervisory and appellate powers now exercised by the Secretary of State in relation to all acts of marshals and others in taking and returning the census of the United States."

The seventh census was governed by the provisions of an act of May 23, 1850, which directed that six schedules be used to collect the information requested by the Congress. The enumeration began on June 1, 1850, and was to be completed, with the results returned to the Secretary of the Interior by November 1, 1850.

The Census Board prepared and printed six schedules for the 1850 census as follows:

■ Schedule No. 1 - Free Inhabitants.

■ Schedule No. 2 - Slave Inhabitants.

■ Schedule No. 3 - Mortality. This schedule collected dataincluding name, age, sex, color, and place of birth-on persons having died during the year ending June 1, 1850. Additional data were collected on constitutional and marital status; profession, occupation, or trade; month of death; disease or cause of death; number of days ill; and any suitable remarks.

Measuring America