Public Health Then and Now
History Counts: A Comparative Analysis of Racial/Color Categorization in US and Brazilian Censuses
ABSTRACT Melissa Nobles, PhD
Categories of race (ethnicity, color, State officials, international organizations, sessing the contemporary production and uses or both) have appeared and continue to ap- and the general public today consider census- of racial data and, in particular, the recent de- pear in the demographic censuses of nu- taking an indispensable component of modern cision by the US Office of Management and merous countries, including the United governance. Most national census bureaus em- Budget (OMB) to allow Americans to check States and Brazil. Until recently, such cat- ploy similar statistical methods and adminis- more than one racial category. egorization had largely escaped critical trative procedures, and international guidelines scrutiny, being viewed and treated as a tech- have advanced this uniformity. Since 1946, for nical procedure requiring little conceptual example, the United Nations has sponsored 4 American Censuses: Race is clarity or historical explanation. Recent world population programs whose express pur- Fundamental political developments and methodologi- pose was to improve and standardize national cal changes, in US censuses especially, censuses.1 The race question and racial categories have engendered a critical reexamination An unavoidable consequence of this ap- have appeared in every US decennial census, of both the comparative and the historical parent uniformity and universality has been from the Republic’s first in 1790 to the 2000 dimensions of categorization. the obfuscation of the very particular dimen- census. Although the term “color” actually ap- The author presents a comparative sions of census-taking within and between peared on 19th-century census schedules, it analysis of the histories of racial/color countries. That is, while census bureaus may was synonymous with “race” in meaning. The categorization in American and Brazilian use the same statistical methods to produce nu- history of racial categorization can be divided censuses and shows that racial (and merical data, they often do not employ the same into 4 periods (see Table 1). color) categories have appeared in these categories (e.g., race, language, ethnicity) to The first period is 1790 through 1840, censuses because of shifting ideas about generate these data. Moreover, even if the same when categorization was shaped by represen- race and the enduring power of these categories are used, they usually do not bear tational apportionment, slavery, and racial ideas as organizers of political, eco- the same meanings. Indeed, the basic incon- ideas. The second period is 1850 through 1920, nomic, and social life in both countries. sistency of official categories and their mean- when categorization was used expressly to ad- These categories have not appeared sim- ings can even be observed within the census vance the racial theories of scientists. The third ply as demographic markers. The author history of one country. In American censuses, period is 1930 through 1960, when census def- demonstrates that censuses are instru- for example, census and other state officials initions of racial categories were identical to ments at a state’s disposal and are not have changed racial categories and their defi- those of Southern race laws. The fourth period simply detached registers of population nitions several times since the first census in is 1970 to the present, during which categori- and performance. (Am J Public Health. 1790 and on nearly every census since 1890. zation has been shaped most profoundly by 2000;90:1738–1745) In short, the political impulses behind cen- civil rights legislation, the implementation of sus categories vary across national settings and OMB Statistical Directive No. 15, and the lob- within national settings across time. Whether bying efforts of organized groups. Before the the terms used are “race,” “ethnicity,” “color,” introduction of self-identification on the 1960 or some combination of these depends largely census, enumerators determined the person’s on historical circumstance. What these terms race by observation on the basis of the defini- mean, to whom they apply, and how they are tions provided in official instructions. employed in public policies are most intelligi- ble in terms of specific national experiences. In this article I provide a succinct history and analysis of racial and color categorization The author is with the Department of Political Sci- in American and Brazilian demographic cen- ence, Massachusetts Institute of Technology, Cam- suses.2 This history clearly illustrates the fun- bridge, Mass. damental political and historical contingency of Requests for reprints should be sent to Melissa such categorization. It also shows that the pro- Nobles, PhD, Department of Political Science, Mas- sachusetts Institute of Technology, 77 Massachusetts duction of racial/ethnic data has served shift- Ave, E53-453, Cambridge, MA 02139 (e-mail: ing political and social purposes. Knowledge [email protected]). of this history is absolutely essential to as- This article was accepted June 30, 2000.
1738 American Journal of Public Health November 2000, Vol. 90, No. 11 the American Revolution competed, at the time, TABLE 1—US Census Race Categories, 1790–2000 with other ideas that claimed natural hierar- 1790—Free White Males; Free White Females; All Other Free Persons; Slaves chies and limitations. Further, the deepening 1800—Free White Males; Free White Females; All Other Free Persons, except Indians Not entrenchment of slavery in economic and po- Taxed; Slaves litical life rendered moot abstract commitments 1810—Free White Males; Free White Females; All Other Free Persons; except Indians Not to universal equality and liberty. Taxed; Slaves To be free and White and to be free and 1820—Free White Males; Free White Females; Free Colored Persons, All other persons, except Indians Not Taxed; Slaves Black were distinctly different political expe- 1830—Free White Persons; Free Colored Persons; Slaves riences. Whites were presumptively citizens. 1840—Free White Persons; Free Colored Persons; Slaves Although free Blacks were also citizens by a 1850—Black; Mulatto birthright, they did not enjoy the same rights 1860—Black; Mulatto; (Indian)b 1880—White; Black; Mulatto; Chinese; Indian and entitlements as Whites, precisely because 1890—White; Black; Mulatto; Quadroon; Octoroon; Chinese; Japanese; Indian Blacks were deemed inferior and unfit for re- 1900—White: Black; Chinese; Japanese; Indian publican life on the grounds of race.4,5 The cit- 1910—White; Black; Mulatto; Chinese; Japanese; Indian; Other (+ write in) izenship status of Native Americans was de- 1920—White; Black; Mulatto; Indian; Chinese; Japanese; Filipino; Hindu; Korean; Other termined by the particular status of tribes as (+ write in) 1930—White; Negro; Mexican; Indian; Chinese; Japanese; Filipino; Hindu; Korean; (Other spelled out in law and treaties. The federal gov- races, spell out in full) ernment considered most tribes “quasi- 1940—White; Negro; Indian; Chinese; Japanese; Filipino; Hindu; Korean; (Other races, sovereign nations,” thereby disqualifying their spell out in full) members from American citizenship.6 The cen- 1950—White; Negro; Indian; Japanese; Chinese; Filipino; (Other race—spell out) 1960—White; Negro; American Indian; Japanese; Chinese; Filipino; Hawaiian; sus schedules of 1800 through 1820 explicitly Part-Hawaiian; Aleut Eskimo, etc. reflected these arrangements in their category 1970—White; Negro or Black; American Indian; Japanese; Chinese; Filipino; Hawaiian; “all other free persons, except Indians Not Korean; Other (print race) Taxed.” 1980—White; Negro or Black; Japanese; Chinese; Filipino; Korean; Vietnamese; American The censuses from the years 1790 through Indian; Asian Indian; Hawaiian; Guamanian; Samoan; Eskimo; Aleut; Other (specify) 1990—White; Black or Negro; American Indian; Eskimo; Aleut; Chinese; Filipino; Hawaiian; 1840 asked few questions beyond those related Korean; Vietnamese; Japanese; Asian Indian; Samoan; Guamanian; Other API to population. They counted free White males (Asian or Pacific Islander); Other race and free White females, subdivided into age 2000—White; Black, African American, or Negro; American Indian or Alaska Native; Asian groups; free colored persons (in some years, Indian; Chinese; Filipino; Japanese; Korean; Vietnamese; Native Hawaiian; Guamanian or Chamorro; Samoan; Other Asian (Print Race); Other Pacific Islander all other free persons, except Indians not taxed); (Print Race); Some other race (Print Race) and slaves. The earliest censuses registered race, as it was then understood. Race was con- Note. Categories are presented in the order in which they appeared on schedules. sidered a natural fact, though its political and Source. US Bureau of the Census. social significance was still being sorted out. aIn 1850 and 1860, free persons were enumerated on schedules for “free inhabitants”; slaves were enumerated on schedules designated for “slave inhabitants.” On the free- To be sure, colonial racial discourse had long inhabitants schedule, instructions to enumerators read, in part: “In all cases where the regarded Africans as different from and inferior person is white leave the space blank in the column marked ‘Color.’” to the English, whatever their common hu- b Although “Indian” was not listed on the census schedule, the instructions read: manity.6Yet political elites did not regard these “ ‘Indians’—Indians not taxed are not to be enumerated. The families of Indians who have renounced tribal rule, and who under State or Territorial laws exercise the rights of differences as permanent. By the 1850s, in this citizens, are to be enumerated. In all such cases write ‘Ind.’ opposite their names, in respect, racial discourse had changed mark- column 6, under heading ‘Color.’” edly. So too would the role of census-taking.
1850–1920 Censuses
The 1850 census marked a watershed in 1790–1840 Censuses free or slave and whether one was taxed or not. census-taking in several ways. For our pur- The race question was included because race poses, a large part of its significance rests in the The initial reasons race appeared at all are was a salient social and political category. introduction of the “mulatto” category and the not transparently connected to demographic Eighteenth-century political elites regarded reasons for its introduction. This category was concerns, because the principal impetus for race as a natural and self-evident component of added not because of demographic shifts, but US census-taking was political. The Constitu- human identity, in keeping with European En- because of the lobbying efforts of race scien- tion of the United States mandated that “an ac- lightenment thought. Observed differences in tists and the willingness of certain senators to tual enumeration” be conducted every 10 years physical appearance and cultural practices were do their bidding. More generally, the mulatto for the purposes of representational appor- the result of differences in natural environ- category signaled the ascendance of scientific tionment. How slaves would be counted was es- ment.3 authority within racial discourse. pecially contentious. Delegates to the Consti- These ideas, most robust in the years im- By the 1850s, polygenist thought was tutional Convention eventually agreed on the mediately preceding and following the Amer- winning a battle that it had lost in Europe. The three-fifths compromise, meaning that for ap- ican Revolution, were gradually subordinated “American school of ethnology” distinguished portionment purposes, a slave would count as to theories of polygenesis and the widespread itself from prevailing European racial thought three fifths of a person. belief in the existence of innately and perma- through its insistence that human races were The question remains, then: Why did the nently superior and inferior races. It is impor- distinct and unequal species.7,8 That polyge- census count by race? After all, representation tant to emphasize, however, that the egalitarian nism endured at all was a victory, since the depended on civil status—whether one was ideas emerging from the Enlightenment and very existence of racially mixed persons had led
November 2000, Vol. 90, No. 11 American Journal of Public Health 1739 European theorists to abandon it. Moreover, Both the 1870 and 1880 censuses were Congressional documents and enumer- there was considerable resistance to it in the designed to amass statistical proof for this the- ator instructions for the 1890 census again re- United States. Although most American mo- ory, as enumerator instructions reveal. Enu- veal scientific interest in the census. Bureau nogenists were not racial egalitarians, they were merators were expected to determine, through officials and social scientists wanted to know initially unwilling to accept claims of separate visual inspection, the traces of African blood in “[w]hether the mulattoes, quadroons, and oc- origins, permanent racial differences, and the persons counted. The 1870 instructions read: toroons are disappearing and the race be- 14 infertility of racial mixture. Polygenists delib- It must be assumed that, where nothing is writ- coming more purely negro.” Therefore, erately sought hard statistical data to prove that ten in this column, ‘White’ is to be understood. “quadroon” and “octoroon” were added to the mulattoes, as hybrids of different racial species, The column is always to be filled. Be particularly categories “White,” “Black,” “Mulatto,” “Chi- were less fertile than their pure-race parents careful in reporting the class Mulatto. The word nese,” “Japanese,” and “Indian.” The instruc- here is generic, and includes quadroons, octoroons, and lived shorter lives. and all persons having any perceptible trace of tions read: Racial theorist, medical doctor, scientist, African blood. Important scientific results depend Write white, black, mulatto, quadroon, octo- and slaveholder Josiah Nott lobbied certain upon the correct determination of this class in roon, Chinese, Japanese, or Indian, accord- 11(p26) senators for the inclusion in the census of sev- schedules 1 and 2 [italics in original]. ing to the color or race of the person enu- eral inquiries designed to prove his theory of Schedule 1 was for population, and Schedule 2 merated. Be particularly careful to distinguish 9,10 between blacks, mulattoes, quadroons, and mulatto hybridity and separate origins. In was for mortality. The 1880 instructions for octoroons. The word ‘black’ should be used the end, the senators voted to include only the “color” were nearly identical. to describe those persons who have three- category “mulatto,” although they hotly de- How was polygenism able to withstand fourths or more black blood; ‘mulatto,’ those bated the inclusion of another inquiry— Darwinism’sclaim that all humankind had de- persons who have from three-eighths to five- eighths black blood; ‘quadroon,’ those per- ”[d]egree of removal from pure white and black scended from a common evolutionary ancestor? sons who have one-fourth black blood; and races”—as well. Instructions to enumerators Polygenists profited from the fact that Darwin’s ‘octoroons,’ those persons who have one- for the slave population read, “Under heading main claim left unaddressed 2 of polygenism’s eighth or any trace of black blood [italics in 11(p36) 5 entitled ‘Color,’ insert in all cases, when the central concerns: the effects of racial mixture original]. slave is black, the letter B; when he or she is a and the capacities of races.12,13 As polygenists For 50 years, from 1850 to 1900, the census mulatto, insert M. The color of all slaves should saw it, common ancestry did not erase the ev- contributed directly to the formation of scientific be noted.” For the free population, enumera- ident fact of human diversity, nor did it explain ideas of race. These ideas were the backbone of tors were instructed as follows: “in all cases the content of those differences or the effects of a racial discourse that justified and sustained slav- where the person is black, insert the letter B; if racial intermixture. That Whites and Blacks ery and then de jure and de facto racial segrega- mulatto, insert M. It is very desirable that these could mate did not mean that they should. More tion. At the same time, social scientists studied particulars be carefully regarded.”11(p23) information was needed about the physical and race because of their scientific interest in it, for The 1850 census introduced a pattern, es- psychological effects of racial mixture on reasons distinct but not disconnected from larger pecially in regard to the mulatto category, that Whites, Blacks, and their mulatto offspring. political, social, and economic developments. lasted until 1930: the census was deliberately Moreover, if humankind had evolved from However, in the 20th century there was a used to advance race science. Such science was common ancestors, that did not mean that the marked change. Twentieth-century censuses fundamental to, though not the only basis of, races had followed similar or even comparable ceased to play such a prominent role in the for- racial discourse—that is, the discourse that ex- evolutionary processes. Indeed, polygenists ar- mation of racial theory; instead, for the most part plained what race was. Far from merely count- gued that Whites and Blacks had evolved so they simply counted by race, presuming race to ing race, the census was helping to create race differently that it rendered their common an- be a basic fact.Theorizing about race continued by assisting scientists in their endeavors. Al- cestry practically meaningless. in social science circles, but scientists and theo- though scientific ideas about race changed over By the 1890 census, polygenism and Dar- rists did not deliberately enlist the census in their those 80 years, the role of the census in ad- winism came to coexist. Darwinism had not theorizing, as they had in the past. Census cate- vancing such thought did not. replaced polygenist thought but rather had gorization continued to sustain racial discourse The abolition of slavery and the reconsti- combined with it. Race scientists and social inasmuch as categorizing and counting by race tution of White racial domination in the South theorists were convinced, according to their in- gave race an official existence. The use of the were accompanied by an enduring interest in terpretation of Darwin, that all races were en- “mulatto” category in racial theorizing until the race. Predictably, the ideas that race scientists gaged in a struggle for survival. They trans- 1930 census was an important exception to this and proslavery advocates had marshaled to de- lated Darwin’s biological idea of natural overall trend. By 1930, the definitions of “non- fend slavery were used to oppose the recogni- selection into a social theory of racial strug- White” categories became consistent with legal tion of Black political rights. Blacks were nat- gle. Yet, in keeping with their polygenist pre- definitions of non-White racial membership. urally inferior to Whites, whether as slaves or occupation with “mulattoes,” the same scien- Since 1970, the census has once again as free people, and should therefore be dis- tists and social theorists considered mulattoes emerged as a venue for directly enabling pub- qualified from full participation in American to be at a distinct disadvantage and thought lic policies and for shaping debate about the economic, political, and social life. Although they would die off. Mulatto frailty would prove concept of race itself.The census now supports scientists, along with nearly all Whites, were that racial mixture engendered racial disad- civil rights legislation, and racial discourse once convinced of the inequality of races, they con- vantage and would result in eventual disap- again turns on the same basic question that tinued in their basic task of investigating racial pearance or reversion to the “dominant type.” 19th-century social scientists were driven to origins. Darwinism presented a challenge to The “dominant type” was, of course, presumed answer: What is race? As in the past, the cen- the still dominant polygenism, but the mulatto to be Black; at no point before or after 1890 sus is being used to answer that question. How- category retained its significance within po- were mulattoes considered “mixed Whites.” ever, there is now a much wider circle of par- lygenist theories. Data were needed to prove Blacks and other non-Whites were mixed; ticipants, including census bureau officials, that mulattoes lived shorter lives, thus proving Whites were not. These ideas emerged power- politicians, social scientists, civil rights advo- that Blacks and Whites were different racial fully in the 1890 census, and certain of them cates, policymakers, and organized groups species. persist today. within civil society who are seeking recognition.
1740 American Journal of Public Health November 2000, Vol. 90, No. 11 Three interrelated fundamental shifts in were dissatisfied with the quality of 1890 mu- Mixed Races” meant that any mixture of White American intellectual, institutional, and polit- latto, octoroon, and quadroon data.) The basic and non-White should be reported according to ical life account for the more constrained in- idea that distinct races existed and were en- the non-White parent. Similarly detailed in- fluence of the census. First, race science settled duringly unequal remained firmly in place. structions, of paragraph length, were provided into a set of ideas that would dominate for What happens when superior and inferior races for “Mexicans” and “Indians.” nearly 40 years and would then be challenged mate? Social and natural scientists still wanted In contrast, legal definitions of “White” for decades thereafter: that discrete races ex- to know. But the advisory committee to the did not change, when they existed at all. In isted; that these races possessed distinctive in- Census Bureau decided in 1928 to terminate general, Southern laws conceived of White as tellectual, cultural, and moral capacities; and use of the mulatto category on censuses. the complete absence of any “Negro or non- that these capacities were unequally distrib- The stated reasons for removal rested on white blood,” down to the last drop and as far uted within and between racial groups. Social accuracy. Had the advisory committee pos- back generationally as one could go.20 Again, scientists no longer used the census to sort out sessed confidence in the data’saccuracy or the the census reflected legal practices by never the basic questions of race science. Instead, the Census Bureau’s ability to secure accuracy, providing a definition of White. census registered the evident existence of race. “mulatto” might well have remained on the The state of racial discourse was more Second, the Census Bureau’s gradual census. The committee did not refer to the ev- unstable than the 1930–1960 census instruc- institutionalization changed perceptions ident inability of the mulatto category to set- tions would lead us to believe. By the 1940s, about the purposes and limits of racial enu- tle the central, if shifting, questions of race sci- the scientific foundations of the discourse had meration. The bureau would eventually be- ence: first, whether “mulatto-ness” proved that shifted noticeably. Cultural anthropologists, come a full-fledged bureaucracy, its meth- Whites and Blacks were different species of under the guidance of Franz Boas, com- ods soundly grounded in statistical science. humans, and then, whether mulattoes were pellingly challenged the basic tenets of race Its mission was to provide racial data, with- weaker than members of the so-called pure science. Nazism forced social scientists world- out explicitly advancing racial thought and races. The exit of the mulatto category from wide to reexamine their thinking on race. How- without being beholden to political inter- the census was markedly understated, espe- ever, a change in the thinking of social scien- ests. Counting by race would come to be cially when compared with its entrance in 1850 tists did not alone account for changes in racial widely viewed as an administrative task and and its enduring significance on 19th-century discourse. The decline of the South’seconomy, technical procedure, not a tool of scientific censuses. the massive migration of Southern blacks to investigation. Moreover, decision making Beginning with the 1890 census, all Native Northern and Midwestern cities, an increase about racial categorization became an even Americans, whether taxed or not, were counted in political participation and agitation, suc- less public process and purportedly a less on general population schedules.16 Much as cessful legal challenges to segregation, and the political one, as Congress deferred to the racial theorists believed that enumerating mu- onset of the Cold War transformed the politi- internal decision-making processes of the lattoes would prove their frailty, they thought cal landscape. This new landscape was far less Census Bureau. that Native Americans were a defeated and van- nourishing to the prevailing variants of racial In 1902, the bureau became a permanent ishing race. Given the weight of these expecta- discourse. The acceptance of race did not mean federal agency under the Department of Com- tions in the late 19th century, it is not surpris- that American social, political, and economic merce and Labor.15 In 1918, an advisory com- ing that census methods and data reflected them. life would or should continue to be organized mittee was formed to assist in the development As the historian Brian Dippie observed, “the around race in the ways it had been. Ideas of of schedules and inquiries, including the race expansion and shrinkage of Indian population race, the census, and the attendant (and proper) question. This committee advised the bureau estimates correlate with changing attitudes about public policies had long been inseparable; they until the mid-1940s. In 1954, all census legis- the Native American’s rights and prospects.”17 were no longer. lation becameTitle 13 of the United States Code. The idea of the vanishing Indian was so perva- At the same time, it became increasingly Third, the hardening of racial segregation sive that the censuses of 1910 and 1930 applied difficult to discuss what race was in a coherent and subordination, both de jure and de facto, a broad definition of “Indian” because officials way, other than to state that it did not exist, bi- paralleled the hardening of scientific thought. believed that each of these censuses would be ologically. Civil rights discourse has focused Southern law had largely settled on the “one drop the last chance for an accurate count.18 exclusively on racism, discrimination, and of non-White blood” rule of racial membership equality, leaving aside the question of race it- by 1930.The definitions of non-White categories 1930–1960 Censuses self. Census-taking in the post–civil rights pe- as spelled out in census enumerator instructions riod has reflected this tension: census data are were identical to those of Southern race laws. It With removal of the mulatto category, cat- used to remedy racial discrimination, while is important to emphasize, however, that the def- egories and instructions for the 1930 census census categories are themselves supported by initions of White and non-White racial mem- mirrored the racial status quo in law, society, a decentered, conflicting, and, in certain ways, bership were not limited to the South or its legal and science. Southern statutes that had defined anachronistic racial discourse. regime. They were imposed and assumed na- Negroes and other non-Whites by referring to tionwide, thereby explaining their appearance a specific blood quantum now defined them 1970–2000 Censuses on the federal census. But census categories did broadly. Any person with any trace of “Black not simply reflect race laws, scientific thought, blood” was legally Black and subject to all the The civil rights movement and resulting and social customs.The “mulatto” enumeration disabilities the designation conferred. Census civil rights legislation of the 1960s dramati- shows that census-taking followed its own path definitions followed suit, and enumerator in- cally changed the political context and pur- to the same destination of the “one-drop” rule. structions in 1930 read, in part: “A person of poses of racial categorization. Federal civil The mulatto category remained on the mixed white and Negro blood should be re- rights legislation—most notably the Civil 1910 and 1920 censuses for the same reason turned as a Negro, no matter how small the Rights Act of 1964, the Civil Rights Act of that it had been introduced in 1850: to build percentage of Negro blood. Both black and 1968, and the Voting Rights Act of 1965—dis- racial theories. (Census officials removed the mulatto persons are to be returned as Negroes, mantled the most egregious discriminatory category from the 1900 census because they without distinction.”19 The category “Other mechanisms, namely, Black disenfranchise-
November 2000, Vol. 90, No. 11 American Journal of Public Health 1741 Source. US Census Bureau. (Available at: http://www.census.gov/dmd/www/houser.htm. Accessed September 6, 2000.)
Promotional poster depicting a sculpture by Allan Houser and encouraging Native American cooperation with the 2000 census. ment in the South, rigid residential segrega- been the issuance of Statistical Directive No. 15 offspring of interracial marriages. In its review, tion, and wholesale exclusion of Blacks from by the OMB. Since 1977, the directive has the OMB actively sought public comment certain occupations and American institutions. mandated the standards that govern all statis- through congressional subcommittee hearings These new laws and programs required racial tical reporting by all federal agencies, includ- in 1993 and 1997 and notices posted in the and ethnic data for monitoring legislative com- ing the Census Bureau. The directive defines Federal Register. Not surprisingly, well- pliance and the delivery of new social services Hispanic as an ethnic category, meaning that established civil rights organizations lobbied and programs. For example, the Civil Rights there are, for example, White Hispanics and against major changes in the directive, while Act of 1964 and the Voting Rights Act of 1965 Black Hispanics. As for persons of mixed racial newly formed organizations of multiracial and their subsequent amendments, extensions, or ethnic origins, the directive instructs that Americans lobbied for the addition of a sixth and court interpretations require population such persons be classified according to the cat- official racial category, “multiracial.” They ar- tabulations by race at the level of city blocks for egory that “most closely reflects the individ- gued that the “one-drop rule” of non-White the purposes of redistricting and the possible ual’s recognition in his community.” According racial membership was no longer valid and that creation of minority and majority congressional to the directive’s preamble, these categories census categorization should reflect new un- electoral districts.21 were devised to standardize “record keeping, derstandings of race. The now positive benefits of racial cate- collection, and presentation of data on race and Numerous other groups presented the gorization and racial data have stimulated and ethnicity in Federal program administrative re- OMB with their own suggestions, each de- sustained organized attempts to have categories porting and statistical activities.” The defini- signed to enhance the recognition of a partic- protected, changed, and added. The Hispanic- tions, the directive cautions, “should not be in- ular group. For example, the Celtic Coalition, origins question, for example, was added to terpreted as being scientific or anthropological the National European American Society, and the 1980 census in response to lobbying by in nature”; they were developed to meet con- the Society for German-American Studies all Mexican American organizations,22 and sev- gressional and executive branch needs for called for the disaggregation of the White cat- eral Asian categories were added to the 1980 “compatible, nonduplicated, exchangeable egory. The Arab American Institute lobbied for and 1990 censuses in response to lobbying by racial and ethnic data.”24 Thus, these categories the reclassification of persons of Middle East- Asian American organizations.23 Civil rights are both statistical markers and political ern origin from White to a new Middle East- advocates took racial categories (legal and cen- instruments. ern category. sus) as they were, arguing that such categories In 1993, the OMB began a comprehen- At the OMB’s request, the National Acad- had been the basis of discrimination and should sive review of the directive. According to OMB emy of Sciences Committee on National Sta- thus serve as the basis of remedy. officials, this review was prompted by growing tistics conducted a 1994 workshop that in- Perhaps most politically consequential for public criticism that the directive was incapable cluded federal officials, academics, public census-taking in the post–civil rights era has of accurately measuring new immigrants or policy analysts, corporate representatives, and
1742 American Journal of Public Health November 2000, Vol. 90, No. 11 phasized racial mixture with the same vigi- TABLE 2—Brazilian Color Questions and Categories, 1872–2000 lance that their American counterparts have 1872—White (branco), Black (preto), Mixed (pardo), Caboclo (Mestizo Indian) exercised in emphasizing racial purity and ex- 1880—No census clusivity. Brazilian social scientists largely ac- 1890—White (branco), Black (preto), Caboclo (Mestizo Indian), Mestiço cepted the scientific truth of races and their in- 1900—No color question equality, though not with the same intensity as 1910—No census Americans and Europeans. Like American 1920—No color question, but extended discussion about “whitening” 1930—No census (Revolution of 1930) elites, Brazil’s elites were obsessed with racial 1940—White (branco), Black (preto), Yellow (amarelo)a mixture, but they concluded that Brazilians 1950—White (branco), Black (preto), Mixed (pardo), Yellow (amarelo) were becoming a whiter race, not a racially de- 1960—White (branco), Black (preto), Mixed (pardo), Yellow (amarelo), Índio (Indian) graded and disadvantaged one. 1970—No color question 1980—White (branco), Black (preto), Mixed (pardo), Yellow (amarelo) 1991—White (branco), Black (preto), Brown (pardo), Yellow (amarelo), Indígena 1872–1910 Censuses (indigenous) 2000—White (branco), Black (preto), Brown (pardo), Yellow (amarelo), Indígena Although the 1872 Brazilian census was (indigenous) conducted 1 year after the passage of major Source. Instituto Brasileiro de Geográfia e Estatística. abolitionist legislation, neither census inquiries aIf the respondent did not fit into one of these 3 categories, the enumerator was instructed nor census data were used in slavery debates. to place a horizontal line on the census schedule. These horizontal lines were then Likewise, although Brazilian intellectual and tabulated under the category pardo. political elites were preoccupied with the per- ceived calamity of racial mixture, they did not use the census to examine the problem, unlike their American counterparts. The categories secondary-school educators. In March 1994, brown or mixed (pardo), and Black (preto) on both 19th-century Brazilian censuses were the OMB established the Interagency Com- used in nearly every census. The history of nearly identical: white (branco), black (preto), mittee for the Review of Racial and Ethnic color categorization can be divided into 3 pe- brown or mixed (pardo), caboclo (mestizo In- Standards. This committee included represen- riods. The first is from 1872 through 1910, dian). The 1890 census added to these 4 cate- tatives from 30 federal agencies, including the when categorization largely reflected elite and gories the category of mestiço (racially mixed). Census Bureau, the Department of Justice, and popular conceptions of Brazil’s racial compo- Paradoxically, the census was one of the the Department of Education. sition. The second is from 1920 through 1950, few late-19th-century undertakings that was In the end, the committee’s recommen- when census texts actively promoted and re- not preoccupied with or used to discern the na- dations to the OMB ruled the day. In October ported the whitening of Brazil’s population. tional disaster that Brazilian elites were con- 1997, the OMB announced its final changes The third is from 1960 to the present, when vinced would accompany racial mixture. As to the directive and to census methods. Most categorization methods have been questioned the Brazilian historian Lilia Moritz Schwarcz significantly, the OMB decided for the first and contested by statisticians within the Cen- has richly documented, museums, historical time in the history of American census-taking sus Bureau and by organized groups within societies, law schools, medical schools, and to allow respondents to choose more than one civil society. scientists all focused on racial mixture because race on their census schedules. It therefore de- Brazilian censuses have included a color it was the key to understanding Brazil and its cided against the adoption of a single “mul- question for the same basic reason that Amer- national possibilities.26 The silence of the cen- tiracial” category. It also made slight alterations ican censuses have included a race question. sus was likely due to the modest state of the in the wording of existing categories. The issue Brazilian elites viewed race as a natural com- statistical institute and the underdevelopment of racial categorization is temporarily settled, ponent of human identity and as an independ- of statistical methods. The establishment of the until preparations for the 2010 census begin. ent factor in human affairs. Brazilian censuses General Directory of Statistics accompanied have not counted by race as such, but by color. the abolition of slavery in 1888 and the estab- Color has referred to physical appearance, not lishment of the Old Republic in 1890. Histo- The Brazilian Census: White Is racial origins. Racial origins, however, are not rians consider all 3 of the censuses conducted Better disconnected from color, because color is de- by the General Directory of Statistics (1890, rived from the mixture of Brazil’s 3 “original” 1900, and 1920) unreliable.1 Brazil’s modern Compared with the American experience races: European, African, and Indian. Color federal census bureau, the Brazilian Institute of census-taking, that of Brazil seems relatively and race are conceptually distinguished, but of Geography and Statistics (Instituto Brasileiro simple, if erratic. The color question has ap- related: color refers to appearance, race refers de Geográfia e Estatística [IBGE]), was es- peared inconsistently on Brazilian censuses to origin.25 While this distinction is hardly un- tablished in 1938. from the first modern census in 1872 up to the ambiguous, it lies at the heart of Brazilian racial 2000 census (see Table 2). The two 19th- discourse and the support of a census that 1920–1950 Censuses century censuses, in 1872 and 1890, included counts color. a color question. Of 20th-century censuses, the The thinking has gone as follows: Bra- In the 20th century the role of the census 1940, 1950, 1960, 1980, and 1991 censuses zilians are racially mixed, of different colors. changed dramatically, as did the Brazilian asked a color question, although the 1960 color This racial mixture has made counting by race elites’ideas about racial mixture. In a sharp re- data were never fully released. The 1900, 1920, exceedingly imprecise. However, the census versal, intellectuals posited that the disastrous and 1970 censuses did not contain a color ques- question and categories have themselves or- consequences of racial mixture would be tion. No census was taken in 1910 or 1930. ganized the fluid boundaries of the racial mix- averted because Brazilians would become Categorization has been more consistent, ture presumed to exist. Brazil’s intelligentsia, whiter over time. Racial mixture was not de- with the 3 color categories of White (branco), political elite, and census officials have em- generative but fortifying for whites and cleans-
November 2000, Vol. 90, No. 11 American Journal of Public Health 1743 ing for nonwhites.27 Whitening would also be activists raised the issue of terminology most mining levels of educational attainment, em- achieved through European immigration. forcefully as the IBGE prepared for the 1991 ployment prospects, and income.32,33 It is hard to overemphasize the centrality and 2000 censuses. The National Census Commission, of census data to 20th-century claims of a The 1940 census also celebrated whiten- which was appointed by the military gov- racially mixed Brazilian people and the polit- ing. The author and esteemed educator Fer- ernment, removed the color question from ical and social arguments that have flowed from nando de Azevedo wrote the census text, which the 1970 census—against the recommenda- such claims. In the first half of the century, was also published separately as a book and tions of 2 experts the military itself had con- census texts reported, as a positive develop- (this time) translated into English.31 Azevedo sulted. In the late 1970s, scholars and black ment, that Brazilians were becoming whiter. concluded the chapter “Land and Race” (race, activists lobbied to have the question re- The 1920 census included an extended dis- like land, was assigned a natural and funda- stored to the 1980 census. It was restored, cussion of the whitening of Brazil’s popula- mental status) with the observation, “If we although the statistical institute’s president tion. In a section of the census titled “Evolution admit that Negroes and Indians are continuing remained opposed to it and called the ques- of the Race” (which was later published sepa- to disappear, both in the successive dilutions tion “unconstitutional.” rately as a book), the social theorist Oliviera of white blood and in the constant progress of Since Brazil’s redemocratization in the Vianna explained that the “aryanization” of biological and social selection,” Brazil would mid-1980s (after 21 years of military rule), ac- Brazilians was under way.28 Within mestiço soon be white.31(p41) The pardo category was tivists and scholars have aggressively challenged groups the “quanta of barbaric bloods” were reinstated in the 1950 census schedule, making the discourse of racial democracy. They have decreasing, while the quantum of “white blood” the 4 choices white (branco), black (preto), also, necessarily, challenged census methods was increasing, each time refining the Brazil- brown or mixed (pardo), and yellow (amarelo). and terminology. Their efforts have prompted ian race. Self-identification replaced enumerator deter- reexamination within the IBGE. In the early Given the pervasiveness of the elites’ be- mination in 1950 as well. 1980s, for example, a group of statisticians and lief in whitening, it is not surprising that this be- analysts within the IBGE’sDepartment of So- lief was communicated in the census text. But 1960–2000 Censuses cial Studies and Indicators decided to pool the text is surprising, because the 1920 census pardo and preto data under the term negro did not include a color question. Therefore, its From the 1950s onward, Brazilian census (black) in socioeconomic analyses and tables. predictions of whitening were not based on texts spoke little about whitening. The pro- They decided that this action was appropriate data collected contemporaneously, however found shifts in scientific racial thought after because the 2 groups had similar socioeconomic unreliable and ambiguous such data certainly World War II largely account for this change. profiles and because negro is the preferred term would have been. Vianna most likely wrote her Census texts spoke less aggressively and less of black activists and certain academics.34 text to assure elites that Brazil’s future as a frequently of both whitening and the regener- Activists and academics again raised the white country was certain, thereby making the ative and redemptive powers of racial mixture. issue of terminology through a grassroots cam- continued recruitment of European workers Instead, racial mixture was reported in a mat- paign surrounding the 1991 census. The cam- unnecessary. By 1920, industrialists and politi- ter-of-fact way and was not equated automat- paign “Náo deixe sua cor passar em branco: cians were fed up with the militancy of immi- ically with whitening. However, Brazilians still Responda con bom C/senso” (“Don’t Let Your grant workers.29 believed in distinct races, if not in their inher- Color Pass in White: Respond With Good The 1940 census was the first 20th- ent superiority or inferiority, and in racial mix- Sense”) urged Brazilians to check a darker century census to ask a color question. Cen- ture. Moreover, Brazilian elites have used color color on their census schedules. It publicly sus enumerators were to check white, black, data to promote the image of Brazil as a racial raised 2 fundamental issues. First, the cam- or yellow for each respondent. If the respondent democracy. According to this view, Brazilian paign confronted the IBGE by asking why the did not fit into one of these 3 categories, the citizenship has been neither enhanced, dimin- term “color” was used and not the term “race,” enumerator was to place a horizontal line on the ished, nor stratified because of race. Presumed and why the terms preto and pardo were used census schedule. These lines were later tabu- racial differences are not a way of distinguish- and the term negro was not. Second, the cam- lated under the category pardo. Indigenous per- ing among Brazilians, because Brazilians are paign questioned the preference of most Bra- sons were counted as pardo as well. The IBGE racially mixed. They are simply Brazilians, zilians to choose a lighter color, especially their eliminated the category pardo in response to with their different colors. decision not to select black (preto) to describe the rise of European fascism. According to The census, in counting by color rather themselves on census schedules. IBGE documents, the category’s elimination than race, has thus been instrumental to the The 1991 color question was like previous would assure Brazilians that census data would discourse of racial democracy. Moreover, the questions, with one important exception: the not be used for discriminatory purposes.30 IBGE has been reluctant to cross-tabulate color terms raça (race) and indígena (indigenous) It is important to note also that the mean- categories with socioeconomic variables or to were added. The question was rephrased to ask, ing of pardo was then, and remains, ambigu- release color data in a timely fashion. Until the “What is your color or race?” and “indigenous” ous. Portuguese-language dictionaries define it early 1980s, the lack of such socioeconomic was added to the colors white, black, brown, as both “gray” and “brown.” Its connotations data made it impossible to test the claim that and yellow. (Since 1940, indigenous persons are equally ambiguous, because Brazilians use color was economically and socially inconse- had been classified as pardos.) These 2 new the word infrequently in common parlance. Its quential in Brazil. It also stymied the advocacy terms were linked: race applied only to the in- most significant use is as a census term. Al- efforts of scholars, policymakers, and activists digenous population. Indigenous persons be- though controversy did not then surround the for remedial and positive public policies. It was long to one race, Brazilians to another race, category preto (black), it has also been a pe- not until the 1976 Household Survey that the with its many colors. The IBGE’s decision to culiar term for the IBGE to use. Brazilians usu- IBGE produced data that correlated color with include “indigenous” was reportedly made after ally use the term in the third person, not the income, health, education, and housing. Since consultations with anthropologists and repre- first person, as the census requires. Even more then, there has a been a veritable boom in quan- sentatives of the Federal Indian Affairs Bureau. illuminating, Brazilians use it most commonly titative research, all of which has clearly shown Campaign organizers speculated, however, that to describe objects, not human beings. Black that color is a significant variable in deter- the term was included at the request of the
1744 American Journal of Public Health November 2000, Vol. 90, No. 11 World Bank, which wanted demographic in- port to be. The recent efforts of Americans and vival. Norman: University of Oklahoma Press; formation for World Bank initiatives on the Brazilians to have categories changed, added, 1987. protection of indigenous territories. or maintained have had the happy effect of 17. Dippie B. The Vanishing Indian: White Attitudes In the midst of preparations for the 2000 forcing census bureaus to account publicly for and U.S. Indian Policy. Middletown, Conn: Wes- census, there was growing public and schol- their methods and rationales. There are no sim- leyan University Press; 1982:xv. arly debate about IBGE methods and terms. ple, obviously right or obviously wrong an- 18. Historical Statistics of the United States: Colonial In these debates the IBGE had to explain and swers to the question of whether American or Times to 1957. Washington, DC: US Bureau of often defend its past and current methods. The Brazilian censuses should continue to count the Census; 1960:3. sources of pressure on the IBGE include de- by race or color. However, we are better 19. Instructions to Enumerators. Washington, DC: mographers, black activists, academics, and equipped to think about such questions once US Bureau of the Census; 1930. politicians. With the unraveling of racial de- we understand the complex relationship be- 20. Murray P, ed. States’ Laws on Race and Color. Cincinnati, Ohio: Women’s Division of Christ- mocracy, the question of who Brazilians tween race and censuses. ian Service; 1951. “really” are, racially, has reemerged power- 21. Edmonston B, Schulze C. Modernizing the U.S. fully. There is a clear reason for this connec- Census. Washington, DC: National Academy tion. The image of a racially democratic and References Press; 1995. nondiscriminatory society has hinged on the 1. Goyer D, Domschke E. The Handbook of Na- 22. Choldin HM. Statistics and politics: the “His- idea of racial mixture. In fact, a causal link tional Population Censuses: Latin America and panic issue” in the 1980 census. Demography. was drawn that was often presented tautolog- the Caribbean, North America and Oceania. 1986;23:403–418. ically: Brazilians are racially mixed and there- Westport, Conn: Greenwood Press; 1983. 23. Espiritu YL. Asian-American Panethnicity: fore there can be no discrimination, or there 2. Nobles M. Shades of Citizenship: Race and the Bridging Institutions and Identities. Philadelphia, can be no racial discrimination because Bra- Census in Modern Politics. Stanford, Calif: Stan- Pa: Temple University Press; 1992. zilians are racially mixed. The acceptance of ford University Press; 2000. 24. Office of Management and Budget. Standards the existence of discrimination—an existence 3. Horsman R. Race and Manifest Destiny: The for the Classification of Federal Data on Race substantiated by census data—has led un- Origins of American Racial Anglo-Saxonism. and Ethnicity. Available at: http://www. avoidably to the abandonment of the idea of Cambridge, Mass: Harvard University Press; whitehouse.gov/OMB/fedreg/directive_15.html. 1981. Accessed August 29, 2000. racial democracy and to a rethinking of cen- 4. Kettner JH. The Development of American Citi- sus terms and methods. 25. Nogueira O. Tanto Preto Quanto Branco: Estudos zenship, 1608–1870. Chapel Hill: University of de Relações Raciais. São Paulo, Brazil: TA The discourses of whitening and racial North Carolina Press; 1978. Queiroz; 1985. democracy have resided in census methods and 5. Finkelman P. Prelude to the Fourteenth Amend- 26. Schwarcz LM. The Spectacle of the Races: Sci- texts as much as they have existed in the real ment: black legal rights in the antebellum north. entists, Institutions, and the Race Question in world. As Brazilians now consider whether Rutgers Law J. 1986;17:415–482. Brazil. New York, NY: Hill & Wang; 1999. 6. Jordan W. White Over Black: American Attitudes their society is composed of distinct racial 27. Skidmore T. Black Into White: Race and Nation- Toward the Negro, 1559–1812. Chapel Hill: Uni- groups rather than one racially mixed people, ality in Brazilian Thought. Durham, NC: Duke versity of North Carolina Press; 1968. the census will undoubtedly be involved in ad- University Press; 1995. vancing a new racial discourse. However, the 7. Stanton W. The Leopard’s Spots: Scientific Atti- tudes Towards Race in America. Chicago, Ill: Uni- 28. Vianna O. Evolução do Povo Brasileiro. Rio de terminology on the 2000 census was the same versity of Chicago Press; 1968. Janeiro, Brazil: José Olympio; 1956. as in past censuses: color was used, not race; 8. Gould SJ. The Mismeasure of Man. New York, 29. Maram S. Labor and the Left in Brazil, 1890– pardo and preto were used, not negro. NY: WW Norton & Co; 1996. 1921: a movement aborted. Hispanic Am His- 9. Cong Globe. 31st Cong, 1st sess, April 9, 1850: torical Rev. 1977;57:254–272. 671–675. 30. Estudos Sobre a Composição da População de Conclusions 10. Horsman R. Josiah Nott of Mobile: Southerner, Brasil Segundo a Cor. Rio de Janeiro, Brazil: In- Physician, and Racial Theorist. Baton Rouge: stituto Brasileiro de Geográfia e Estatística; 1950: What are the larger lessons of the Amer- Louisiana State University Press; 1987. 11. Estudos de Estatística Teórica e Aplicada: ican and Brazilian experiences? As I see it, the 11. 200 Years of Census-Taking: Population and Estatística Demográfica. lessons are several. First, these experiences re- Housing Questions, 1790–1990. Washington, 31. Azevedo F. Brazilian Culture: An Introduction to veal the sinuous relationship between racial DC: US Bureau of the Census; 1989. the Study of Culture in Brazil. New York, NY: ideas, census taking, and public policy. They 12. Haller JS. Outcasts From Evolution: Scientific Macmillan; 1950. teach us that racial categories on censuses do Attitudes of Racial Inferiority, 1859–1900. Ur- 32. Lovell PA. Desigualidade Racial no Brasil Con- not merely capture demographic realities, but bana: University of Illinois Press; 1971. temporâneo. Belo Horizonte, MG: MGSP Edi- tores; 1991. rather reflect and help to create political reali- 13. Stocking G. Race, Culture, and Evolution. New York, NY: Free Press; 1968. ties and ways of thinking and seeing. The cat- 33. Lovell PA, Wood CH. Skin color, racial identity, 14. Cong Rec. 50th Cong, 2d sess, 1889:Vol 20, and life chances in Brazil. Latin Am Res Rev. egories are themselves intellectual products, p 2246. 1998;25:90–109. social markers, and policy tools. 15. Anderson M. The American Census: A Social 34. Oliveira LE, Porcaro RM, Araujo TC. O Lugar They also teach us that census bureaus History. New Haven, Conn: Yale University Press; do Negro na Força de Trabalho. Rio de Janeiro, must be viewed as the political insiders that 1988. Brazil: Instituto Brasileiro de Geográfia e Esta- they are, not the detached recorders they pur- 16. Thornton R. American Indian Holocaust and Sur- tística; 1985.
November 2000, Vol. 90, No. 11 American Journal of Public Health 1745