History of Enumeration Procedures, 1790-1940
Diana L. Magnuson
This essay presents a short history of enumeration practices from 1790 to 1940. It aims to describe those aspects related to organization, personnel, and oversight that might have affected the outcome of the census as reflected in the manuscript returns. The essay does not cover the processing of the returns, only the elements that affected the manuscripts from which the modern census samples were created. The essay is divided into four parts: personnel recruitment and quality, training and oversight, public preparedness for the census, and mechanisms for correcting the final returns.
SELECTION AND QUALITY OF LOCAL ADMINISTRATORS AND CANVASSERS
The process of selecting local administrators and canvassers for the federal decennial censuses from 1790-1940 can be divided into two distinct periods. The first period, which lasted from 1790-1870, is distinctive because of the use of marshals and their appointed assistants to enumerate the population once every decade. Beginning in 1880, after several decades of lobbying for reform of the census machinery, marshals were replaced by supervisors and enumerators, initiating the second period of census taking.
Selection of Marshals, 1790-1870
The census machinery of the first six decennial censuses of population (1790-1840) largely reflected an infant federal bureaucracy. The most expedient means of enumerating the population was through the use of marshals of the judicial districts of the United States (which coincided with state lines). By law then, marshals were "authorized and required" to enumerate the U.S. population (omitting Indians not taxed). Marshals were further authorized "to appoint as many assistants, within their respective districts, as to them shall appear necessary," and to assign to each assistant "a certain division of his district, which division shall consist of one or more counties, cities, towns, townships, hundreds, or parishes, or of a territory, plainly and distinctly bounded by water courses, mountains, or public roads." Ideally, assistants were to be persons of "assiduous industry, active intelligence, pure integrity, great facility and accuracy of computation" with "an intimate knowledge of the division allotted to them."
Before commencing their duties, marshals and their appointed assistants each took an oath of affirmation before their district judge or justice of the peace. Failure on the part of the assistant to make the returns, or worse, to make false returns, resulted in a hefty fine. Before turning over the results of their enumeration, the assistant marshals were required to deposit a signed copy of the schedules "to be set up at two of the most public places" within his division for the inspection "of all concerned" for an unspecified amount of time. Non-compliance resulted in the forfeiture of his pay.
The marshals in turn were required by census law to file the returns of their assistants "with the clerks of their respective district courts." Marshals were fined the considerable penalty of $800 for failing to file the returns of their assistants with district court clerks or for failing to return the aggregate enumeration of each district to the President of the United States within the allotted time.
The census of 1850 marks the first time at which concern regarding the nature of the canvassers was discussed publicly. Up to this point, debate surrounding the census had focused almost entirely around the scope and content of the enumeration. Congress and the Secretaries of State had not expressed any concern regarding the effectiveness of marshals and their assistants in their job of enumerating the population.
By 1869 the complaints of former Census Superintendents Kennedy and DeBow were echoed with greater stridency by the Special Committee of the House of Representatives on the Ninth Census, a committee appointed to consider reform of the 1850 census act.
While the efforts of the Special Committee were ultimately defeated by Congress in 1870, several outcomes from the 1870 debate laid the foundations for change in 1880. First, the Special Committee's attempt to improve upon the 1850 census act did secure small changes which made the 1870 count more valuable than it otherwise may have been. These changes, however, were clearly stop-gap measures. Second, debate surrounding the proposed 1870 census act generated a considerable literature both inside and outside of Congress. Third, Francis Amasa Walker was appointed Superintendent of the 1870 census. Walker's experience as Superintendent in 1870 would put him in a better official position to argue for reform at the Tenth enumeration. As Superintendent of the Census, Walker, like Superintendent Kennedy before him, would "expose, with unsparing hand" the "barbarous" and "grossly defective" census of 1870 in his lengthy introduction to the population volume of the Ninth census. In Walker's introduction to the Compendium, he outlined what he considered to be four serious problems with enumeration procedure: supervision by United States marshals; the formation of subdivisions; the appointment of assistants; and the compensation of assistant marshals.
By 1880, three factors favored monumental changes like those suggested by the Special Committee in 1870. First, Congress was willing to listen to the concerns of the national and international statistical communities--communities which had grown in importance and stature since their early efforts to influence population data collection in the 1850s and 60s. Second, Walker, retaining his job as the Superintendent of the Census, was in a better position to argue for a massive overhaul of census taking (both strategically and experientially). And lastly, public outcry in the popular and scholarly press over that "wretched [census] machinery . . . of ours" became much sharper.
The first census hurdle Congress had to leap with regard to the 1880 enumeration was whether to pass a new census law. Most in the legislature were inclined to agree with the views of Senator Morrill of Vermont: " . . . let us unite in the passage of an act commensurate with our growth, that will not only illustrate our present condition and power, but point to that still more eminent rank and condition which we may hope to reach in ages yet to come." With regard to amending the 1850 census law, Walker perhaps said it best in his opening statement to the Committee of the Whole: "It will be utterly impossible to take a respectable census under the old law. We have outgrown it. We may as well clothe ourselves in the garments that we wore when we were small boys, and appear in them in public and consider ourselves well dressed, as to go on with the old methods and the old schedules which did very well in their time."
Selection of Census Supervisors, 1880
The use of specially-appointed supervisors, rather than judicial marshals, to direct census-taking below the national level is commonly cited as the outstanding innovation in the census act of 3 March 1879. Three justifications were put forward to support creating the new post of supervisor rather than relying on U.S. marshals. First, Walker claimed that the burden of work was too great to be adequately performed by officials who were already "crowded to the limits of their time and strength by prior official duties." A second advantage of census supervisors was that these new officials would be under the direct control of the Census Office and the Department of the Interior, while marshals were answerable to the Justice Department. Finally, supervisors could be selected whose qualifications were well suited to the particular requirements of administering the census.
The Superintendent had a sound point that those who supervised census-taking should not be burdened with other duties. Supervisors themselves adamantly stressed the time-consuming and demanding nature of their post. Given other, competing duties, marshals could hardly have devoted the time and care to overseeing the census that the supervisors claimed to have dedicated in 1880. Indeed, Walker claimed that a marshal had no choice but to "entrust the whole census work thus brought into his office to a deputy . . . who may be well chosen or ill-chosen for the purpose, [and] does the work anonymously and without any appreciable degree of official responsibility."
Whether the administrators of census-taking were under the control of the Department of Justice (as U.S. marshals) or the Department of the Interior (as census supervisors) may seem a minor detail. Nonetheless, the Superintendent argued convincingly that the power to appoint and remove supervisors was critical to the quality of the enumeration. The Census Office had no power to veto marshals' plans for dividing up territory, and several marshals had insisted "against the advice of the Census Office . . . on assigning to assistant marshals districts which could not possibly be canvassed in compliance with law in the prescribed time, the result being either the undue protracting of the enumeration, or else the illegal letting out of the work to unauthorized parties." The Journal of Social Science noted that the division of territory by marshals was "Perhaps the greatest of all the defects, or false provisions of the law," and that "under the pressure of local or congressional political influence, five marshals out of six will constitute their subdivisions, as a rule, of a size utterly inconsistent with a prompt and thorough enumeration." Similarly, lacking the power to veto enumerator appointments, the Census Office had only been able to express "its entire disapprobation" at the appointment of southern enumerators in 1870 "whose appointment was disgraceful to the government and detrimental to the service," leading to "mischievous and even scandalous results."
The qualifications and character of the supervisors obviously had an important effect on the quality of the 1880 enumeration, for those individuals had to "select, appoint, commission, instruct, supervise, and finally correct the work of . . . enumerators." Researchers today may wonder whether the men appointed to the post were indeed well qualified for "the special and highly technical work of the census." The pre-eminent concern for discussants at the time was whether partisan loyalties would lead these officials to falsify returns to affect apportionment, swelling the count in areas where their parties held sway and purging the rolls elsewhere. Among U.S. Congressmen, so intense were fears about politically-motivated doctoring of the returns that the legislators enacted provisions in the 1879 census law to limit partisan influence. In practice, however, political considerations were prominent in the selection of census supervisors.
Once the nomination and confirmation process was completed, observers concluded that the quality of the supervisors was high. Though highly critical of the confirmation process, the New York Times concluded, "Probably owing to the influence of Prof. Walker . . . the selections under the law have been made with as much intelligence and impartiality as could be obtained." The Census Office stated that "no difficulty was experienced in securing good men for the position," and Walker pronounced himself well satisfied with the supervisors' performance. "The very difficult and critical duties of that office [of supervisor] have been discharged, with but inconsiderable exceptions, in a manner most satisfactory. The zeal, energy, and prudence displayed by these officers, their provision against the accidents of enumeration, and their intelligent comprehension of the wants of their districts, entitle them to the highest commendation." Walker's successor, Charles Seaton, concurred that "The supervisors of 1880 . . . did that work in general with marked ability, high conscientiousness, and a laudable ambition to cause their respective districts to be fairly and well represented in the census." Even in the three cases where supervisors were removed from office, the causes for removal did not "reflect upon the personal or official integrity of the supervisor."
The main result of the partisan conflicts in the confirmation of census supervisors was delay in filling the posts. Although Superintendent Walker had compiled a list of nominees by mid-January, 1880, conflicts between Senators and the President led to some slots still remaining unfilled in late May. By early April, 132 of the 150 supervisors had been confirmed, so almost 90 percent of the supervisors had at least two months to complete preliminary census work. In a few areas, however, preparation time was short.
Whether delay marred enumeration quality depended on the compensatory actions of the supervisor and local citizenry. In Philadelphia, a supervisor who replaced his incompetent predecessor a week before the census and "was obliged to build the whole service from the ground up," found help from prominent volunteers. This expedient worked, for the paper later gloated, "Philadelphia was the last of all the great cities in being furnished with a census supervisor, but is the first in completing her enumeration; and thus it again comes to pass that the last shall be first." The last-minute replacement of a Louisville supervisor who had resigned had a less satisfactory outcome. That supervisor was "compelled to begin the work of enumerating without time to familiarize himself with the division of work in the city and without co-operation, suggestion, or advice." The results of the Louisville census evoked such dissatisfaction that the city's board of trade employed a statistician to oversee re-enumeration of one ward and petitioned Walker for a general recount.
Selection of Enumerators, 1880
One of the most important tasks of the census supervisors was the selection of enumerators. As discussed in detail below, the census law and Census Superintendent set general guidelines for these officials to follow in choosing canvassers. Nonetheless, press coverage of local census-taking indicates that supervisors and communities varied in their implementation of these guidelines. The following section examines two pieces of the process of selecting enumerators. First, the efforts of supervisors to solicit and encourage qualified applicants for the enumerator position are investigated. Second, three issues related to the appointment of enumerators--methods of soliciting enumerator applications, use of female enumerators, and handling of partisanship in enumerator appointments--are examined to illustrate the variation in practice within common general parameters. More broadly, this variation suggests that it is misleading to generalize about the quality and completeness of the enumeration on the basis of evidence from either the national office or a single community. While the national census office exercised far greater control and imposed more uniformity in practice in 1880 than in previous enumerations, the Tenth Census was still the end product of different practices carried out within 150 districts.
For the most part, supervisors had little difficulty in attracting applicants to enumerator posts in 1880. Rather than complaining of difficulties in filling posts, the supervisors resented the burden of processing the large number of applications and recommendations and the antagonism of disappointed candidates.
The only areas of the country suffering from a lack of qualified applicants were sparsely-settled regions, where "the work will be heavy and the pay light." In some of these locales, supervisors faced the choice of either placing less qualified applicants or hiring persons who lived outside the enumeration district and hence were less familiar with the people and territory.
The Superintendent of the Census offered supervisors some general guidelines in choosing enumerators: "The appointments should be made with reference to physical activity, and to aptness, neatness, and accuracy in writing and in the use of figures," to "active" and "energetic" young men "of good address." Walker warned, "Unless the officer appointed be fairly proficient in all clerical exercises, he will find his duties very trying and his pay very meager." And he favored those with previous experience as officials or in certain occupations. "Township assessors and other local officers" were "almost beyond the reach of error" due to their familiarity with "the names, residence, occupations, personal characteristics, and to a degree the history of the inhabitants"; postmasters at small offices had learned precision in filling out forms; country physicians appreciated "the value of reliable statistics" and understood "vital conditions . . . and the history of families"; and schoolteachers were "accustomed to keep lists and make reports."
Formal checks also encouraged the appointment of qualified enumerators. Supervisors forwarded their lists of names to Washington for review, and the Census Office could deny confirmation to incompetents and "political hotheads." Walker labeled the absence of such checks on the appointment of enumerators one of the main weaknesses of the 1870 census. The Washington Census Office did exercise its veto power on some enumerator appointments. Supervisors swore to discharge their duties in accordance with the census law (which demanded enumerators be chosen on the basis of competence), and were liable to two years' imprisonment for any dereliction of duty.
The most effective safeguard on the quality of enumerator appointments was a supervisor's awareness that hiring the unfit would reduce his community's count and incur the ire of the citizenry. Walker reinforced this point in his directive on enumerator selection: "If it is badly done, in any district, the service will be discredited, the district will be disparaged in the result, and the Supervisor will not escape blame." The Superintendent's warning was well-founded. In cities where local boosters' expectations of population increase were disappointed, even supervisors who had performed conscientiously found themselves reviled by newspapers and citizens' committees.
Whether these recommendations and sanctions were successful in securing qualified enumerators of course depended on the predilections of individual supervisors. Most big-city newspapers commended the quality of their city's canvassers, describing them as "carefully selected," "talented," "well fitted for performing the work with accuracy and dispatch," of "eminent fitness," and "good men" selected with "the greatest care." In St. Louis and St. Paul, the assessment changed after the population total proved disappointing. Such ex post facto condemnations, however, are more evidence of the degree of disappointment engendered by intercity rivalries than of the quality of the enumerators.
In the South, in particular, the qualifications and performance of enumerators were said to have improved in 1880. Unfortunately, such judgments were heavily tinged by racism. More important than changing racial composition was the shift from en masse to household enumeration. Many southern enumerators in 1870 "did not go out into the country and visit the homes of the people to make the enumeration, but made the count as people came into the towns to market or to court or muster." In 1880, "they went from cabin to cabin and did what the census laws require --paid personal visits to every place where it was likely that a person could find shelter." Whether because of greater completeness or rapid growth, the 1880 census count in the South was so much greater than the 1870 count that Northerners suspected politically-motivated fraud.
The attempts of supervisors to solicit and encourage qualified applicants for the enumerator position had mixed results at best. Formal checks encouraged the appointment of qualified enumerators, but in sparsely populated areas, supervisors had to rely on a relative scale of qualification.
Once supervisors got the attention of potential enumerators, it became a matter of appointing men, and in rare instances, women, to the position of census-taker. Three issues related to the appointment of enumerators are examined here: methods of soliciting enumerator applications, use of female enumerators, and handling of partisanship in enumerator appointments.
Evidence on the point is scanty, but the specific method of eliciting and judging enumerator applications seems to have varied across areas. For example, the supervisor for Atlanta publicized the positions through his local newspaper, but this was not done in any of the other newspapers examined in this essay. Such advertisements were not necessary to secure applicants. Supervisors from North Carolina claimed, "as soon as names were proposed for appointment of Supervisor, and before confirmation, applications for appointments [as enumerator] were daily poured in." The particular information to be solicited from applicants may have been left to supervisors' discretion. The Atlanta supervisor ordered hopefuls to make out applications in their own handwriting, to specify their occupation, address, and length of residence, and to include "testimonials as regards education, and assurances that the duties would be thoroughly and honestly performed." The supervisor in Philadelphia relied on advice from prominent businessmen and filled the posts by recruitment. In New York City, successful applicants were selected on the basis of "a personal examination as to the comparative fitness of each individual."
In the employment of female enumerators, the Superintendent explicitly directed supervisors to exercise their differing judgments. Acknowledging that the law did not expressly prohibit women enumerators, Walker stated, "It is clear that in many regions such appointments would be highly objectionable; but the Supervisor is not prepared to say that localities may not be found where a canvass of the population by women could be conducted without any disadvantage." In the cities of Atlanta, Philadelphia, Washington, and New York, between one and ten women were hired. In Baltimore, the supervisor declared that only one woman had applied for (and received) the job. "Several" women were employed as enumerators in St. Paul. In San Francisco, over 200 women sought positions as enumerators, but the supervisor hired none.
Supervisors also differed in their handling of partisanship in enumerator appointments. Driven by concern about fraud and the distribution of patronage, Congress had written into the 1879 census law the requirement that enumerators be "selected solely with reference to their fitness, and without reference to their political and party affiliations." (In the course of the debate, some legislators admitted this provision was unlikely to be honored). In his directions to supervisors, Francis Walker reinforced and elaborated on this point, and threatened to reject nominations of party extremists. Wrote the Superintendent, "It is hoped that so plain a provision of the law will command the cheerful obedience of all; but the Census Office will, if it shall appear to be necessary, insist, on its part, upon a full compliance, in good faith, with the requirement." In Walker's view, "the intention of the law is not to be carried out merely by appointing indifferently from the political parties which divide the country. The men selected . . . should be men so fair and moderate in their political feelings as to give an assurance" that they will "not pervert their trust to partisan purposes."
Even given "so plain a provision," supervisors differed in their method of achieving non-partisanship. The Atlanta supervisor ordered applicants to say nothing about "their political associations," and promised to remove any enumerator who "prostitutes his office to partisan uses." Until his removal, the first Philadelphia supervisor had intended to adopt the Congressional "division of the spoils" model. Reported the Philadelphia Public Ledger, "The agreement entered into at Washington, by which the Supervisors are equally divided between the two great political parties . . . will be continued in this city as to the matter of appointments by them . . . Half of these are to be Republicans and half Democrats; the city and county committees of the latter party to divide its quota of appointments equally." In Baltimore, only 18 out of 104 enumerators were Democrats, but "from most counties no applications were received from Democrats."
As the latter two examples indicate, political calculations as well as competence influenced the hiring of enumerators, the census act notwithstanding. This is hardly surprising. The supervisors--who made the selections--had to have political connections, albeit as party moderates. Would-be enumerators needed to supply testimonials and probably sought them from the socially and politically prominent. Those who could not produce such testimonials or who were not allied to the party of the supervisor would be less likely to apply, as in the Baltimore case. And the Superintendent's own directives reinforced the relevance of prior partisan experience. Non-partisanship, as Walker saw it, demanded considering the applicants' political views, if only to screen out "hotheads." By urging the hiring of those with previous experience as local officials (discussed below), Walker further encouraged the involvement of party activists. All of this really matters only if partisanship led either to the appointment of incompetents or to the filing of fraudulent returns.
Selection of Census Supervisors and Enumerators, 1890-1940
The process of selecting census supervisors and enumerators at the Eleventh through the Sixteenth Decennial Censuses changed less dramatically than it had in 1880. Small improvements or refinements marked the process with each succeeding census. Some things, however, did not change--the difficulty in securing qualified applicants, the suspicion of patronage in supervisor and enumerator appointments, and the continued quest for greater accuracy.
The 1890 census of population followed the model of supervisor and enumerator appointment devised in 1880. Yet what had been heralded as a laudable innovation in 1880, was already condemned by 1890. Civil service reformers in the latter decades of the nineteenth century wanted to see the majority of the census positions come under new civil service legislation. In other words, civil service reformers wanted the clerks, supervisors, and enumerators to take competitive examinations as a prerequisite to hiring. Reformers staunchly argued that as long as the census remained under the control of the spoils system it would be enumerated by incompetents, biased and hopelessly inaccurate.
The process of enumerator selection was significantly improved with the innovation of enumerator exams at the Twelfth Decennial Census (1900). The implementation of enumerator exams was more or less a compromise with civil service reformers. Whereas in 1880 and 1890 supervisors were allowed to make enumerator appointments based upon their own judgment of the character and competency of each enumerator, beginning in 1900 would-be enumerators were required to submit to a written examination. Persons seeking to fill the post of enumerator were required to submit an application and "evidence of the capacity . . . to perform the duties contemplated" in the form of a test schedule, "writ[ten] out in full." The test schedule was simply an exact copy of the population schedule mailed to each candidate and "filled out in hypothetical manner." In the privacy of their own home then, hopefuls filled out a sample schedule using family data presented in a narrative form. Upon completion of the examination, candidates returned the test schedule to their respective supervisors, complete with certification that they had not received any assistance in filling out the schedule. In 1900 some 300,000 persons filled out applications and test schedules in the hope of being one of the 53,173 persons chosen as enumerators. Nearly one-sixth of these candidates were rejected after the results of the test were tabulated.
Despite the new procedures for procuring enumerators, census officials continued to lament the difficulties in obtaining "properly trained men" for census fieldwork. The suspicion that supervisors "[took] advantage of the discretion which is allowed them, to prefer one candidate over another for political or personal reasons" added to the administration's discomfort over the non-competitive nature of appointments. Administrators were uncomfortably aware that data quality was directly correlated with enumerator "character and efficiency." However, in the first quarter of the twentieth century, no other method of selection and appointment seemed reasonable. The Director of the Census (formerly titled Superintendent of Census) could not possibly have the connections to nominate qualified supervisors for three-hundred districts; he had to rely on local (and unavoidably political) recommendations.
The selection process at the 1910 Census was similar to that carried out in 1900. Supervisors were required to apply for the positions and were theoretically chosen based upon their qualifications. As in 1900, persons seeking enumerator posts were required to submit an application to their respective supervisors and to take a competitive examination. Unlike the 1900 examination procedure however, prospective enumerators were not allowed to fill out the test schedule without supervision. While candidates in 1900 simply certified that they had filled out the schedule completely of their own accord, census officials were highly suspicious that many of them had received substantial assistance. In 1910, anyone wishing to be considered for an enumerator post was required to submit to a proctored examination. Prior to the test date, candidates were mailed census of population and census of agriculture schedules, together with enumeration instructions to be studied "carefully in advance." Then, on 5 February 1910, at over 5,300 locations across the country--usually a school house or other public building--more than 160,000 candidates submitted to an examination proctored by either a supervisor, civil service commission employee, or postmaster. The successful completion of the examination was not necessarily followed by an appointment. Supervisors were required to "give due regard to the relative excellence of the test papers in making their selection," but also to those qualifications for an enumerator "which can not be determined by a written test." Those qualifications to be considered by the supervisor included age, sex, character, habits, and "standing in the community." While supervisors were "expressly instructed" not to allow partisan politics to play a part in the selection of enumerators, it was generally acknowledged that such practices continued unabashedly. Thus, despite the gains in the method of obtaining qualified enumerators, the Census Bureau continued to seek "radical" improvements in the methods of selecting enumerators."
The Fourteenth Census Act (1920) departed from the legislation governing the appointment of supervisors from the previous four censuses. Under the census acts for the Tenth through the Thirteenth Censuses, it will be recalled, supervisors were appointed by the President, by and with the advice and consent of the Senate. The Fourteenth Census Act, however, stipulated that supervisors were to be appointed by the Secretary of Commerce, upon the recommendation of the Director of the Census. Under this arrangement the Director of the Census had more control over the supervisors and thus obtained better overall results. Further, this shift in the responsibility of supervisor appointments with the "advice and consent" of the Senate to the Director of the Census helped to speed up the selection process. The need for speedy appointments had often been noted by the Bureau, but in the case of the Fourteenth Census, it was especially necessary; the enumeration date had been moved ahead six months from 1 June to 1 January 1920.
The method of selecting competent enumerators in 1920 generally followed the practice devised in 1900 and 1910. Anyone interested in becoming an enumerator was issued an application blank, illustrative example of the enumeration schedule, and several other forms of information.
Those that applied were required to take a written exam at a predetermined examination site. Exams were held in public buildings, usually a school house, and tests were held two or three times during a day for several days. As in 1900 and 1910, the exams consisted of filling out a sample schedule of population from a narrative description of several families and, in rural districts, a sample agriculture schedule from information furnished. And again in 1920 the examinations were proctored by either supervisors, the supervisors' assistants or postmasters. In cases where applicants simply could not attend the written exam, the applicant was allowed to take the written exam at home or to arrange for an individual test date. According to the Director, the fact that in some cases the test was not competitive was not the issue, to secure "any applicants at all" was more important. Once supervisors had administered the written exams, they marked and rated them and then mailed them to the Census Bureau where they were examined "as fast as received." Supervisors were notified promptly as to the number of candidates approved by the Director.
From this list supervisors' made their appointments. Whenever possible, encouragement to apply and special preference in hiring was given to "honorably discharged soldiers, sailors, or marines who have seen service overseas, and whose test papers indicated that they are capable of doing the work."
The method employed for appointing supervisors and enumerators in 1920 remained virtually unchanged for the Fifteenth Census (1930). Again the Census Bureau sent out a press release soliciting the applications of would-be supervisors. Rather than considering every applicant a potential supervisor, the Census Bureau sent a copy of the general instructions for supervisors (a pamphlet of approximately seventy pages) along with a cover letter requesting confirmation that the applicant still wanted to be considered for the supervisor post. The Bureau added this extra step to the appointment process on the assumption that "in many instances persons apply for this position without much idea as to what the duties of the office are, and possibly with the impression that it is something in the nature of a sinecure." If the applicant still wanted the job, he or she was sent an application blank along with a test schedule. Both documents were to be filled out and returned to the Census Bureau. The Bureau then in turn carefully examined the completed population schedule and returned the corrected copy to the applicant for his or her information. If the applicant passed the test "satisfactorily" and "his record and credentials are likewise satisfactory" he or she was hired.
The solicitation and application process for would-be enumerators in 1930 followed that of 1920. The difficulty in obtaining qualified persons to apply for the post of enumerator was again an issue in 1930. Supervisors in districts where the number of applications were inadequate to fill the number of posts were responsible for actively recruiting persons to apply.
The only material change in the selection of enumerators in the Sixteenth Decennial Census was that an interview was used to supplement the application, test scores, and a subjective notion of competency, propriety and standing in the community. In 1920 and 1930, supervisors granted interviews of candidates only if the circumstances warranted it (i.e. the candidate could not attend the written examination). In 1940 district supervisors were encouraged to interview applicants "whenever possible" in order to facilitate an atmosphere whereby "applicants could speak freely about their qualifications." After the interview and written examination, each supervisor compiled a list of candidates whom they regarded as "reasonably expected to qualify" as enumerators. This list became an "eligibility" list from which applicants were "screened." The screening process eliminated "reasonably" qualified applicants who had poor handwriting and those who demonstrated an inability to follow written directions. Other ineligibles screened out through this process included persons not citizens ofthe United States; current or retired federal employees; employees of the Census Bureau who were engaged on other inquiries of the Bureau; employees of states, municipalities, and other local government bodies where federal employment was prohibited; or anyone under age 18. Once these less-than-desirable applicants were screened from the list, supervisors were instructed to rank applicants according to a hiring preference. Supervisors used this ranked list of applicants to select enumerators as well as substitute enumerators in case hired applicants were unable to complete their duties.
All qualified applicants (and then some) were sent a packet of enumeration instructions to carefully study before their written examination. Supervisors arranged for reasonably located testing sites and proctors and administered written exams based upon the instructions previously mailed enumerator hopefuls. After the examinations supervisors corrected the tests and notified the candidates of their scores.
TRAINING AND OVERSIGHT PROVIDED BY THE NATIONAL CENSUS OFFICE
The training and oversight provided by the national census office can be divided into three distinct periods from 1790 to 1940. The first period, characterized by minimal training and oversight, encompasses the first six decennial censuses of population. These enumerations were technically supervised by the President (as in 1790) or the Secretary of State (as in 1800-1840). Upon closer scrutiny, however, it becomes clear that minimal oversight of census work was carried out by whomever the secretary of state happened to appoint to the work. The second period, 1850-1870, was half as long as the first and characterized by vigorous lobbying for reform of training and oversight of canvassers on the part of those close to the census. These efforts for reform went unrequited until 1880, which marked the beginning of the third period.
The Tenth Decennial Census of population (1880) ushered in a new era of training and oversight of census-takers by the national census office. Beginning in 1880, and continuing through 1940, the training and oversight provided by the national Census Office became increasingly hierarchical, organized, and systematized. The organization of a permanent Census Bureau in 1902 (within the Department Commerce, also created in 1902) facilitated greater administrative control over population data collection.
These three broad periods of training and oversight provided by the national census office are illustrative of a growing sensitivity on the part of census administrators to the accuracy of the decennial enumeration. Generally speaking, census administrators in the early period of census-taking (1790-1840) seemed satisfied with merely completing and tabulating the enumeration for apportionment. Beginning in 1850, and growing in magnitude with each decennial census, administrators at the national census office became more attuned to issues of accuracy in data collection. It was generally acknowledged by those close to the census--such as census administrators, statistical experts, and even some members of Congress--that the census had grown far beyond its constitutional mandate of a simple "nose count." Thus the census, as a tool for gathering social statistics, needed to be precise for the multitudes of users of aggregate and tabulated data. The interested public was also increasingly demanding a complete and accurate count. The story of these attempts to improve accuracy are not only interesting in their own right, but are also relevant for present-day users of tabulated census data.
Marshal Training and Oversight, 1790-1840
Evidence of marshal training and oversight in the early period of American census-taking is scanty, in part because the effort to do either was minimal at best. The first decennial census was technically supervised by President George Washington, and census lore credits the president or the Secretary of State with providing copies of the census act and perhaps instructions to the marshals.
Beginning in 1800, and through the census of 1840, the decennial enumeration was supervised by the Secretary of State, a significant refinement in the administrative machinery of census-taking. According to Section 8 of the census acts, the secretary of state was "authorized and required" to "transmit" to all marshals "regulations and instructions" as well as "the forms contained therein of schedule to be returned, and proper interrogatories."
Marshal Training and Oversight, 1850-1870
The second period of training and oversight of the census fieldstaff by the national census office was characterized by a growing sense of inadequacy in the census machinery. The ensuing efforts to reform the machinery at the Seventh through Ninth Censuses, while unrealized, would lay the groundwork for major reform at the Tenth Census.
The training and supervision of marshals and their assistants by the nascent census administration in Washington took a decided turn for the better before the seventh census. The establishment of the Census Board and the creation of the Department of Interior (which shifted responsibility for the decennial enumeration away from the Secretary of State) in 1849 dramatically improved some aspects of census administration. First, marshals and their appointed assistants received informational circulars explaining each inquiry in addition to written enumeration instructions. Second, personal correspondence with the office of the Secretary of the Interior provided marshals and assistant marshals with a minimum of training and oversight. Superintendent Kennedy reported in 1852 that "The utmost care has been exercised to insure correct returns, and the manner of taking our Census has been calculated to effect such a result. . . . In all cases where error or inconsistency could be detected, real or imaginary, the individual has been written to in order that the discrepancy might be corrected." The actual procedure of canvassing the population, however, remained virtually unchanged from the first census. These supervisory innovations would remain the high point of enumerator training and oversight for another quarter century.
The frustrated attempt by the Special Committee to change the census machinery at the Ninth Decennial Census and thus improve the accuracy of the count did not preclude Superintendent of Census Walker from initiating some reform of the training and oversight of marshals and their assistants. Two refinements are of note. First, for the first time, assistant marshals were required to "report progress to the Marshal regularly once a fortnight." Second, the instruction manual that assistant marshals received in 1870 was considerably more detailed than that of 1850. The twenty-eight page manual detailed general responsibilities of assistant marshals.
Census Supervisors Districts, 1880
The Tenth Decennial Census of population ushered in a new era of training and oversight of census-takers by the national census office. Beginning in 1880, and continuing through 1940, the training and oversight provided by the national census office became increasingly systematic and bureaucratized.
One consequence of the change in census law at 1880 was the division of the country into specially-drawn supervisors districts, rather than reliance on existing judicial districts, as administrative units for overseeing census taking. Superintendent of Census Walker stressed that quite different criteria were salient in defining census administrative units and judicial districts. For example, Southern Florida was "very properly" made a judicial district" on account of the great facilities for smuggling and the frequent occurrence of maritime disasters." Its limited population of five to six thousand people made the same area inappropriate as an administrative unit for census-taking. In the same vein, Northern New York, "with two and a half millions," Western Pennsylvania, "with 1,750,000," and Idaho each constituted a judicial district.
Responsibility for dividing the country into 150 census districts--each with its own resident supervisor--lay with the Census Superintendent, contingent upon the approval of the Secretary of the Interior. Supervisors' districts did not cross state or territorial lines. Within states, the number and boundaries of districts were set on the basis of "the extent of territory, the compactness or sparseness of settlement, especially the occurrence of cities and large towns, the existing facilitates of transport and postal communication, the various constituents of the population, and the nature of the principal industries." Walker's plan, approved in November, 1879, created districts that were allegedly determined by "the exigencies of enumeration" but that varied considerably in extent and population." Reasonably enough, one supervisor directed census-taking in Wyoming Territory, where 30 enumerators counted twenty-one thousand people in 1880. Surprisingly, the supervision of 901 censustakers who enumerated 1.8 million residents of Massachusetts was also the task of a single individual. However uneven the administrative burden across supervisors' districts, the 1880 plan did divide the population more equally than under judicial districts and, by more than doubling the number of administrative districts, permitted "a minuter subdivision of the work while in progress."
Enumeration Districts, 1880
Another innovation which improved the quality of the Tenth Census, relative to earlier censuses, was the use of smaller and more clearly defined enumeration districts. For the 1870 census, enumeration districts were limited to 20,000 people. The 1879 census act cut the size down to 4,000 (as measured in the 1870 returns). A directive from Walker further restricted their size to 2,500 residents. In practice, enumeration districts for the Tenth Census were even smaller, encompassing 1,600 inhabitants on average.
The task of defining enumeration districts fell upon census supervisors, but the basic rules for setting the boundaries were set by Francis Walker. Enumeration districts were to maintain the territorial integrity of counties and other civil divisions (towns, townships, parishes, militia districts, hundreds, election districts, precincts, or beats). A civil division was to stand as an enumeration subdistrict if the population exceeded seven or eight hundred, but it could be divided to keep the population from exceeding 2,000 to 2,500. Clear boundaries like streams and roads were to mark off these divisions. And if a civil division were broken up, its subdivisions were to be of approximately equal population size. Supervisors' plans for defining enumeration districts and estimates of their population were forwarded to Walker, who sometimes demanded revisions.
Walker's specifications for subdistricting precluded a problem observed in the preceding census: "township divisions were not then recognized, and it is highly probable that many enumerators did not know where their county stopped and another began." If such confusion existed, enumerators would necessarily miss some areas and count others twice.
Limiting the population of enumeration districts to between 700 and 2,500 people ensured that the territory could be covered quickly. Providing for a larger number of smaller districts was a necessary complement to the shortening of the enumeration period, from the five months allowed in the 1850 census act to the two to four weeks specified in the census law of 1879.
While a shorter enumeration period was believed to increase the accuracy of the census, the resort to small enumeration districts had its own advantages. Enumerators were expected to reside in the districts they canvassed, and, in the case of such limited territory, to be familiar with the area and its inhabitants. Explained Walker, "The advantage to the Government of such close limitation of districts will be found in the high degree of local knowledge secured. The enumerator knowing, as will presumably be the case, every house and every family of the town of which he is a resident, will be placed almost beyond the danger of omissions, which are liable to occur in the canvass of larger districts, and will also be above being imposed upon by false statements which to a stranger might appear plausible enough." The provision also expanded the pool of qualified applicants by reducing the time and costs of service. Enumerators "working at short range through a district extending in no direction far from their places of residence, will not be obliged, as was generally the case under the Act of 1850, to fit themselves out expensively for traveling, [or] to close up their business or make arrangements for its being carried on by others in their absence."
However well justified the resort to smaller enumeration districts, supervisors found dividing up their territory to be "a work of some difficulty," requiring "much time, travel, and labor" or "a long and thoughtful correspondence."
Estimating the size of the population within small areas proved to be the most difficult element of districting. Ohio supervisors contended, "It would have been comparatively easy to do this by maps and charts had it not been necessary to a nearly equal division to ascertain the distribution of population within the precinct to be subdivided. The population of a precinct could be estimated by its vote, [but] not so with a part of a precinct. This was accomplished by correspondence and consultations with citizens from all parts of our territory." In some states, like Illinois, counties were not broken into townships, and the census of 1870 provided no indicator of the population size for alternative civil divisions.
Estimating the population of small areas was further complicated by the disturbingly permeable boundaries of civil divisions. Even county boundaries changed between 1870 and 1880. Illinois supervisors testified that voting precincts in their state "cannot be regarded as 'permanent civil divisions of the country,' as they are liable to be changed from year to year to serve the convenience of any neighborhood which may wish to be attached or detached for the purpose of bring brought nearer to the voting place, as to avoid a river, a swamp, or other obstacle."
The shifting and indistinguishable nature of some civil divisions may partially explain enumerators' failures to clearly mark the change from one civil division to another on their schedules, as required by the instructions to enumerators.
In urban areas, enumeration subdistricts were commonly political wards or voting precincts, but these sometimes exceeded the specified population limits. In an era when Western cities sometimes more than doubled in population in a decade, estimating the number of residents within parts of an urban area proved challenging. In some rapidly growing cities, "the task of districting for enumeration districts was not only arduous, but in some cases took longer than the enumeration."
Enumerator Training and Oversight, 1880
However carefully enumerators might be selected, the quality of their returns was shaped by the mechanisms in place to "instruct, supervise, and finally correct the work of . . . enumerators." The responsibility for defining and carrying out these tasks was split between the Washington Census Office and the district supervisors.
In early May, supervisors received from Washington advance samples of the schedules; in late May, the actual blanks arrived. The schedules were given to enumerators when they received their commissions and were sworn into office. For their guidance, enumerators also received a letter of instruction that spelled out pay rates in their district, commissions and oaths, a pamphlet containing "clear and minute" instructions on how to conduct themselves and fill out the schedules, a model completed schedule, and, in at least some districts, copies of the census law.
Because few records from the central Census Office have survived, it is impossible to reconstruct all the directions coming from Washington on the training and oversight of enumerators. It is clear that Walker was determined to change the situation from 1870, when enumerators "were entirely independent of me, and I had no control of their work." The Superintendent also required all enumerators to report each day, via a standard form on postal cards, to the central office in Washington and their district supervisor. These daily reports indicated the number of hours and minutes engaged on the service, and the number of persons, farms, manufacturing establishment, and deaths enumerated that day.
Whatever Walker's desire for uniform practice, press reports indicate that the details of oversight varied across localities. As a Missouri supervisor put it in a letter to the Superintendent, "each Supervisor devised such plans and resorted to such proper means as in his judgment would best subserve the interests of the Government and best promote the efficiency of the service in hand."
So widespread were efforts to check the quality of the first few days' fieldwork that Walker probably dictated this step to the supervisors. But supervisors carried out the check in different ways: in Baltimore, the supervisor ordered enumerators to report to his office; the Atlanta supervisor had enumerators send in their schedules; in Philadelphia, prominent citizen volunteers and a contingent from the city paper met with enumerators in local schoolhouses to check their work. By whatever the method adopted, in those cities for which information is available, this early fieldwork was pronounced satisfactory.
In some areas, other periodic checks on the enumeration were in place. Philadelphia was divided into ten districts overseen by "Commissioners." An Arkansas supervisor reported that he had sent to his enumerators "an average of ten letters and instructions." The District of Columbia supervisor claimed that enumerators were ordered to bring their work to his office every other day. Other supervisors employed positive incentives, like the Missouri man who awarded an honorary "diploma" to the enumerator in each county in his district who had done "the best work." Enumerators themselves reviewed and corrected their schedules during the enumeration process, being "frequently obliged to pass an hour or two in the evening in putting the lists which they have taken during the day in proper order." These corrections by the enumerators presumably entailed recopying illegible pages and filling in information on those who were found on a second visit.
Enumerator Training and Oversight, 1890-1940
Enumeration procedures put in place for the 1880 census continued to serve as a model for succeeding U.S. censuses well into the twentieth century. In other domains, the Census Bureau implemented major changes, such as introducing automatic tabulating (Hollerith) machinery to process the returns of the 1890 census. But in the areas of local administration, enumerator selection, and oversight of fieldwork by the Washington office, successive enumerations largely refined and elaborated methods introduced at the Tenth Census.
The 1890 census was taken under the model of training and oversight instituted by Francis Walker at the Tenth Census. The only discernible change from 1880 to 1890 in the training and oversight of fieldstaff was the creation of a "supervisors' correspondence" position within the Census Office (held by a single special agent) and an increase in the detail of enumerator instructions.
With a greater intensity than at previous censuses the Superintendent of the Census lamented the lack of a permanent census office to facilitate all aspects of the census work. It was agreed by all those close to the census and those outside the Office interested in census work, that the gains made in 1880 needed to be merely the beginning of procedural reform. One of the difficulties the Census Office faced in facilitating a reasonable degree of training and oversight of fieldstaff in the latter part of the nineteenth century was the inadequate amount of time given to procedural preparation. For example, the 1880 census act was not approved until 3 March 1879, and legislation affecting the work of the enumeration of the population was not passed until 20 April 1880, thus leaving only two months to prepare for the Tenth enumeration. In 1890 a similar problem was faced by the Census Office. The census law for the Eleventh enumeration was passed in early March of 1889, and legislation inserting additional inquiries on the population schedule was not passed until 22 February 1890, about three months before the census began. The Census Office had a mere ninety days to coordinate the printing (with the Government Printing Office) and the delivery (with the U.S. Post Office) of 20,000,000 schedules, as well as to print and deliver instructions, oaths, and the census law to over 46,000 enumerators. The pressing nature of this effort and the logistics involved in coordinating all of the pre-enumeration tasks left little time for supervisors to train enumerators and for enumerators to study the instructions and schedules.
An important innovation in the oversight of enumerators--the street book--was instituted at the 1900 census. The street book remained part of the enumerator's portfolio for succeeding censuses. The Census Office used the street book to aid in the enumeration of larger municipalities and to counter any complaints against the "completeness and the correctness" of the returns. The street book facilitated organization and completeness of the canvass on the part of the enumerator; canvassers used it to record "the number of families and persons found in each house or building," houses or buildings where no persons were found, and the date visited. The street book had another important purpose: supervisors used it to verify the completeness of each enumerator's canvass.
Just as in 1890, the 1900 decennial census of population increased the volume of instructions for facilitating training and oversight of supervisors and enumerators. According to a 1900 article on American census methods, through roughly a two month period, "The amount of material sent out to the field from the central office [was] . . . about two car-loads a day. In all, three hundred tons were sent out by registered mail." Included in this avalanche of material was an instruction booklet that each enumerator received from Washington via his or her supervisor. The instructions booklet was a whopping fifty-one pages, delineating everything from general instructions--the census act, care of schedules, enumerator's rights use of the telegraph--to special instructions regarding the schedules of population, agriculture, and manufactures.
The establishment of a permanent Census Office, renamed the Census Bureau (or the Bureau of the Census), in 1902, did not significantly alter enumeration procedures.
Hopes were high on the eve of the Thirteenth (1910) Decennial Census that this enumeration would represent the most accurate one ever produced by the Census Bureau. The permanent Census Bureau had been up and running for over six years. For the first time in the history of the census the administration in Washington believed it would have ample time to prepare for all aspects of the enumeration.
The 1910 decennial census administration refined the training process of census supervisors. In addition to "very full written and printed instructions" issued to each supervisor, conventions were held to train supervisors "in convenient cities in different parts of the country." At these conventions the Director or Assistant Director of the Census, accompanied by the chief statistician for population or the chief statistician for agriculture, gave oral instructions to the supervisors and "answered their numerous inquiries." The Census Bureau's initial enthusiasm over the length of time for preparing the Thirteenth Census was already waning in the fall of 1909. "The supervisors' districts were determined upon almost immediately after the passage of the census act, and a considerably number of supervisors were appointed by the President as early as August, 1909, and most of the remainder by September. Even though the appointments were made thus promptly, the supervisors had scarcely enough time to prepare satisfactorily for the enumeration of their respective districts." The desire of the Bureau to designate supervisors "at least one year in advance of the enumeration" was dashed for yet another enumeration, thus precluding a thorough preparation for the census of population.
The decennial difficulty in training and supervising a nationally dispersed and temporarily employed enumerator force, numbering over 71,000 in 1910, is evident in the 1907-08 report of the Director of the Census to the Secretary of Commerce and Labor. The Director lamented that "The enumerator is . . . the most troublesome problem connected with a decennial census." Enumerator training and oversight was refined for the 1910 canvass on several fronts. First, the pamphlet of instructions was "revised with great care." Second, as supervisors prepared for the Thirteenth enumeration, they were directed by the national Census Bureau to make arrangements to give oral instructions to as many enumerators as possible. It was reported after the enumeration that "In the larger cities practically all of the enumerators were thus assembled and given oral instructions either by the supervisor himself or by his special agents." Only in a few cases, however, did enumerators receive these oral instructions more than a day or two before the actual enumeration began. Finally, during the enumeration, arrangements were made in the large cities for "a continuous personal supervision and instruction" as work progressed. This "continuous personal supervision and instruction" was carried out by the supervisor and his special agents, called inspectors. Enumerators were expected to report to their inspector or supervisor within a day or two of the first day of work. The supervisor or inspector examined the day's work and "gave such instructions as were found necessary." It was the responsibility of the inspector to keep "constant" contact with the enumerators as the work progressed, "answering such questions as arose from time to time, checking the work of the enumerators at random, and otherwise assisting and directing them." In rural areas such supervision was a logistic impossibility, thus enumerators in the country were instructed to send a copy of "part or all of their first day's work" to the supervisor through the mail. Supervisors examined the schedules, corrected them if necessary, and returned them to enumerators.
While the census administration believed that the "actual number of inhabitants [was] ascertained with approximate accuracy" in 1910 and "that the principal interrogatories" on the population schedule were "answered with a fair degree of accuracy," it was the "less important inquiries" that had "less than satisfactory results." For example, Bureau administration calculated that enumerators in 1910 "failed to make any report at all regarding the citizenship of more than one-tenth of the foreign born males of voting age." Second, enumerators "failed to report the year of immigration of about one-tenth of the foreign-born population." Further, it was discovered by the Census Bureau upon tabulation of the veteran status question that the number of veterans reported was "very much smaller" than the number of veterans "actually on the pension rolls." Finally, the age statistics presented "numerous peculiarities" that could only be attributed to "a considerable margin of error in the returns."
The 1920 decennial census of population continued the process of refinement of the training and oversight of census fieldstaff. Schedules, forms, and instructions for supervisors, enumerators and special agents or inspectors were revised after "much time and thought." Continuing the trend toward greater contact between the Director of the Census and the supervisors, and thus ideally increasing the accuracy of the census, the volume of correspondence increased in 1920.
The innovation of supervisors' conferences, directed by the census administration, begun in 1910, was continued at the 1920 census of population. A total of eleven conferences were held across the United States, at which all but thirty-four of the 372 total supervisors attended. For those supervisors who had districts near Washington, their conference was held in the office of the Director of the Census. This conference was of a "general character" and was attended by the director, the assistant director, the chief statisticians for population and agriculture, the geographer, and the disbursing clerk. The remaining ten conferences were directed by the chief statistician for population. Presumably these conferences were similar in nature to those held in 1910, at which the Director or his chief statisticians gave oral instructions and answered the "numerous" inquiries of supervisors.
As in 1910, supervisors were directed by the Census Bureau to orally instruct enumerators whenever possible "in convenient numbers and at convenient points." Supervisors sent postal cards to twenty-five to fifty enumerators at a time, notifying them of their instruction location and time. The points of oral instruction included the legal obligations of enumerators and emphasized the importance of familiarity with enumeration instructions. Certain enumeration instructions were highlighted, including: who was to be enumerated; "carefully distinguish[ing] civil divisions; the "family" and "dwelling" distinction; year of immigration and citizenship; place of birth and mother tongue; and occupation and industry. The emphasis on these questions reflected concerns with their accuracy from the previous census. Apparently, the Census Bureau wanted to direct enumerator attention to these problematic questions to insure greater accuracy than had been the case in 1910.
Inspectors were again a factor in the oversight of enumerators in 1920. Inspectors received their instructions orally from their supervisor. "In general," anywhere from twenty-five to forty enumeration districts were put under the supervision of an inspector. These special agents were in essence an extension of the role of supervisors. In large cities supervisors were authorized to employ one or two special agents (inspectors) for the purpose of "exercis[ing] as close supervision as possible over the work of the individual enumerators" and "assist[ing] and instruct[ing] them for day to day . . . to make sure that they are performing their duties intelligently, industriously, and faithfully." It was "suggested" to enumerators that they report "in person" to the inspector in charge of their district within two days of the start of enumeration. To these meetings enumerators were requested to bring their schedules and street book for inspection, at which time they were also encouraged to ask questions about the work.
The Fifteenth Decennial Census of Population (1930) continued to refine and elaborate methods of training and oversight introduced at the Tenth Census. The number of supervisors increased while the number of enumerators actually decreased slightly. As in previous twentieth- century censuses, supervisors received advance instructions for carrying out the details connected with the enumeration. These instructions again took the form of pamphlets, supplemental written instructions in the form of letters, and oral instructions given at conferences. With only one exception all supervisors attended a conference.
The training and oversight of enumerators at the Fifteenth Census also elaborated on procedures implemented at previous censuses. Enumerators received a revised pamphlet of instructions, a record book (the old street book), and oral instruction from their respective supervisors. In 1930, inspectors were now called "field assistants," but served the same purpose of instructing and advising enumerators in the field.
Just as 1880 marks a turning point in the administrative machinery of the census at the local and national levels, so too does the census of 1940. The Sixteenth Decennial Census significantly expanded local administration and enumerator training and oversight.
The planning for the Sixteenth Census, including the population schedule, administrative machinery, the structure of the field organization, and post-enumeration processing and tabulation, all represent a radical departure from previous censuses. The Sixteenth Decennial Census significantly expanded local administration and enumerator training and oversight. All field activities in connection with the 1940 census were directed by the chief of the Field Division. Assisting the chief of the Field Division were three regional assistants--each representing one-third of the United States. Each region was then divided into 104 areas, which in turn were each administered by an area manager. The territory under the purview of each area manager was further divided into districts, each of which was headed by a district supervisor. The Sixteenth Census introduced the use of "squad leaders" to coordinate and assist enumerators. Squad leaders were used in cities of 50,000 or more population and were appointed for every twenty enumerators. Area managers, with the approval of the chief of the Field Division, appointed squad leaders. Interpreters were also used in districts or territories where a considerable number of non-English speaking people resided. Supervisors were to hire interpreters only in "extreme cases" where no persons could be hired as enumerators who spoke the particular foreign language. Interpreters were not regarded as enumerators, they simply accompanied enumerators on their rounds. Rounding out the field organization was of course the enumerator.
The training and oversight of fieldwork staff expanded significantly with the Sixteenth Census. In the words of the Chief Statistician of the Division of Statistical Research, " . . . we have gone further than previously in attempting to train employees for the various jobs they are to perform." Supervisors and enumerators were inundated with forms, instructions, memoranda supplements, guidebooks, and manuals. Workshops, training seminars, practice exercises, and constant correspondence were required of fieldwork staff at all levels. Area managers, district supervisors and enumerators all were given training programs varying from about two months for area managers to one or several days for enumerators. Training films, filmstrips, and other audiovisual aids were also used extensively. An expanded hierarchy meant more checks on procedure at all levels of fieldwork.
PUBLIC PREPAREDNESS FOR ENUMERATION
Evidence regarding the extent of census publicity is scanty for the first eight decennial censuses of population. Part of the reason for the paucity of census publicity during the early years of the federal census lies in the fact that daily newspapers and magazines were relatively rare and had a narrow readership until the middle of the nineteenth century. The technologies of the railroad and telegraph, changes in printing, and the development and growth of journalism began converging in the United States in the 1840s and 1850s. These developments significantly affected daily newspapers and magazines. Press coverage of the United States census to a large degree follows the growth of daily newspapers and magazines. The census did not become significantly newsworthy until the Tenth Census.
The appearance of an enumerator on the doorstep in June of 1880 could hardly have surprised those who read their local newspaper. Every one of the eleven city newspapers examined for this study in 1880 carried extensive coverage of the upcoming census, usually including the names and districts of the enumerators and the content of the population schedule. While some papers reported on the census without prompting, some of this coverage came at the petitioning of census supervisors, enumerators, and public-spirited citizens. Stories about the census in 1880 commonly included pleas for full public cooperation. Most newspapers even encouraged people to prepare by writing down their answers in advance.
The public relations campaign that city papers voluntarily carried on for the census in the last two decades of the nineteenth-century can partly explain the rarity with which enumerators encountered resistant respondents. Superintendent of the Tenth Census Francis Walker credited the importance of favorable press publicity enough to tell supervisors, "I think it would be quite the correct thing for you to make acknowledgments through the press of the aid which has been given you by the public-spirited cooperation of the press of the city."
In the twentieth century the Census Bureau became more sophisticated in its campaigns to publicize the enumeration and the work of the Bureau at each decennial census. The Bureau was acutely aware that "The completeness and accuracy of the census depend[ed] in no small measure upon the interest and intelligent cooperation of the people themselves." Yet the Bureau administration candidly admitted that "Considerable difficulty is experienced at every census in securing a complete enumeration and full answers to the interrogatories, by reason of indifference on the part of many people and distrust or fear on the part of a considerable number."This "distrust and fear" of a "considerable number" of people was "especially likely to appear among recent immigrants." Thus, "With a view to reducing these difficulties to a minimum," and facilitating "the interest and intelligent cooperation of the public," the Census Bureau undertook an "extensive" publicity campaign.
These publicity efforts consisted of a number of means of drawing attention to the census and educating the public. First, a proclamation by the President of the United States was considered "the most important feature" of the campaign. In 1910 proclamations were translated into twenty-four languages and hung by supervisors and their office assistants in post offices and other conspicuous locations. Second, "numerous" articles on the census were furnished by the Census Bureau to newspapers and magazines. Finally, circulars (brief memos of factual information for general circulation) were given to supervisors to distribute to school teachers, preachers, and employers, instructing them to inform those whom they came in contact regarding the census.
Another form of publicity used for the first time at the 1910 census, was the advance schedule. These schedules were distributed by enumerators a day or two prior to 15 April (the first day of enumeration) to heads of families. It was hoped that the advance schedule would "prepare the way" for the enumerator by "announcing his approaching visit" and informing the public of the questions to be answered. Secondly, the advance schedule would not only save time for the enumerators but also garner more accurate returns. The use of advance schedules was experimental, and they were only employed in cities of 100,000 or more inhabitants.
The decennial censuses of population for 1920, 1930 and 1940 built on the innovations of publicizing the enumeration. Supervisors of the census were instructed to publicize the work of the enumeration and to "keep the pot boiling" all during the enumeration period. In addition to the presidential proclamation (which continued to be issued in numerous languages), newspaper and magazine articles, and distribution of circulars and the advance schedule, the Census Bureau gave public addresses to various groups and organizations. Radio programs and "moving picture theaters" provided other mediums for the Census Bureau to publicize the coming census and to encourage the public to cooperate with the canvass.
The Sixteenth Decennial Census of Population received a media blitz unprecedented in census history. In August of 1939 a Division of Public Relations was organized "under the specific authority of Congress to plan and execute an educational campaign for enlisting Nation-wide cooperation" in the decennial census of population. The Bureau also employed the "consultative service" of "technical experts" on the various media of publicity. In the six months prior to the April enumeration, the Census Bureau "sow[ed] down the county" with folders, posters and proclamations and radio programs. More than 2,000 local committees sponsored by chambers of commerce and state, county, and municipal officials cooperated with the publicity campaign in their localities. School officials, church leaders, fraternal organizations, editors of newspapers and magazines, radio broadcasting chains and "other agencies and groups, too numerous to detail" also contributed to the 1940 census publicity campaign. In all, the Census Bureau spent over $90,000 on publicity for the 1940 census of population.
PROVISIONS FOR CORRECTION OF FINAL RETURNS
The returns canvassers had collected by the end of the enumeration period were subject to three general correction mechanisms throughout most of the decennial censuses through 1940 (the exceptions are noted). The first drew upon participation by the general public; the second was carried out by the marshal (as in 1790-1870) or the district supervisor (as in 1880-1940); and the third was completed in Washington by the national census office staff.
From the first decennial census of population the public was strongly encouraged to cooperate with the process of insuring a complete and accurate enumeration. The method of initiating public cooperation remained unchanged from the First through the Ninth Censuses. Beginning in 1880, and continuing through 1940, minor refinements in the mechanisms for public participation were implemented. At the first decennial census of population assistant marshals were required by the 1790 census act to post a copy of the returns "at two of the most public places" within his division to "remain for the inspection of all concerned." These exact directives were given to assistant marshals through the 1870 census. The 1880 census act refined this instruction by requiring enumerators to "give written advertisement at three or more public places in his district that he will be at the court house of said county on the fifth day after filing said list, not including Sunday, from nine o'clock ante meridian to six o'clock post meridian and the following day, for the purpose of correcting his enumeration by striking out or adding the designation of persons improperly enumerated or omitted." Until the Tenth Census, assistant marshals were paid a specified amount to recopy the names from their population schedules into lists for public inspection. In 1880, at a pay-rate of ten cents per one hundred names, enumerators recopied the names from their population schedules into an alphabetized ledger, together with information on each individual's age, sex, and race.
Beginning at the Tenth Census, those who believed themselves unenumerated were also invited to report to the supervisor's office by mail or in person. These opportunities were well publicized by the press.
The logic behind the provision to require public inspection of returns was threefold: to provide a check on politically-motivated fraud, to secure an additional copy of the returns lest any were lost by fire or other accident, and to record the unintentionally uncounted. While the first two goals may have been met, the results of the third remain questionable.
The second mechanism for correcting final returns was carried out by the marshal (as in 1790-1870) or the district supervisor (as in 1880-1940). After sitting in a conspicuous public place to amend their schedules, canvassers turned their work over to their superior, for a second round of corrections. At the first six decennial censuses of population, assistant marshals turned their work over to the marshals who in turn filed them with the clerks in their respective district courts. No mention is made in the census law of the expectation by the Census Office that marshals were to scrutinize the work of their assistants before filing the returns with the district court clerk. It was merely the responsibility of marshals to file the returns and mail the aggregate figures back to Washington. The only check on the correctness of the returns for the decennial censuses of population from 1790 to 1840 were the penalties for false statements and the oaths required of assistants marshals and marshals.
The 1850 census implemented formal checks on assistant marshal work by their respective marshals. The instructions to marshals specifically directed that it was their duty to "carefully examine whether the returns of each assistant marshal be made in conformity" with the 1850 census law and "where discrepancies are detected, require the same to be corrected." The instructions to assistant marshals in 1870, as written by Superintendent of Census Francis Amasa Walker, were even more pointed in their instruction regarding provisions for correcting final returns. Under the "General Directions" heading Walker instructed assistant marshals to read over entries made "to the party giving the information, that all mistakes may be corrected on the spot, at the time." Assistant marshals were further directed to take "great pains" in comparing the second copy intended for the Census Office and the original, "point by point." Finally, the instructions emphasized that "at the end of each set of returns" the assistant marshal was to "certify that they were made according to law and instructions." The refinements installed by Walker were small indeed, but foreshadowed the significant change he would initiate at the Tenth Census.
Beginning in 1880, enumerators turned their completed schedules over to their district supervisor for a second round of corrections. In the supervisor's office, the returns underwent "a thorough revision and correction" for two to four weeks. Supervisors were charged by census law to "examine and scrutinize the returns of enumerators" to "ascertain" whether the work was "performed in all respects in compliance with the provisions of the law, and whether any town or village or integral portion of the district [was] omitted from enumeration."
Also checked, counted, and recorded were errors of form in filling out the various schedules. A supervisor had authority to compel re-enumeration of neglected or mis-enumerated territory and could refuse to counter-sign the form an enumerator needed to secure payment from Washington.
Both the 1880 census and the supervisors' subdistricting plans provided the staff with yardsticks to use in checking for gross over- or under-counting. The central office had also calculated, and sent to supervisors, figures on the average number of people enumerated per day in the rural and urban sections of their state in 1870. These provided another potential yardstick for checking individual enumerators' returns. In the event of too great a disparity, the Superintendent could demand re-enumeration of an entire city or census district.
Supervisors continued to be responsible for the "scrutiny" of enumerator work at succeeding censuses, and, in the event that "discrepancies or deficiencies" came to light, they were also responsible "to use all diligence in causing" the returns to be corrected. Beginning in 1920 supervisors received a specific set of detailed instructions (a seven page Bureau publication) regarding the examination and correction of schedules. These instructions gave three directives with regard to the examination of the population schedules. First, supervisors were required to check that the canvass was thoroughly completed by each enumerator. Again the effectiveness of this correction mechanism depended upon the extent of the supervisor's familiarity with the territory. Second, supervisors were instructed to carefully examine the schedule inquiries for correctness and completeness. Lastly, it was the duty of supervisors to require enumerators with faulty returns to correct "without delay" the schedules, whether the issue was incomplete returns or erroneous queries. In the case of districts which were "too faulty to admit of their proper correction" by the enumerator, the supervisor was required to write up a full report to the Director.
Up until the 1930 Decennial Census of Population, the returns were forwarded to Washington where they were counted and the totals announced. Beginning with the Fifteenth Census, however, supervisors were required to compute a preliminary tabulation and announce their local population "as rapidly as the enumeration is finished." In the past, the "considerable lapse of time" between the canvass and the publication of the returns caused two problems for the Census Bureau which ultimately hindered a relatively complete and thorough canvass. First, complaints of "defects" in the enumeration by individuals and municipalities could not be addressed promptly. Second, "in many cases" the enumerator responsible for the work had already been paid and was "long gone." So too with district supervisors: in some cases the supervisor "was not available to assist the bureau in the verification" of his or her work; and in other cases he or she was "disinclined" to acknowledge responsibility for the work. Thus it was hoped that by allowing supervisors to publish preliminary population totals that local publicity would afford the Census Bureau the opportunity to enlist the aid of the public, address canvassing problems as they arose, and utilize enumerators and supervisors before the work disbanded.
The third and final setting for corrections was the national Census Office in Washington. The efforts to install mechanisms to correct final returns of the census of population in Washington were halting at best. In keeping with the minimalist approach that governed all early efforts at census-taking, the first four decennial censuses of population were published "exactly as they were received from the marshal, without any attempt to correct them or to make them uniform." This state of affairs is not surprising, given that a Census Office, per se, did not exist in Washington until 1830. Throughout the history of the census the work of collection, correction and tabulation was far more than the administrative machinery could handle. Even after the establishment of a permanent Census Bureau in 1902, the national census office continued to struggle with the editing of aggregate returns.
It was not until the Fifth Decennial Census (1830) that the Secretary of State was authorized to "note all the clerical errors in the returns of the marshals and assistants, whether in the additions, classification of inhabitants, or otherwise, and cause said notes to be printed with the aggregate returns of the marshals, for the use of Congress." Forty-three clerks were temporarily employed in the office of the Secretary of State for the purpose of revising the returns of the Fifth Census. That this law was carried out is illustrated by Secretary of State Edward Livingston's two presentations to Congress regarding corrections to the Fifth Census.
Again in 1840 the Secretary of State was directed to "note all the clerical errors in the returns of the marshals and assistants" and to employ the necessary clerks. A refinement to the 1830 census law gave the Secretary of State additional authorization to employ several professional clerks: a "superintending clerk" at $1500 per annum; a "recording clerk" at $800 per annum; an "assistant clerk" at $650 per annum; and "packer and folder" at $650 per annum. The appointed superintending clerk, William A. Weaver, was in turn directed by census law to employ temporary clerks for the work of correcting and tabulating census returns. Weaver's qualifications and commitment to census work are suspect, and it is not surprising that not long after the 1840 enumeration was completed and copies of the totals distributed to all interested parties, inaccuracies in the count surfaced. While Weaver is not entirely responsible for the problems in the 1840 count (the inept census machinery which employed marshals and their assistants to canvass the population can take some of the blame), his unwieldy population schedule and less-than-Puritan work ethic surely were contributing factors. The problems encountered at the national level for correcting and tabulating the final census returns provided the impetus for significant refinement of correction mechanisms in Washington.
The creation of the Department of the Interior in 1849 transferred all "supervisory and appellate powers" with regard to "the taking and making returns of census" from the Secretary of State to the Secretary of the Interior. The significance of this shift is noted in W. Stull Holt's census history as the first "notable advance" in which "a new level of attainment in scope, accuracy and administrative methods" was achieved. Under the umbrella of the Department of the Interior then, the Census Office began to develop into a bureaucratic enterprise. Two important changes occurred in 1850 which impacted correction mechanisms at the national level. First, the forty-five marshals employed at the Seventh Census were required to correspond with the office of the Secretary of the Interior in cases of error or inconsistency in their returns. The volume of correspondence between marshals and the office of the Secretary of Interior attests to the concerted effort with which the Census Office sought to secure an accurate census. Second, increased responsibilities of the Census Office necessitated the employment of a larger temporary clerical force in Washington.
The growth of the Census Office in Washington and its temporary nature posed a particular set of problems with respect to the correction and tabulation of census returns. The most troubling problem was the inexperience of most of the clerks. Superintendent of the Seventh Census J.D.B. DeBow explained that "There is a peculiar education required for these labors which neither comes from zeal or genius, but is the result only of experience." The clerks employed at the Seventh Census had no such experience; in fact, "such was the pressure of work, that almost any one could at times have had a desk." According to DeBow, all stages of census taking, from planning to correction and tabulation, had "taken care of itself." DeBow elaborated: "Every ten years some one at Washington will enter the hall of a department, appoint fifty or a hundred persons under him, who, perhaps, have never compiled a table before, and are incapable of combining a column of figures correctly. Hundreds of thousands of pages of returns are placed in the hands of such persons to be digested. If any are qualified it is no merit of the system." By the time clerks were trained and had acquired "familiarity with statistics" the Census Office was disbanded until it came time to prepare for the next census. Despite these difficulties, the Census Office staff continued to grow. At the Tenth Census Superintendent Walker required for the first time that would-be clerks submit to an examination to test their qualifications. The Census Office in Washington would continue to struggle, however, with the difficulties a temporary office force posed in correcting and tabulating the final returns until the establishment of a permanent office in 1902.
The establishment of the permanent Census Bureau facilitated the development of systematic correction mechanisms that were refined throughout the twentieth century. At the most basic level, the returns were analyzed, corrected if necessary, and then tabulated. Specifically, after the returns of the enumeration of population were received by the Census Bureau, "a number of experienced clerks" set to the work of "critically examin[ing]: the schedules. The manuscript population schedules were carefully analyzed for the tell-tale signs of padding. Clerks were instructed "to watch for enumeration districts in which unusually large numbers of persons were enumerated, or in which exceptionally large numbers of boarders and lodgers were reported, or in which there were exceptionally large numbers of families of unusual size." The clerks kept "detailed memoranda" on the results of their analysis of "suspicious" returns. In instances that were deemed questionable, "expert employees" were sent out from Washington to investigate. According to the Census Bureau, no corrections were made in Washington without first carefully investigating on-sight the city or district in question.
As alluded to above, of all types of census errors, unintentional underenumeration was least effectively addressed by administrative oversight. At best, the mechanisms of supervision and correction instituted at the 1880 census (and continuing with refinements through the 1940 census) precluded the practice of assistant marshals compiling their lists at public meetings instead of household visits. From drawing up enumeration subdistricts, a supervisor had a rough idea of population size and was likely to notice if the count were substantially lower. The review of returns by supervisors and their clerical assistants, and the innovation of the enumerator street book in 1910, made neglecting an entire section of territory or entering names out of order of residential propinquity somewhat risky. But the enumerator who failed to revisit or query neighbors about temporarily empty dwellings, to secure a translator, or to hunt down those living in business quarters or "obscure rookeries" would not be caught out. Securing a relatively complete count thus depended largely on pay incentives for enumerators, public cooperation, and the vigilance of well-chosen canvassers.