Social Dimensions of Race: Mexico City, 1753 Open Access

The relationship among race, social position, and economic roles was one of considerable significance in colonial Latin America. Race and presumed racial differences were the basic criteria for social differentiation. Initially the elite were (white) European-Spanish peninsulars, the slaves were Black, and the peasants were Indians, commonly residing apart from both Europeans and Blacks in semiautonomous communities. The economic organization of colonial society was based upon similar divisions with specific necessary tasks associated with various racial groups. The Indians were the tribute payers, the producers of food; the Blacks were the skilled and unskilled laborers on sugar plantations and in mines and servants in the great houses of the major cities. The peninsulars were the economically and socially privileged sector, benefiting from the labor of the Black and Indian populations. This tripartite division soon collapsed as miscegenation produced a series of intermediate racial categories.

By the middle of the eighteenth century, the process of racial mixture in New Spain had been taking place for more than two hundred years. There were thus sufficient numbers of individuals in the category of mixed-bloods to be considered significant segments of the social order. Not only were the descendants of mixed racial origin present in sufficient numbers to make an impact on the social hierarchy, but they had lost much of their original definition as strictly biological types. By the middle of the eighteenth century, the relationship between race and the division of labor, or the socioeconomic dimension of race, was no longer as straightforward as it had been before.

Where each of these new racial categories fit into the division of labor is a question that has interested several historians. Magnus Mörner suggested a series of clear-cut distinctions whereby Spaniards were bureaucrats and merchants, creoles were large landowners, mestizos were artisans, shopkeepers, and tenants, mulattoes were urban manual laborers, and Indians were peasants and unskilled laborers.¹ Several subsequent studies, however, have suggested that the relationship between race and the division of labor was less clear-cut. David A. Brading’s work on Guanajuato partially supported Mömer’s vision of colonial society, finding that the peninsulars and creoles occupied elevated positions in society and mulattoes occupied the bottom (Indians figuring only slightly in the mining society of Guanajuato).² Brading modified Mörner’s picture of the mestizo by showing that in Guanajuato they occupied what he called an “ambiguous middle layer” with no particular occupational pattern.³ He suggested an association between racial categories and the division of labor at the poles of the social hierarchy, but a weaker relationship between occupation and the intermediate racial categories. More recently, John K. Chance and William Taylor’s analysis of Antequera broadened Brading’s “ambiguous middle layer” to include the creoles and mulattoes in addition to the mestizos.⁴ Chance and Taylor indicated that creoles, mestizos, and mulattoes could all be found principally in the same three occupational categories: high-and low-status artisans and servants.⁵ A critique of Chance and Taylor by Robert McCaa, Stuart Schwartz, and Arturo Grubessich suggested that the mestizos and mulattoes were somewhat overrepresented in the artisan and servant groups and underrepresented in the professional and elite groups, but supported Chance and Taylor’s contention that the creoles lacked a distinct occupational pattern.⁶

Although differing in their interpretation of degrees of differentiation, these recent studies have suggested that the relationship between racial categories and the division of labor in late colonial Mexico was less definite than previously thought. The issues raised by the relationship of creoles, mestizos, and mulattoes to the division of labor clearly merit further consideration.

The aim of this study is to examine the extent to which the racial labels continued to be associated with the division of labor in the viceregal capital of New Spain toward the end of the colonial period. For such an analysis it is necessary to examine not only the intermediate groups resulting from miscegenation, but the parent groups as well in order to place the intermediate groups in context. To accomplish this task, this article will define the racial terms used in eighteenth-century Mexico and reevaluate the significant racial distinctions within the social division of labor. The major categories within the division of labor here have two central determinants: the role of a group of individuals in production, and the relationships among such groups in the overall social organization of production. This focus specifically rejects approaches that attempt to examine aspects of distribution and consumption (such as income and residence) stemming from production, while failing to consider the social totality.⁷ Since the racial labels given have been observed to shift, the article will examine the relationship between variability in racial labeling and the location of racial groups within the division of labor. Finally, in an  appendix, the utility of “caste” and “class” as characterizations of Mexico’s social order will be evaluated. The study addresses these issues using empirical evidence from Mexico City in the middle of the eighteenth century.

Mexico City is a significant area for the examination of the social dimensions of race for several reasons. The most coveted positions in the Spanish colonial bureaucracy were to be found in the capital.⁸ The combination of the potential for political influence with the traditional urban orientation of Spanish society meant the city was a magnet for the well-to-do. The wealthy and the powerful established their town residences, employed nearly 450 Black and mulatto slaves as house servants, and created demand for a host of other services.⁹ Demand for these services, such as tailoring, shoemaking, and weaving, in turn created employment for a variety of people of less substantial economic and political standing. This meant the city attracted not only the rich, but the less wealthy as well.

The rural poor of Mexico flooded the city’s outlying areas during times of food shortages or outright famine. These displaced people were primarily Indian and lived largely in the Indian sections of the city (called parcialidad.es) and the vacant lots and empty spaces of the center of the city. The men provided the unskilled labor, and the wood, water, and food for the wealthier inhabitants of the city; the women worked as servants in the homes of the rich and powerful. The economic attraction of the viceregal capital extended from the wealthy to the poor to those in between. This scope of economic opportunity meant that a range of racial groups came to the city as well, including Black slaves, rural Indians, and European and American Spaniards.

As a geographic locale, the capital city provided the opportunity for far greater contact between individuals of different races than did the rural regions and smaller towns of New Spain. Mexico City was thus an area in which racial mingling took place on a large scale and the assortment of displaced rural groups and ethnic mixtures was among the greatest in the empire. This environment therefore provides the opportunity to examine an unusual sweep of racial categories not present in most places in large enough numbers to merit separate study. At the same time, the size of Mexico City’s racially mixed population makes it possible to examine the less common intermediate racial groups in greater detail than can be achieved in cities of smaller populations.

Over the three centuries of the colonial era, there emerged in Mexico an intermediate group of people of mixed racial origin. The mixed-bloods were known collectively as castas, but a variety of distinct names was applied to each of the individual mixed combinations. The process of mixing began with the three major groups (Spanish, Black, and Indian). From these came three basic combinations: Spanish-Black, Spanish-Indian, and Black-Indian. The designation for the first combination was mulato (Spanish-Black); for the second, mestizo (Spanish-Indian); and for the third, at least initially, zambo or zambaigo (Black-Indian). All these terms applied in Mexico in the sixteenth century to the offspring of mixed ancestry.¹⁰Zambo and zambaigo never gained widespread currency in colonial Mexico and were seldom used by the end of the sixteenth century.¹¹ No other term emerged to replace them, although a variety of other designations was used sporadically in the colonial period to refer to offspring of mixed Black-Indian parentage (including the zoologically inspired lobo, or “wolf,” and coyote). No single appellation, however, emerged to refer specifically to this group. In the eighteenth century a plethora of fanciful terms were created for successive generations of Black-Indian offspring, including ahí te estás (“there you are”) and tente en el aire (“hold yourself suspended in mid-air”), but these were never used in ordinary communication.¹² By contrast, the first two terms, mestizo and mulatto, gained widespread currency in the seventeenth and eighteenth centuries as the principal means of identifying individuals of mixed racial ancestry. The five basic terms in use in colonial Mexico by the middle of the eighteenth century were Spanish, Black, Indian, mestizo, and mulatto.

In the seventeenth century, two additional terms appeared: castizo and morisco.¹³ These represented a refinement of the earlier terms of mestizo and mulatto. Castizo referred to a light-skinned mestizo and morisco to a light-skinned mulatto. These two terms introduced a second stage on a scale from each of the two darker groups (Black and Indian) to white. In this fashion, the continuum Indian—mestizo—castizo— Spaniard paralleled the progression Black—mulatto—morisco—Spaniard.

In the popular conception, race is identified with inherited physical characteristics. In relying upon race as a basic categoiy of social distinction, colonial society subscribed to the belief that the physical traits inherited by an individual formed an unchanging and objectively identifiable group of characteristics. It was thus possible to use racial labels to distinguish among individuals and to separate them into stable categories since the bases of these distinctions were presumed to be inherited and hence unalterable.

What was not apparent to members of colonial society was the extent to which their own perceptions influenced the system of racial classification. Although the presence of specific physical features is undeniable, it is not the physical characteristics themselves that are the crucial elements of racial identity, but their social perception and definition.¹⁴ The laws governing the inheritance of physical characteristics, for example, can produce a theoretically infinite range of colors, hair textures, and other features, but colonial Mexican society recognized only four intermediate shadings beyond the basic Black, white, and Indian. These shadings—castizo, mestizo, mulatto, and morisco—represented only a tiny fraction of the range of possible physical features; but it was only those four that were seen as different racial categories. The recognition of only four groups as separate depended on social selection of the relevant categories of groupings. This selective social attention focused on those groups that approached the dominant white group and in practice disregarded the remaining categories. The introduction of the terms castizo and morisco for lighter individuals of mixed parentage in the seventeenth century has been seen as an attempt to preserve white exclusivity and to maintain the boundary between white and mixed-bloods (castas).¹⁵

Since racial categories were selected by their social perception and identification, they were categories of social not biological race.¹⁶ In colonial Mexico the social race of an individual was related to the combination of physical appearance, economic status, occupation, and family connections, in other words, to his overall socioeconomic position as well as to physical features.¹⁷ A series of portraits of racial categories painted in the early eighteenth century reflects the importance of socioeconomic criteria in racial classification.¹⁸ Each of the racial groups was depicted in a social setting that included the dress, furnishings, home, and occupation thought to be associated with each racial category. Each racial group, including the more obscure categories, was characterized in this series of portraits by a different and unique style of dress and occupation, suggesting that members of colonial society may have believed that each racial category occupied a distinct socioeconomic niche.

To examine the reality of the relationship between race and socioeconomic position, the census of 1753 was selected.¹⁹ That census was an indirect result of rapid internal migration into the city of Mexico in the years after 1749. In that year a poor harvest and several severe epidemics struck the rural population of the areas near the city.²⁰ Faced with a shortage of food and an outbreak of disease, many rural residents made their way to Mexico City where those who survived commonly resorted to begging, working at odd jobs, and theft.²¹ The 1753 census was taken as a preliminary step in resolving some of the social problems produced by the inflow of immigrants.

The sudden invasion by Indians from the countryside placed a variety of strains on Mexico City. There was the problem of housing. Longstanding municipal ordinances required Indians to construct their dwellings on the outskirts of the Spanish section in one of two designated Indian areas. The massive influx of immigrants evidently so strained the supply of housing in the Indian areas that the homeless filtered into the Spanish area of the city. The unusual movement of Indian population into the traditionally Spanish section threatened social control over Indian population and prompted royal officials in May of 1753 to issue an order requiring all Indians not employed in bakeries, as servants, or as apprentices of master craftsmen to vacate the central city within eight days and to take up residence in the two specified areas.²² Recent arrivals from the countryside were ordered to return to their places of origin. It is doubtful that the orders had much impact on residence patterns since such regulations were periodically reissued whenever similar crises occurred.²³

Nonetheless, the terms of the decree are a help in understanding some of the limitations of the census. The census takers were not interested in the Indian residents of the area they surveyed, since from their viewpoint Indians did not rightly belong there. Hence the marginal notations on many blocks: “jacales of Indians.” Those Indians who were engaged in the occupations considered by the bureaucracy as proper to Indian residents of the central area (such as baking) were, on the other hand, studied extensively and with great care, as were Indian apprentices. Many of the Indians residing in Mexico City were water carriers and fuel suppliers and bearers, but these were required to return to their own segregated areas of the city at night. Thus, although they worked in the city’s core, they do not appear in the census because the census only recorded residents of the center.²⁴

Besides immigration, the second major problem leading to the census was the increase in urban crime.²⁵ Responsibility for the rise in crime was attributed to a group of people the royal government characterized as “beggars, vagabonds, loiterers, shiftless and illegally employed. ”²⁶ For better control of this unemployed or vagrant criminal element, whose numbers had been swelled by the influx of recent unemployed migrants from the countryside, the city council and royal government decided to improve the method of patrolling the streets. It was for this purpose that the city council members sought and received permission from the royal government to undertake a survey of the streets and the residents of the city.²⁷ Municipal officials revived an old plan to divide the city into sections, which could then easily be closed off by law enforcement officials for the more efficient apprehension of criminals.²⁸ The census was thus designed for a specific purpose (control of crime), and was implemented by the members of Mexico City’s council, part of the city’s political elite. The census therefore reflects elite perceptions of the city’s social and racial groups.

The extant parts of the census give a reasonably complete picture of three-quarters of the central, predominantly Spanish, area of Mexico City. These data, however incomplete, provide valuable information not available elsewhere in the social dimensions of race. The 1753 census of Mexico City is one of a few colonial Mexican censuses to provide detailed information on both race and occupation. For example, annual surveys of the parishioners of Mexico City’s parishes failed to note occupations and recorded the race of the residents sporadically (principally only toward the end of the colonial period). The necessary information classifying individuals simultaneously on the basis of race and occupation is, therefore, difficult to obtain, although it is crucial for analysis of the relationship between race and social position.

The 1753 census is valuable for yet another reason. It provides simultaneous information on occupation and on race while using a range of racial classifications found in few other censuses. Six basic racial categories were employed: Spanish, mestizo, castizo, mulatto, Black, and Indian. The census further distinguished between Spaniards born in the New World, designated simply “Spanish,” and those born on the Iberian Peninsula (for example, as “Spanish from Extremadura”). More important were the distinctions among the racially mixed population. In the next major civil census the category of Black disappeared entirely, leaving only the appellation mulatto for the population of African descent.²⁹ Differences or similarities between the Black and mulatto population cannot be examined. The most detailed later census, that taken in 1811, did not use the categories mestizo or castizo for the bulk of the territory covered in the census, but the word casta instead.³⁰ Since casta was the generic term for all the racially mixed population, it could be interpreted as castizo, mestizo, mulatto, or presumably any of the other racially mixed categories. Distinctions within the castas were obliterated, making it impossible to study differences among these groups.

In order to examine the division of labor in colonial society, it is necessary to consider the total organization of production rather than isolated aspects of the economic process. Such an examination, not surprisingly, shows that between those who owned the means of production (rural property holders and mine owners) and those who lived solely by their labor (servants and day laborers) there existed a fundamental social cleavage. This was the essential division of class. Classes can also be subdivided, however, on the basis of significant differences in their role in production. While these dissociations are not class divisions in the strictest sense, they are significant social divisions within the economic structure.³¹

In the dominant class were what might be called the residual elites, the various levels of religious, civilian, and military bureaucrats. These have been classified together with the rural property owners and mine owners as elite.³² A second significant group within the dominant class were the merchants and shopkeepers, the owners of the means of distribution. A final less important group within the dominant class were the managers (mayordomos) who did not own the means of production, but administered them for the owners.³³ In the laboring class, on the other hand, day laborers can be distinguished from servants, since both filled fundamentally different functions in production. Artisans, in turn, were divided into two classes, those who owned both shops and the tools of production and those who owned at most only the tools of their trade. The ownership of tools was important for the bargaining position of the worker on issues of wages, hours, and working conditions. The worker who was forced to use his tools as collateral for a loan had only his labor to sell, and at lower rates.³⁴ Since the artisans actually belonged to two separate classes, they have been subdivided accordingly. Those who owned shops were classified with the shopowners; the propertyless were simply labeled artisans.³⁵

Table I shows the occupational breakdown by race for free males in each group.³⁶ (Slaves, who were predominantly employed as servants in the houses of the elite, were excluded from this analysis.) Blacks excepted, a small number of each of the other racial groups were members of the elite. Not all elites enjoyed the same status. The creole and peninsular Spanish members of the elite were the major officials in the governmental and religious hierarchies, while the mulatto elites were leaders of their own segregated military units. Mestizo and Indian members of the elite tended to be lower level officials within the religious and civil bureaucracies, sacristans rather than priests, interpreters in courts rather than judges. Not surprisingly, creole and peninsular Spaniards were more numerous among the elite. A second group, the managers of production, were almost entirely creole and peninsular. The largest single group, comprising over 80 percent of all peninsulars, were merchants.

The well-known predominance of merchants among the peninsular Spaniards requires no further explanation here. Of greater interest is the distribution of creoles among merchants, shopowners, and artisans. Chance and Taylor’s study of Antequera found large numbers of creoles concentrated in the artisanal category.³⁷ By separating those artisans who owned shops from those who did not, the balance between merchant and artisan among creoles is substantially altered. Although artisans formed the largest single occupational group among the creoles of Mexico City, as they did in Antequera, the elite and merchant category together accounted for just over half of all employed creoles. Thus, although Chance and Taylor rightly call attention to the humble occupational pursuits of many creoles, Mexico City’s creoles were far from falling within an “ambiguous middle layer.”³⁸ Their most notable characteristic was their employment as merchants and shopowners. In this they resembled their parent group, the peninsulars, more than any other racial group. Creole merchants differed from peninsulars in that the former were more frequently retail merchants, while the latter controlled wholesale trade.

The next racial group, the castizos, were unmistakably artisans, and by contrast with the creoles were far less frequently merchants or even owners of the artisan shops. Not only was a castizo more likely than a creole to be an artisan, but he was far less likely to have owned the shop in which he worked. The greatest difference between creoles and castizos lay in the absence of ownership of productive property by the latter.

In both Chance and Taylor’s study of Antequera and Brading’s study of Guanajuato, mestizos constitute an “ambiguous middle layer” and, indeed, they appear to have occupied that position in Mexico City as well.³⁹ Substantial numbers of castizos, mestizos, and mulattoes were artisans: 72 percent of the castizos, 54 percent of the mestizos, and 43.6 percent of the mulattoes. Although at first glance this distribution indicates an undifferentiated middle layer, it is possible to distinguish among these largely artisan groups by seeing in what other occupations they were employed. Mestizos, for example, were more often laborers or servants than either creoles or castizos. Well over a third of all mestizos were laborers or servants, compared with 16 percent of castizos and 6 percent of creoles. While the greater numbers of mestizo laborers and servants distinguished mestizos from the castizos and creoles, the large number of laborers distinguished them from the mulattoes and free Blacks. Mestizos had the highest percentage of laborers of any of the intermediate racial groups. In this respect they resembled their parent population, the Indians, more than any other group.

The Indian population of Mexico City in 1753 was clearly differentiated from the other racial groups by the preponderance of laborers in it. The Indians were the bearers of water and wood and labored in the city’s bakeries. They were less frequently employed as artisans than any other group in the city, Blacks excepted. Given the primitive agriculture of the period, it should come as no surprise that rural male migrants to Mexico City were employed largely as laborers since these tasks were related to those they carried out in the rural areas. Contemporary rural migrants to Mexican cities often are engaged in similar occupations.⁴⁰ Only a third of the Indians were servants; presumably these were not the most recent migrants, but those who had become accustomed to urban life.

Many free mulattoes were artisans, as were some members of the other racial groups, including the Spaniards, mestizos, and castizos. What distinguished the mulattoes from the other intermediate groups was that some 50 percent of them were servants. More mulattoes than mestizos or castizos worked in the city’s homes and convents. Looking at the mulattoes’ parent population, the Blacks, we find an even larger percentage of servants (81.5 percent) and their virtual disappearance from the artisan group. The pattern of employment of mulattoes was closely related to that of the Blacks.

The explanation for this pattern of employment can be found in the role of slaves in Mexico City. Slavery in the plateau surrounding Mexico City was an urban institution. Blacks and mulattoes in Mexico, if slaves, were typically house servants; when freed they mostly continued to work in the same occupation. In the antebellum United States upper South, slavery was a rural institution and Blacks were largely unskilled agricultural labor. When freed and in an urban environment, they were employed largely in unskilled occupations.⁴¹ In colonial Mexico, Indians rather than Blacks performed the bulk of the unskilled rural labor and Indian migrants were the city’s laborers. Emerging from this predominantly unskilled group, mestizos were more likely than mulattoes to be thrust into unskilled labor. This reasoning suggests that the differences in employment between mestizos and mulattoes resulted from the different economic roles of the parent groups, urban slavery on the one hand and rural agricultural labor on the other, rather than any distinctive set of social attitudes in Latin America.

This explanation of the pattern of employment among men suggests an occupational division of labor founded upon several separate organizational principles. Indians and Blacks were at the bottom of the free population, but occupied different economic niches. Blacks were domestic servants, Indians were laborers, both relatively unskilled forms of employment. Men in the intermediate groups, both mestizos and mulattoes, were more likely to be skilled craftsmen than either of their parent groups. They differed, however, in their other employment, with mestizos more commonly laborers and mulattoes more commonly domestic servants. Clear differences separated these intermediate groups from the Spaniards. Creole Spaniards were craftsmen, but they were more often merchants and shopowners than were the mestizos, mulattoes, or castizos. In this respect they, too, were like their parent population, the peninsular Spaniards, who were overwhelmingly engaged in commerce. If we were to establish a hypothetical pattern of social mobility based on the occupational structure, it would contain two separate transitions. The movement from Black or Indian to mestizo or mulatto would have entailed becoming a craftsman. The movement from the intermediate groups to Spanish would have required not the acquisition of skills, but sufficient capital to establish a small shop. If Black or Indian upward socioeconomic mobility was achieved by the acquisition of a skill, the upward mobility of mestizos or mulattoes was attained by the acquisition of property.

Analysis of the census data suggests a complex relationship between race and the division of labor founded on the fundamentally different economic functions of agricultural labor, urban slavery, and commerce. It is instructive to view this picture briefly in another perspective, that of racial distribution within occupations. The results of this analysis (see Table II) suggest several additional observations. As expected, Indians dominated the category of laborers. Mulattoes dominated the ranks of servants since this was their largest single occupation and the free Black population was small in absolute terms. Also, and not surprisingly, creoles dominated the ranks of the elite as well as those of the merchants and shopowners. The category of artisan, however, also reflects a predominance of creoles. This bears further examination. Since creoles were the largest group in the part of the city here under examination, did they dominate the artisans simply because of their numerical predominance? To answer this question, the racial distribution of artisans was compared with that of adult males in the area. This comparison finds some differences in the racial distribution within the artisans. Indians, Blacks and peninsulars, the three poles of the city’s division of labor, were underrepresented in the craftsmen category; as expected, castizos and mestizos were somewhat overrepresented, while creoles and mulattoes were proportionally represented.⁴² More important, the artisans were substantially more representative of the city’s racial distribution than any other occupational group, and more closely mirrored its racial balance.⁴³

Within the artisans, some differences existed among racial groups. Spaniards enjoyed a near monopoly of a few crafts (such as silver-and goldsmithing) and were more often master craftsmen than were members of other racial groups. These differences, however, do not contradict the basic point that artisans were less obviously racially stereotyped than any other occupation in Mexico City. Tailors and shoemakers were the two largest categories of artisans within nearly every racial group from peninsular and creole Spaniard to mulatto and Indian. Slight differences existed in the percentages of tailors and shoemakers in each category. Among the Indians, shoemakers and tailors were behind carpenters, while among creole Spaniards, tailors were followed by weavers, with shoemakers in sixth place. This information suggests that the economic structure of Mexico City, with its obvious internal demand for the production of shoes and clothing, created the opportunity for the employment of mestizos and mulattoes as well as all other racial groups. Spanish labor was insufficient to fulfill demand, or unable, or perhaps unwilling, to monopolize production. In any case, the artisan category was relatively open to different racial groups. The opportunity for social mobility for Blacks and Indians was related to the availability of employment in craft production, and that in turn depended upon the economic organization of the city as a whole.

One additional characteristic of the occupational structure is worth noting. Two of the intermediate racial groups, castizos and mestizos, failed to dominate any of the occupational groups, while all other racial groups dominated at least one of the major divisions of labor in the city. Although a majority of both castizos and mestizos were artisans, they were represented in this category in numbers approximately proportional to their overall representation in the population. The absence of an exclusive economic niche within the division of labor for mestizos and castizos may have contributed significantly to their failure to become a definite social category. It suggests further that these two may have been the least identifiable racial categories in Mexico City.

To understand better the relationship between race and the division of labor, it is necessary to look not only at men’s roles, but at those of women and children. The percent of female participation in the labor force in 1753 shows significant differences by racial group.⁴⁴ (See Table III.) The group with the lowest percentage of working women was the Spanish creole population, with only 14.9 percent of those women in the census described as having an occupation. Twenty-one percent of the castizo women and 44 percent of the mestizo women were employed. The highest percentages of female participation in the labor force were among the Blacks (87 percent) and mulattoes (69 percent). Indian women occupied an intermediate position between the castiza/mestiza group and the mulatta/Black group. To make sense of these differences, the occupations held by women are examined.

Differences in types of employment accompanied the difference in employment percentages. (See Table IV.) Spanish women were alone in the elite occupations of schoolmistress and estate owner. They worked more often than women of other racial groups as shopowners and as spinners and weavers. It is interesting that the occupations of seamstress and spinner were more often activities of creole women than of other racial groups.⁴⁵ Only castizas came close to levels of creole employment in these “female” occupations. Creoles excepted, two-thirds or more of employed women were domestic servants. Even among the creoles, a substantial number of women were servants. Employment opportunities for women were clearly more restricted than for men.

A traditional alternative to domestic service for urban women was the sale of fruits, vegetables, and cooked food in the city streets. Such vending had been characteristic of Indian women from the earliest days of the colony, yet virtually no Indian women appear as food sellers in the census. The most obvious explanation for their absence from the census is that they resided outside the city or beyond the Spanish area. In either case they would not have appeared in the census, which covered only those who resided in the central area, not those who only worked there. The low percentage of participation in the labor force by Indian women, relative to Blacks and mulattas, likely reflects the census area rather than generally lower levels of employment among Indian women in the city. The lower percentage of employed Indian women in the city’s center, however, probably indicates the relative disadvantage they faced as recent migrants to the city. Domestic service requires an acquaintanceship with urban life unfamiliar to recent rural migrants. Furthermore, Indian women faced competition for domestic service jobs from the groups that traditionally dominated that sector, slaves and free mulattas.

The predominance of mulattas in the domestic service jobs associated with urban slavery can be seen more clearly if the occupation of women is analyzed by racial group. Mulattas appear as the largest single category, accounting for over 45 percent of all female domestic servants. (Spanish, mestiza, and Indian women constituted between 15 and 18 percent of all servants. Owing to their relative scarcity in the population, castiza and Black women together accounted for only 4 percent of all female servants.) The largest racial group an inhabitant of Mexico City in 1753 would encounter as domestic servants were mulattas. Schoolteachers, shopowners, seamstresses, and spinners would have been creole; and the street vendors, more than likely, Indian. The racial patterning of employment among women showed less variety than among the men, but still demonstrated differences among racial groups deriving from the division of labor in colonial society.

The age and employment of children by racial group also reveals significant differences in the division of labor. Among children aged 17 and under, Spanish or castizos were the least frequently employed and mestizos were less frequently employed than mulattoes.⁴⁶ (See Table V.) Relative levels of Indian and Black children’s employment, however, were somewhat anomalous. Black male children had very low employment levels; Black female children, the highest. On the other hand, Indian children showed distinctly elevated levels of employment. The number of Indian boys and girls working was roughly double that of mulattoes.

Indian boys and girls entered the work force far earlier than their counterparts in other racial groups. (See Table VI.) The mean age for all working Indian boys was a relatively low 11.1 years, and for Indian girls it was only 10.1. The mean age for all employed children of other racial groups ranged from 12.8 to 14.3. Indian children not only worked in greater numbers, but at younger ages compared to other racial groups.

To understand the differences in levels of employment among children, analysis of the options open to them is necessary. For boys there were two possibilities: apprenticeship and domestic service. A majority of employed creole, castizo, and mestizo boys were apprenticed to craftsmen, but nearly all employed Indian and mulatto boys were servants. Artisan apprentice contracts customarily began at age 12; domestic service could begin at any age. The earlier age of entry into domestic service partially accounts for the lower age of Indian children in the work force.

For girls, fewer options were available. Between 95 and 100 percent of all castiza, mestiza, mulatta, and Indian girls were servants. Only Spanish girls had the opportunity for employment outside the domestic arena, and a majority of Spanish girls were in the “female” trades of spinning and sewing.

The high rates of employment among Indian children and their relatively early entrance into the labor force probably resulted from a combination of circumstances. Early employment of children is customary in rural agricultural areas; such practices could be transferred to urban life. A second cause may have been the inability of Indian adults to find employment due to the flood of migrants produced by preceding years of epidemic and famine. Indian women were at a disadvantage in particular because of competition from slave labor and from free mulattas. Children’s earnings may have compensated for the inability of adults to find employment under these conditions, and contributed substantially to the family income.

For Black and mulatto boys, the employment possibilities were relatively slight. As residents of the city, it is likely that they aspired to artisan apprenticeships, and their limited participation in the work force probably reflected their inability to obtain such contracts. The association of urban slavery with domestic service allowed adult mulattas readily to obtain jobs as servants, but worked against young men seeking employment as artisans.

Sexual differentiation also influenced female and child employment levels. Rather few creole women worked; when employed, they were more frequently in an elite occupation, owning shops, spinning, or sewing. Substantially more mestiza, Black, and Indian women worked, and almost exclusively as servants. Castizas most closely resembled creoles in their relatively low levels of employment. The resemblance between castizas and creole women carried over into the employment patterns of girls from these groups. Among the Spanish and castizo children, boys were more frequently employed than girls, while among mestizos and mulattoes, boys and girls had nearly equivalent rates of employment. At the other end of the racial spectrum, both Black and Indian girls were more often employed than boys of either group. Sexual differences in employment were apparent at either end of the social structure. Few Spanish and castiza women and girls worked; most Black and Indian women did, as did most Black and Indian girls. The mestizos and mulattoes were clearly intermediate groups in this respect; nearly equal numbers of boys and girls from these groups worked. This evidence suggests that only Spanish and castizo families could afford to support girls and women who did not work. In all other families, females had to contribute to household income.

The trends visible in female and child labor force participation are similar to those seen in male employment, and also add to our understanding of the male patterns. Urban Black slavery and Indian rural agricultural labor were the determining influences on levels of female and child employment. Free mulatta and Black women were employed in the domestic service jobs associated with urban slavery. Mulatta women dominated this job market. Indian women as recent migrants were less well acquainted with urban life and were less frequently employed as servants. Indian women residing outside the city’s center sold food on the streets. For Indian families, high levels of juvenile employment probably compensated for lower income from female labor. For Blacks and mulattoes, the situation was reversed, with low levels of child employment for male children and high levels of adult female employment probably resulting from the difficulties encountered by young Blacks and mulattoes in obtaining apprenticeships. Spanish and castizo households were characterized by low levels of both female and child labor, presumably since the income of creole men was sufficient to carry women and children as economic liabilities. Black and Indian families could ill afford such a luxury.

Economic need may have played a substantial part in shaping the employment rates of women and children, but it did not succeed in obliterating the distinctions among races. The way in which each racial group resolved its economic problems was limited by the economic structure of the larger society. Social differences were not eliminated by economic hardships, but rather were the basis of different strategies for the resolution of those difficulties.

Since racial groups were categories of social rather than biological race, they were not indelibly fixed by birth. Instead, social race was determined by a combination of physical appearance and economic position. In such circumstances, variability in racial labeling was inevitable, reflecting differences in the relative weight different individuals attached to physical appearance and economic position. The critical question, however, is how much agreement there was on the racial categories. To what extent did members of colonial society agree about who was a mestizo, castizo, mulatto, or Indian?

Difficulties in distinguishing social races were well known in colonial society. One seventeenth-century observer remarked that there was no way “to distinguish the mestizo from the pure Spaniard except the hair, which is not modified for two or three generations. Otherwise there is no difference, not even in the features of the face nor in the form of the body nor in the way of speaking or pronunciation.”⁴⁷ If there were so little physically to distinguish an individual, how readily were categories of social race perceived? Another observer suggested that all an Indian had to do to become a mestizo was to comb his hair, wash his face, cut his nails, and put on a clean shirt.⁴⁸

Changes in social race are sometimes called “passing.” “Passing,” however, has normative connotations; it suggests movement in one direction, toward “white.” Our interest is broader than that, extending to movement in any direction. Consequently, we will use the term variability to refer to changes in racial labels.

A major stumbling block in the analysis of variability in labeling racial groups has been the difficulty in going beyond casual anecdotes to systematic analysis. To resolve this problem, a new approach was needed. The method chosen for measuring the degree of variability in the perception of racial groups was to contrast the reporting of race in the 1753 census with the judgment about race under another set of circumstances or conditions. A number of documents were examined to find a set of reports that would facilitate a comparison with the individuals examined in the census. It was essential to make the comparison as systematic as possible in order to avoid the pitfalls of anecdotal comparison. Experience indicates that the best method of assuring a systematic comparison was to examine a list of individuals from the same area of the city who also had been classified by race, and to check their names against the census. By linking two sets of lists through the use of names (a process called nominal record linkage),⁴⁹ a basis for the systematic comparison of racial reporting under two different circumstances was established. After a survey of several different kinds of lists, I found the marriage register for the parish of the Sagrario best suited for the purposes of the study. (See  Appendix II.)

Examination of the 1753 census with the procedure just outlined turned up a total of 106 couples, out of a total of 831 marriages taking place in the Sagrario parish between 1752 and 1753. This signifies an overall trace or location rate of 12.8 percent, approximately half the rate for nominal record linkage in other historical studies.⁵⁰ Part of the explanation for the relatively low trace rate resides in the stringency of our requirements for a match, part in other difficulties.

The Sagrario’s registers in 1752-53 were divided into two; one section for Spaniards and one for castas, or mixed-bloods. Nearly twice as many Spanish as casta marriages were performed; not surprisingly, more Spaniards than castas were matched with individuals in the census of 1753. The proportion of Spaniards in the linked population was also increased because the trace rate for such individuals was double that of the castas (8.1 versus 16 percent). The low trace rate for castas was one cause of the low overall trace rate. The low rate for castas in turn may have resulted from several influences. First, Spaniards may have been more stable residentially than castas, and hence more easily located by the census takers. Second, the census version used for matching was weighted toward household heads. These were largely Spaniards, so castas residing in Spanish households were not always cross-listed. A third possibility is that the census may have underrecorded the population of mixed racial origin. This discrepancy is hardly unusual, since even the most sophisticated modern censuses have substantial undercounts of the Black population.⁵¹

The amount of agreement on racial categories is shown in Table VII. As expected, the highest number of matches and the greatest amount of agreement are within the Spanish group. Seventy-seven of the 108 individuals traced are Spaniards; disagreement exists on 11.7 percent of this group. For the racially mixed population, disagreement is 16.1 percent, and the overall level of disagreement is 13 percent. These are statistically significant differences, meaning that the racial misclassifications cannot be attributed to random chance.⁵² To understand how and why these differences were produced, close inspection is needed of the numbers themselves.

Of the eighty-four individuals classified as Spaniards on the parish register, four were listed as mestizos on the census and two were classified as castizos. Only one Spaniard fell into the mulatto category; no Spaniards fell into the antipodal Black or Indian groups. There were thus no extreme examples of misclassification, but rather significant shifts between the intermediate racial groups (castizo and mestizo) and the Spaniards.

For the Black and Indian population, slight information was available. Only three individuals classified as Black on the census could be identified from the marriage record, but one such Black woman was listed on her marriage record as a mulatta. The disagreement between the marriage and census was over a difference between Black and the nearest lighter category, mulatto. No Blacks were misclassified as either white or Indian.

For all practical purposes, no matches could be made on the Indian population. Although there was perfect agreement on the lone Indian identified on both records, the individual was himself exceptional, the owner of a barber shop, and qualified as a master barber. The absence of matching on the Indian population can be attributed to several defects in the census itself. The first has already been alluded to, the deprecatory references to “jacales de indios” located within the Spanish area and the lack of careful counting of individuals residing in those jacales. The second defect is a more general one and has to do with the refusal of the census takers to refer to the Indians, with perhaps unconscious condescension, by anything other than first names. Parish registers more often listed both the first and last names of these individuals. This would suggest that perhaps some other form of linkage (within parish records themselves, or, for example, between marriage and birth records) might provide a useful alternative strategy to marriage-census linkages for testing the ethnic identification of the Indian population.

Difficulties resulting from first name usage were not a significant barrier to the tracing of the Black population in Mexico City because the Blacks, even when identified only by first names, were also listed as the servant or slave of a local resident, making them relatively easy to locate. In rural areas, the use of first names for Blacks might prove a severe hindrance to tracing, and some other strategy, such as the one suggested for Indians, might prove advantageous.

Disagreement was substantial on the racially mixed categories of mestizo and mulatto. Four of those termed mestizos in the census were listed as Spaniards in the parish records; one was listed as a mulatto. The dominant pattern of mestizo reclassification was into the lighter (Spanish) category. Of those described as mulattoes in the census, one appeared as a Spaniard, one as a castizo, and two as moriscos in the parish registers. The general tendency was for lighter classifications (morisco and Spaniard) to be attributed in the marriage registers to mulattoes as well as mestizos. Thus more mestizos and mulattoes were reclassified in the parish registers as lighter (mestizo as Spanish, mulatto as morisco or Spanish) than were reclassified in other racial groups. Although the numbers are small, they suggest that marriage registers showed a tendency toward the recording of a lighter racial status than the census.

Although both mestizos and mulattoes were listed as lighter in the parish registers than in the census, only one mulatto was classified as Spanish, contrasted with four mestizos, indicating that it might have been easier for a mestizo to pass for Spanish. Mulattoes who were classified differently were more often listed in the marriage registers as moriscos (light mulattoes) than as Spaniards.

The tendency to lighter classification in matrimonial records was also reflected in the classification of an entire group of people, the moriscos. By definition moriscos were the light-skinned mulattoes and hence the people closest to Spanish or white. In the parish books, “morisco” was the second most commonly used designation for a person of African origin (17.6 percent of all persons of African descent); but in the census, “morisco” was used for less than 1.6 percent of the population. In effect, the parish records recognized the existence of a light-skinned mulatto group, whereas census takers did not. It is not surprising, therefore, that no census moriscos appeared in the linked data and the two moriscos listed in the parish registers were both identified as mulattoes on the census.

Several instances, however, contradict the general tendency toward lighter classification in the marriage records. In one instance, the parish priest recorded a woman as a castiza, the census taker as a Spaniard. The reason for this discrepancy appears to lie in the circumstances in which the marriage ceremony took place. This woman was not married at home or in a church as was the normal rule, but in jail, where she had been incarcerated for concubinage. Since elite social prejudices of the day suggested that the criminal element was dark-skinned, it is perhaps not surprising that a light-skinned woman in jail was seen as a castiza. Released from jail and back in her normal environment, she was recorded as white by the census taker.

Another exception was a tailor who married a mulatta in 1752 and was recorded as a castizo by the priest. When the census was taken the following year, he was recorded as a Spaniard. In labeling the man a castizo, the priest or notary reduced the social distance between the mulatta and her husband, an action suggesting the influence of elite prejudice against marriage by Spanish men to mulatta women.⁵³ Spanish elite prejudices about the marriage of light men and dark women and about the socioracial identity of the criminal element altered the usual upward tendency of racial classification in marriage registers.

A final instance of disagreement illuminates the same pattern in different circumstances; a woman listed as a mulatta at marriage was recorded as a mestiza on the census. This shift, in contrast with the two previous examples, was not a directional shift to a lighter classification. Mestizos and mulattoes were both intermediate racial groups originating in different groups in the social structure. When recorded as a mulatta, the woman was marrying a mulatto male who was about to die. On the census taken after her husband’s death, she appeared as a mestiza. Like the previous examples, however, this instance also reflected the tendency to reduce the social distance between spouses. In both cases, it was in a marriage to a mulatto that the gap was noticeably narrowed. Narrowing the social distance when one of the parties was a mulatto reflected Spanish beliefs that Blacks and mulattoes ought to marry each other rather than individuals outside the group.⁵⁴ The difference in classification in the marriage register reflected the attempt to reduce the discrepancy between the anticipated and actual behavior in favor of the former.

The trend toward lighter classification was not the only pattern of disagreement within the categories of mestizos and mulattoes. By conventional definition, both should have been readily identifiable as the offspring of separate socioracial groups: mestizo (white-Indian) and mulatto (white-Black). In the linked data, one mestizo was listed as a mulatto and a mulatto as a castizo. The cross-over between mulatto and mestizo/castizo suggests that the distinction between the two was sometimes difficult to maintain when both occupied the same position within the division of labor.

In addition to general differences between marriage and census records, and elite prejudices about both the socioracial identity of criminals and the proper marriage partners for Blacks, there is a third source of racial variability. Since race and the division of labor were significantly related, variations in racial labels might be attributable to social mobility. If this were the case, we would expect differently classified individuals to occupy positions in the division of labor that differed from those of the majority of their racial group. To make this comparison, the occupations of the linked individuals were contrasted with those of adult males in the census. Excluding individuals about whom there was disagreement on race, Table VIII shows a close relationship between the linked data showing racial agreement and the census. Linked data for Spaniards, mulattoes, and Blacks paralleled the census well. Only in the case of mestizos was there significant deviation from the pattern of employment in the city, since all the traced mestizos were artisans, whereas in the census only half of them were so described. Mestizos excepted, the linked data replicated the occupational structure fairly well.

The population about which there was disagreement on race did not fit the occupational structure nearly as well. Of the three mestizos listed as Spaniards on the marriage registers, one was a muleteer, another a carpenter, and the third the owner of a tobacco shop. Neither the carpenter nor the muleteer can be considered occupationally mobile; other reasons must account for variations in their racial labels. The mestizo owner of a tobacco shop, however, belonged to a category usually associated with creoles, since less than 7 percent of all mestizos were shopowners. Social mobility was a factor in this difference in racial labels.

The only mulatto to be listed as a Spaniard on the parish registers was a painter. Although this was an artisan’s craft, it was one of the few crafts whose membership was almost entirely Spanish. Since few mulattoes entered this craft, it is not surprising to find this man labeled as a Spaniard on the parish register.

Social mobility was also responsible for differences in classifications among castizos and moriscos. The individual classified as a castizo in the parish register and mulatto on the census was named Juan Antonio Chávez, and was unusual in other aspects besides the disagreeement over his race. Chávez was the owner of a small bookstall, an occupation that classified him as a shopowner. Only 5 percent of all castizos and 2 percent of all mulattoes were shopowners. Furthermore, Chávez owned a bookstall, which suggests that he was literate. Literacy was unusual even among the Spanish population. Both Chávez’s literacy and his occupation distinguished him from the majority of either castizos or mulattoes. Failing to fall into the conventional occupational role of either mulattoes (servants) or castizos (artisans), he was by definition upwardly socially mobile.

The two moriscos indicated by the parish register were both artisans; one was a master carpenter, the other a journeyman tailor. Neither was a servant, the occupation of most mulattoes, but both were given this racial label by the census takers. Since 44 percent of mulattoes were artisans, occupation alone indicates only a limited degree of mobility. As artisans, however, one man was exceptional, enjoying the status of master carpenter. Few mulattoes received this appellation. Relative to the majority of the mulatto group, the master craftsman was upwardly socially mobile and was labeled morisco in the parish register.

Social mobility was clearly a consideration in differences in racial labeling. Several castizos, moriscos, mulattoes, and mestizos about whom there was disagreement in racial labeling occupied positions in the social division of labor that were substantially different from those of most of their racial group. Such differences account for a variety of instances of racial mobility, but are not the sole source of such differences. Differences between marriage and census records, influences founded on elite social prejudices, and social mobility were together the principal sources of differences in racial reporting.

This analysis leaves unresolved another issue in the analysis of racial variability, the general level of agreement on different racial groups.

The statistical device designed to measure agreement between two observers who categorize items into the same set of nominal categories is Cohen’s k (kappa).⁵⁵ The index of overall agreement (kappa) ranges from 0 to 1, and when calculated for the data joined through nominal record linkage, is .69.⁵⁶ This figure indicates agreement on the racial categories, but is far from indicating the existence of complete unanimity. Of perhaps greater interest than overall accord is the agreement on each of the separate racial categories. Was there greater or lesser agreement as to who was a mestizo compared to accord on who was a Spaniard? Can we rank order the racial categories to ascertain on which racial groups there was the most accord and on which there was the least? To do so we can calculate partial kappas,⁵⁷ with the results shown in Table IX.

The rank order that results shows the greatest agreement between observers as to who was a Spaniard, the least as to who was castizo or morisco. The absence of concurrence on the moriscos stems from the failure of census takers to use this term for light mulattoes. It is important to note that the two groups that were closest to white, the castizos and moriscos, were the groups about which there was the least agreement. If these categories were created in the seventeenth century to maintain the boundary between white and mixed-bloods, and to preserve white exclusivity, they were failing to do so by the middle of the eighteenth century. These two terms were failing, however, in different directions. Castizos about whom there was disagreement fell into the white (Spanish) group, suggesting that the boundary between castizo and white was disintegrating. The category morisco was disintegrating in another direction, as moriscos were reclassified not as white but as mulattoes. Boundaries between castizo and white were disappearing, but those between mulatto and white were being strengthened. These conclusions are borne out by another aspect of the rank order. As a group, the mixed population of Indian origin (castizos, mestizos) was less readily identified than the population of Black or mulatto origin.

Several explanations are plausible interpretations of these data. Physical clues, those historically impenetrable characteristics of hair, lace, and skin, may have been more readily identifiable among the population of African origin than among those of Indian origin. But while physical traits may have contributed to identification, a still more persuasive explanation derives from the relationship between race and the division of labor. Of all the racial groups in the city, the only ones not occupying a distinctive position in the social division of labor were the castizos and mestizos. Both groups were artisans, but their representation in the artisan population was closely related to their numbers in the total. Despite their overlap in the artisan category, all the other racial groups were identified with a distinct position within the division of labor. The castizos and mestizos alone failed to have a unique economic niche. Since the racial categories were those of social, not biological, race, the absence of a unique position in the division of labor signified a reduced capacity to isolate and identify these racial groups.

The principal findings of the research presented in this article may be summarized as follows. The relationship between race and the division of labor in Mexico City was founded on the essentially different economic roles of Indians, Blacks, and peninsulars. Each of the three racial groups was associated with a specific economic task in the social organization of production: Blacks with domestic service, Indians with other forms of labor, and Europeans (whites) with wholesale trade and commerce. Men in the intermediate racial groups were more likely to have been artisans than any members of their parent racial groups. Despite their common employment as artisans, the intermediate racial groups of Mexico City did not form an undifferentiated middle layer, but rather showed degrees of difference that reflected the social origin of the parent groups in the social division of labor. Some creoles were craftsmen, but they more often were merchants and shopowners than members of other intermediate racial groups, resembling their parent group, the peninsular merchants. Mestizos and mulattoes were more likely to be artisans than either of their parent populations, the Blacks and Indians. They resembled their parent populations (and differed from each other) in the larger number of mestizos working as laborers and the greater number of mulattoes becoming servants. These differences originated in the division of labor between Black urban domestic slavery and Indian agricultural labor, with the racially mixed offspring of Blacks more often being servants and the offspring of Indians more often laborers.

Employment among women was also related to this broad pattern of economic organization. Black and mulatta women, like their male counterparts, were associated with the domestic service jobs characteristic of urban slavery. Indian women, more recently arrived and less acquainted with urban life, faced competition from slave labor and were less able to find jobs as servants. Sexual differentiation was clearest among Spaniards and castizos, with women in both these groups less frequently employed and, when working, employed in a wider range of economic activities than women from other racial groups. Few girls in either group worked, suggesting that these families could afford to have women and children as economic liabilities rather than as contributors to family income.

Racial identification and variability in racial labeling were also related to the division of labor. Creoles and peninsulars, because of their domination of commerce, occupied a distinctive economic position and were readily identifiable as a racial group. Blacks and mulattoes continued to occupy a fairly well defined economic niche because of the persistence of urban slavery and were still identifiable as a racial group in the middle of the eighteenth century. Urban slavery thus preserved distinctions between creoles and mulattoes rather than permitting them to become blurred. Mestizos and castizos were the only racial category to occupy an ambiguous position in the division of labor. Like other intermediate racial groups, they were principally artisans, but unlike other groups, failed to dominate a single major economic category. Consequently, they showed a significant degree of racial variability.

The racial groups of Mexico City were related to significant differences in the division of labor in colonial society. Racial terms were cognitive labels attached to different groups in the economic organization of production.⁵⁸ The labels stressed the degree to which the mixed population approached white; the economic structure showed continuities with the tasks associated with the original tripartite division of labor in colonial society. The reality of social race was formed by the intersection of the division of labor and the cognitive system of racial labels. When they coincided, the level of agreement on social race was high; where they diverged, when specific economic tasks were not associated with racial labels, there was little agreement. Race and the division of labor continued to be associated in the middle of the eighteenth century, but the boundaries among racial groups were disintegrating, as the separation grew between the cognitive system of labels and the economic division of labor. Economic tasks had always been associated with racial labels; as the colonial economic structure dissolved, so did the racial labels it had imposed.

Caste and Class: The Non-Debate

It may be noticed that this article has avoided mention of a traditional debate commonly associated with studies of race and the division of labor, the debate between “caste” and “class.” The reasons for this are two: the definition of these terms, and the concepts behind these definitions that have established the nature of the debate. The American sociological tradition associated with “caste” and “class” was established by William Lloyd Warner in 1936.⁵⁹ Warner defined caste as a social group closed to entry through either marriage or socioeconomic mobility, and class as relatively open to intermarriage and social mobility.⁶⁰ Studies of Latin American social structure have relied implicitly on Warner’s dichotomy to characterize the class system as incipient at the end of the colonial period,⁶¹ or to contrast the two systems as opposite.⁶² Neither term, however, can be seen as absolutely opposite except theoretically since the caste system will have elements of mobility and the class system elements of inheritance of position.⁶³ So, the debate over caste and class resolves itself into the question: how closed must a system be before it is a caste system, how open before it is a class system? These arguments trivialize the discussion of race and class by reducing it to an argument over degrees.⁶⁴

Several historians have eschewed the word “caste” and used “estate” instead. The concept of estate is Weberian in origin, and the word itself is the English translation of Weber’s Stände.⁶⁵ While Weber’s concept has a dual meaning in his own work, his American interpreters, such as Gerhard Lenski, have generally characterized estates as legally defined segments of the population with rights and duties established by law.⁶⁶ John Chance and William Taylor selected Lenski’s definition of estate and mentioned the legal rights and duties associated with the racial labels. In practice, however, they actually use “estate” as a synonym for “closed,” a usage that resembles Warner’s definition of “caste” more than anything else.⁶⁷ Their discussion changes the labels from caste and class to estate and class, but does not alter the basic terms of the argument.

A nearly identical formulation of the word “estate, ” also falling within the confines of Warner’s caste and class distinction, is Lyle N. McAlister’s “American system of estates. ”⁶⁸ McAlister’s construct, based upon social and juridical statuses, was also derived from Weber, and received further elaboration by Magnus Mörner.⁶⁹ Both McAlister and Mörner, more than Chance and Taylor, stressed the centrality of judicial differences as the basis of social organization. While deriving their definition of estates as legally defined groups from Weber, both McAlister and Mörner overlooked Weber’s observation that legal privilege is the manifestation of stability in the definition of economic power.⁷⁰

The use of the term “class” in studies of Latin American social structure has related difficulties. The inspiration for most usages is clearly Weberian, since class is defined pluralistically. Theoretically class is conceived as an amalgamation of different patterns of consumption (such as residence) or the result of distribution of the fruits of social production (wealth or income). While this article specifically rejects such an approach on theoretical grounds,⁷¹ there are significant methodological problems with this focus. For most writers the measurement of distribution is a scale of wealth or income. This scale is subdivided into strata; hence the term social stratification, which is frequently applied to this approach. Strata, however, are not the same as class. Strata can be divided and subdivided indefinitely until the original point of the investigation has been lost entirely.⁷² In order to be useful, class must express a meaningful social unit, not an infinitely small subdivision of a measurement scale.⁷³ Multidimensional analyses multiply the varieties of strata, but fail to alter the approach to social analysis.

A second methodological difficulty with the use of the strata approach is practical. Rarely is consistent information on income or wealth available historically for any but a small fraction of the elite. This produces a double standard for historical analysis: “multidimensionality” is applied to the elite; “unidimensionality” is applied to lower groups.

In recent years, the debate over estate and class has become centered on issues of measurement and the interpretation of statistical indicators of degrees of intermarriage or social mobility. In the process, more fundamental issues have been sidetracked. The variety of economic structures in different areas of Latin America and the relationship between race and the division of labor within those economic structures all are important issues in understanding the colonial social structure. The blame for sidetracking these issues does not lie with modern methods of computation, but rather with the theoretical assumptions that have made issues of degree appear fundamental.

Nominal Record Linkage

The method followed in linking names from the census to the parish registers began with the identification of all couples who had married in the year and a quarter immediately preceding and the eight months following the census in the parish of the Sagrario, the same area surveyed by the census takers in 1753. The marriage register constituted a reliable basis for comparison for a number of reasons. Recently married individuals are for the most part a relatively stable group (not highly mobile geographically). This would make them easy to find. Second, they can also be presumed to have been relatively healthy (near the biological prime), so that their mortality rate would eliminate only a few of the potentially linkable population from the group. A third reason for selecting marriage records was more practical. Such sources provide greater information on the individual than do other parish records, especially birth and death records. (Birth records for the area often fail to mention the parents’ names and are complicated by both the high mortality of young infants and the underregistration of births. Death records are even less useful choices for nominal record linkage because they report only the name of the decedent and in most cases fail to identify even the nearest surviving relative.) A final reason for selecting the marriage records derives from the nature of the procedure used. Since nominal record linkage relies on the accurate use of names, it was important that the records carefully report both the husband’s and the wife’s names. Since marriage joins two separate individuals, the nature of the institution requires an accurate report of the last names.

First names are, by themselves, not a good guide to accurate record linkage. One example may serve to illustrate this point. In 1740, five brothers in Texcoco requested that they be exempted from tribute. To prove their status as nontributary Indians, the brothers presented, among other items, copies of their birth certificates, of which four could be located. Their mother’s first name was recorded in successive birth certificates as María Juana, María Francisca, María de la Cruz, and María Josefa.⁷⁴ To be sure, one of her first names was María. Beyond that, the rest was conjecture. In this instance the last names matched far more reliably than the first names. For this reason, record linkage relied primarily on agreement on the last name.

The basic procedure followed in linking names from the census to the parish register began with the identification of all couples who married in the parish of the Sagrario around the time of the census. The names of all couples married between January 1752 and December 1753 were recorded and alphabetized, first by the husband’s last name. Then the alphabetized census list of 1753 was searched to identify the couple on the list. All linkages were made by hand (computer linkages not being cost effective), and four separate searches were made through the alphabetized census list, each entailing a realphabetization of the names on the marriage register. The first search was by husband’s last name, the second by wife’s last name, the third by husband’s second last name, the fourth by wife’s second last name. The primary criterion for accepting or rejecting a match was the agreement between census and marriage records on both last names of the couple. Thus for example, Don Vicente Blanco, who married María Petra Estrada on January 7, 1753, was linked to Vincente Blanco married to Petra Estrada and residing with her parents on Calle Tacuba when the census was taken.

This process of linkage was easy for the northwest and the southwest sectors of the city, since census takers in both areas were generally consistent in recording the names of both spouses. In the remaining sector (the southeastern quadrant) linkage was more difficult because the most common notation was simply the name of the man, José García, for example, and his marital status (married) but without mentioning his wife’s name. By the criteria used, this was not a match to a Sagrario marriage record of a José García. While the requirement for two names may have reduced the number of matches, it made those matches accepted a more reliable basis upon which to study the issues that concern us here. The difference between this and other sectors stemmed from its having different census takers rather than from substantive underlying differences in the surveyed population. Matches were made in the southeast quadrant in the absence of a wife’s last name when two additional conditions were met: if the husband’s last and first names were both relatively uncommon (don Fernando Páramo, for instance), and if the indicators of social status might be reasonably associated with the occupation listed in the census. These two backup conditions for identification were used only for the southeastern quadrant of the city where matching by husband’s and wife’s last names was impossible. Since our concern was with racial variability, it was deemed critical not to overestimate the degree of variability through unreliable matches. Hence, our figures are conservative estimates.

Author notes