IV Conferencia Iberoamericana sobre Familia
Cartagena de Indias, 9 de septiembre, 1997

Click here to start Powerpoint Web show

The history of family and gender in Latin America has grown rapidly in recent decades both in the number of studies and in the great diversity of lines of inquiry (Gonzalbo 1992; Arrom 1992; Dueñas Vargas 1995; Rodríguez Saenz 1995; Rodríguez Jimenez 1996; Dore 1997; Lavrin 1997). Now, the time is ripe for a renovation of older approaches with more broadly conceived problems, expansive studies, sweeping chronologies, sophisticated methods, and bolder conclusions. Contemporary social concerns should be addressed as well as the imperatives of the historical profession. Before exhorting others to risk their careers on an untested venture, as an example I offer my interpretation of the history of the family in Mexico from the first moments of contact with Europeans to 1990.

The case for renewal is compelling for demographically-oriented studies of family and gender. Precisely because of past successes, the yield from conventional methods diminishes with each new study. The two most important methodologies used by demographically-inclined family historians--French family reconstitution and English household and family studies--have run their course in Latin America.

The "royal road to historical demography," as Rabell (1990:8-9) calls the French "family reconstitution" method, proved to be a dead end in Latin America, aside from the occasional success (Klein 1986). Morin sagely cautioned us of this possibility a quarter century ago (1972). The family reconstitution method is a rigorous procedure for constructing and analyzing family genealogies of ordinary people. The five conditions required for a successful journey on the royal road--stable family names, a small parish, a series of births, marriages, and deaths (or baptisms, marriages and burials) faithfully recorded week-by-week for a century or longer, low illegitimacy, and little migration--characterizes few, if any, of the thousands of parishes in Latin America. At best, reconstitution studies inform us about levels of legitimate fertility, legitimate marriage, and legally recorded infant and child mortality. Unfortunately in Latin America, legitimate events constitute a minor fraction of the demography of family life, and to reconstitute even this fraction requires enormous labor, resources, and leaps of faith. Few of the published Latin American studies even acknowledge using conventional methods for estimating error nor do they report correction rates. Rarely is the power of the method focussed to illuminate gender history or social history in general.

The English method of studying family and household from census lists proved to be more productive in Latin America (Cook and Borah 1971). From a list of inhabitants, the researcher can rather quickly compute the average household size, the frequency of various family types, the occurrence of female headed households and a plethora of other measures (Laslett and Wall 1972; Laslett 1993). Unfortunately, the English approach--as it was implemented by historians of Latin America--, if not a dead-end, proved to be something of a wrong turn. Its central thesis that the average family was always small (4.75) because the nuclear family was the norm for ordinary people in the past set researchers to work on the wrong question--size and structure of families and households rather than the family and household contexts in which people lived--, using the wrong unit of analysis--households rather than individuals--, and the wrong measures--averages rather than proportions. The English method of analyzing census lists evolved over time (Wall 1985), but unfortunately its application in Latin America did not. Household lists are valuable sources, when they are used to address interesting questions using germane categories and appropriate techniques. Fortunately there are a great number of family listings in Latin America and, properly used, they will teach a great deal about the history of family and gender.

I offer the following exploratory analysis of census lists from 1540, 1777, and 1990 to suggest different turns that researchers might take with this source. The essay begins with the conventional approach, units of analysis, and measures of the English school using an Aztec census from 1540. Then, the discussion is reoriented from the household to the individual in the household context to illustrate how census lists may be used to address questions of the gendered dimensions of marriage, family, identities (personal names), and work. My discussion of the second period is based on Rabell's interpretation of a census list of almost 20,000 people from 1777. This example illustrates how we may gain a better understanding of the contingencies of race, gender and family by studying individuals in family contexts over the life course (for a conventional approach see Klein 1996). Findings for these early periods are compared with individual-level data from the 1990 national census sample. In addition to 1990, national samples exist for 1960, 1970 and perhaps 1980 in the case of Mexico (McCaa 1997). Modern census samples, like those of Mexico, are likely to become one of the most valuable sources for studying the family transition in Latin America over the closing decades of the millennium.

For the early modern world some of the richest household listings are from Central Mexico, dated between 1537 and 1544--before the spiritual conquest or the accompanying demographic disaster was far advanced. These remarkable documents, written on fig bark "paper" in Nahuatl by native scribes using authentic Nahua concepts detail name, age, marital status, relationship to head, land holdings, and tax obligations for thousands of ordinary Nahua families (Cline 1993:3-8). Complex families, including joint or multiple families living around a single patio, were the rule for ordinary people in pre-hispanic Mexico, as the ethnohistorian Pedro Carrasco proved more than three decades ago (1964a:373). The maze of kin ties in early sixteenth-century Nahua households was revealed by the work of Carrasco and others disseminated in English, Spanish, and German as well as Nahuatl (e.g., Carrasco 1964a, 1964b, 1976, 1993; Hinz et al., 1983). Ignored by family historians Carrasco's findings handily disprove the uniformitarian creed espoused in the revisionist's bible Household and Family in Past Time: that the nuclear family was always the norm for ordinary people everywhere and for all time (Laslett and Wall, 1972:xi; for contrasting views see Flandrin 1979; Kertzer 1982; Wall 1982; Anderson 1985; Segalen 1986; King and Preston 1990; Hareven 1991; Smith 1993; Ruggles 1994; and Anderson 1995).

This founding myth of revisionist family historians was based on the fallacy of demographic determinism: that in the past high mortality destroyed three generational households before they could materialize (Levy 1965:49). Carrasco's data readily refute this dogma. Nahua mortality in the sixteenth century was worse than anything in Europe since the Black Death, yet many ordinary people in Ancient Mexico lived in complex multi-generational households. Lists for the villages which I have studied, Huitzillan and Quauhchichinollan, reveal 49.6% of the population residing in vertically extended units of three generations or more. Among two generation households fully three-fourths contained lateral extensions. Only 6.3% of all households were "simple," consisting solely of parents and children (Cline 1993; my calculations).

The key to household composition was, and is, culture, society or economy, not simply demography. The fact that expressions of kinship remain essentially comparable in Mexico--even with the passage of five centuries and the turmoil of conquest, colonialism, catholicism, liberalism, capitalism, socialism, nationalism, and a host of other isms--makes a test of this proposition possible.

Four and one-half centuries after conquest, complex families still exist in rural Morelos, the region studied by Carrasco. Figure 1 compares a household transcribed and translated by Cline (1993:281 H#87) with one from the 1990 census computerized sample provided by the Mexican national statistical institute (INEGI 1994).

Both examples are large three-generational households, containing several conjugal unions. The sixteenth century household reveals the ravages of mortality tempered by the flexibility of culture. This household contains four conjugal unions, two broken by death. While six unmarried children reside in the household, only three belong to the head and his wife. The greatest lateral extension is matrilineal, the widowed sister and unmarried brother of the wife of the head. A patrilineal connection is present too, although it is not an offshoot of the household head's lineage. Instead we find the head's niece's husband accompanying his unmarried brother and widowed sister-in-law and her unmarried daughter. In contrast, the "modern" family of 1990 shows no signs of mortality, but points to the continued acceptance of married children in the household. Two daughters and a son reside in free union with their spouses and with four of their own children, grandchildren of the head. A fifth couple is unrelated to the head. The household also contains the head's unmarried son aged 15 and daughter aged 10.

Readers should not be deceived by examples, even carefully chosen ones. While these cases are genuine, a quantitative analysis reveals great differences in the household contexts in which people of the two eras lived (Table 1, graph). In central Mexico of almost a half a millennium ago, kin ties were ubiquitous. Nearly half the population resided in households with at least three generations of kin. Likewise, half the population was related to the head, without taking into account members of the head's conjugal unit. Thus, very few people in these villages--less than one percent--did not live with kin. In recent times by contrast, the nuclear family reigns supreme, the residential locus for nine-tenths of the population of rural Morelos. Relatives of the head (6%) no longer constitute a major fraction of the total while individuals unrelated to the head also do not amount to much (4%). By 1990 the large complex household had become a sociological fossil, even in rural Morelos, where 450 years before it had been the norm. Indeed, the ancient Nahua family was strikingly like the "classical family of Western nostalgia," ridiculed by Laslett and his revisionists as a demographic impossibility, the invention of unsophisticated family historians (Smith 1993:325).

Table 1. Household Composition in Rural Morelos: 1540 and 1990
and the Federal Republic of Mexico, 1990

Sources: my computations from databases constructed from Cline 1993 and from INEGI 1994 one percent individual-level national sample for: acommunities with fewer than 2,500 inhabitants in Morelos state; and bnational totals for the Federal Republic.

Gender was initially of little concern to family historians, revisionists or not. Indeed, Louise Tilly's thorough review of 426 English language journal articles on the history of family and gender concludes that there was a "missed connection" between the two fields (1987). Early gender historians saw the family as "a central institution of women's oppression" and turned their energies instead to political history, particularly that of womens' movements (Tilly 1987). Family historians with a demographic approach missed the gender connection because their earliest studies found few women heading households. In Latin America, on the other hand, the widespread occurrence of female headed families forced historians to take gender into account (Kuznesof and Oppenheimer 1985; Arrom 1985, 1991; Diaz 1997). If one of the fundamental principles of patriarchal systems is that "the senior male controls and protects everyone in the household," then the challenge for historians of the family is obvious (Dore 1997:105).

Here too, Mexico serves as an example. Gender mattered for the ancient Nahua, but there is no scholarly consensus on the nature of relations between the sexes. While most writers agree that "women were subordinate to men" (Cline 1993:40), the meaning of that subordination is disputed (Burkhart, 1992; Rodriguez-Shadow and Shadow 1996). Kellogg argues that parallelism and symmetry were fundamental features of gender relations and that complementary elements outweighed hierarchical ones (1995a:564; 1997:125). Her studies are based on exhaustive readings of a large corpus of sixteenth century notary and judicial documents with native women as principals. Many of her sources were written in Nahuatl, and thanks to her labors and others we are permitted to hear the native voices of ordinary people.

Yet too often the study of family and gender in particular centers on Tenochtitlán and neighboring urban complexes while the country-side is ignored. With the vast majority of the Nahua population residing in rural areas, much of the urban evidence is simply beside the point. While published texts are much more abundant for the core of the empire, the peripheries, where most people lived, should not be ignored. It is possible that in the center there was more parallelism and complementarity with less hierarchy, although Rodriguez-Shadow (1991, 1996) argues otherwise. She reexamines the conventional published texts and concludes that Aztec women were devalued and dominated by males. Her analysis, with its discussion of schools, tianguis (markets), and temples, also focusses on the urban scene. It seems to me that the evidence for Rodriguez-Shadow's thesis is even stronger in the country-side, where the position of women was decidedly subordinate from my reading of the Morelos censuses.

Cline examined the evidence on women with respect to household headship, polygamy, age at marriage, women's position in the household, and work. She concluded that on the whole women were subordinated to men (Cline 1993:31-42). Of 315 households only three were headed by women, and in each case female tenure was clearly transitory. All female household heads were recently widowed and their coresiding children had not yet married (Cline 1993:40-41). In other cases, widowed mothers were listed as subordinate to their sons, often young and recently married. Widows seem to have been barred from remarrying because they made up almost one-fifth of the adult female population while for adult males the fraction was less than one-fiftieth. Widowed or abandoned women remained under the hegemony of a male--a son, brother, in-law, or even more distant kin--, but men readily remarried, typically with a young virgin.

Some men did not wait for a wife to die before taking another. Only males enjoyed access to more than one spouse or concubine, but the frequency of polygamy was low in the rural communities studied here, involving only four men, all local leaders, and sixteen women. For example, the tlatoani (village leader) of Huitzillan had a wife. He also had six concubines, eight children, and two dependents--both female, one a widow with two young children herself. In Quauhchichinollan the local leader was unmarried, but claimed three concubines and an aunt who served as "his noble woman" (Cline 1993:133). He had only one child (two years old). He was accompanied by his mother, three young unmarried brothers, six servants and a female (Indian) slave.

When Aztec marriage is discussed by scholars, polygamy receives favored treatment even though its frequency of occurrence is easily exaggerated (Bernard and Gruzinski 1996:164). Polygamy was almost exclusively a perogative of nobles, if we may generalize from the Morelos evidence. Yet, the possibility of a husband taking a second wife, or simply abandoning the first must have a remained a lingering uncertainty in the minds of many ordinary women.

My rendering of the demographic data on family and marriage adds somber hues to the portrayal of gender relations among the Nahua. Marriage, which in these texts meant cohabitation with the expectation of child-bearing, was extremely precocious for females, averaging less than thirteen years, substantially less than what any previous writer dared to suggest (McCaa 1996:18-31). Adulthood meant marriage for Nahuas, and almost all females completed the transition by age 15. Since most males did not attain adulthood until their twenties (male mean age at marriage is estimated at 19.4 years), this left a large age gap of almost seven years between spouses, a significant difference given the extreme youthfulness of brides. Perhaps this was tempered by the fact that many young brides remained in the paternal home, with the groom moving into the bride's father's household. Other young brides shifted from the dominion of their fathers to that of their husbands, who were often themselves under the tutelage of a brother. In either case this generally happened before the bride reached the age of biological maturity. The timing of marriage and spousal choice must have been a matter of parents and kin, or perhaps the initiative of the young man who simply "took" his woman. For females, autonomy within the household came with widowhood, if at all. Indeed, few widows maintained an independent household for long. For most, dominion passed to a son, or more frequently, son-in-law, brother, brother-in-law, or more distant male kin.

If "patriarchy" is defined as the "manifestation and institutionalization of male dominance over women and children in the family" (Lerner 1986:239), then the Nahuas evolved a peculiar species of this social demon. Patriarchal control explains the absence of female household heads, the relatively early marriage of females, late marriage of males, and the strange ordering of coresidence by marital status and gender. Since marriage was not inevitably neolocal, more than half of all married couples lived under the tutelage of others. This pattern of marriage and coresidence was not random. Fathers and brothers husbanded males in the household--their sons and other brothers. Fathers were much more successful in holding married daughters (and their husbands, n=77), than sons (and their wives, n=37). Married brothers stayed together but somewhat fewer married sisters remained under the authority of a brother (n=98:64).

Young males and females of marriageable age were not distributed randomly within households either. Fathers used unmarried daughters to draw males into the household, but brothers did not or could not. Half of all daughters of marriageable age (10 years or older) were in households headed by their fathers, but the proportion was less than 40% for sons. There were only five unmarried sisters above ten years of age living in households headed by their brothers, but there were 38 brothers in this situation (excluding the heads themselves).

Mortality, specifically of patriarchs, was the lubricant in this system, providing the flexibility for first one gendered code of marriage and residence to apply and then another. Conventional studies of household and family give relatively little consideration to the dynamics of gender within households except for headship. Where neolocality is the rule, the timing of marriage and who marries whom provide insights on family and gender, but when coresidence of married couples is common as among the Nahuas, the analysis must go much more deeply inside the household. Some of these differences which I have briefly catologued here were due to demography--for example, the fact that married sons were more likely to be orphaned than married daughters--but even these are due in part to the workings of a finely tuned system of gendered dominance and subordination applied to offspring.

Traces of dreariness, if not gender dominance, emerges even from a subject as simple as personal names (Table 2). The "earthly" names of Nahua females hint at anonimity and submissiveness, if not outright gender tyranny. The ten most common female names were shared by 86% of the female population, compared with only 27% for the top ten male names. The most common female name, Teyacapan and its variant Teyapan ("La Primera" or "Primogénita"--Spanish translations are from an anonymous text in MNAH "Libro de Tributos," ms. 549 bis), was borne by 313 females (26.8%). Tlaco ("La de Enmedio") and Teicuh ("La Segunda") each contributed an additional 15.4%. Necahual ("La Callada") completes the set of names borne by 10% or more females. More than 70 percent of all females shared one of these four names. Of the remaining six in the list of the top ten, four refer to birth order--Xoco ("La Ultima"), Xocoyotl ("Benjamina, Hija Postrera"), Tlacoehua ("Hija Segunda"), and Cihuaton ("La Pequeñita). The list is completed by Centehua ("La de Uno") and Tepin ("Hermana Mayor"). There are 78 other female names in the database developed from Cline's transcriptions. Fifty-seven appear only once. Twenty-one occur fewer than 15 times but more than once. Combined, these 78 less common names account for 14.2% of the female population.

Table 2. Frequencies of Ten Most Common Nahuatl Names By Sex
Huitzillan and Quauhchichinollan, 1540

Source: my computations from database constructed from Cline 1993

Note: Frequencies differ slightly from McCaa (1996) because of a small number of refinements in standardizing spelling and in sexing names.

Male names follow a completely different logic. There were many more of them, they were shared less often, and they referred to animals, plants, qualities or attributes rather than birth order or kin ties. Four of the most common names were shared by one-fifth of the males--Yaotl ("Rival" or "Enemigo"), Matlalihuitl ("Pluma Celeste"), Nochheutl ("Frijol Ideal") and Coatl ("Serpiente")--, but the dispersion for the remaining names was considerable. One-third of the male population had unique names, and an additional two-fifths were known by names that were shared by fewer than 15 individuals. Names occurring 12-19 times were: Tototl ("Pajaro"), Quauhtli ("Aguila"), Tochtli ("Conejo"), Çolin ("Codorníz"), Matlal ("Añil"), and Xochitl ("Flor"). Male names individualize the person whereas female names allude to birth order, kin connections, submissiveness or being possessed by a male. In addition male names were more likely to carry an honorific suffix, while very few females carried any distinction whatsoever, other than birth order. The study of Nahua "earthly" or "local" names remains something of a "linguistic thicket." My quantitative analysis suggests that such a study may reveal new insights on Nahua gender relations (Lockhart 1992:117-119; Cline 1993:92).

Work, like personal names, was also "generally divided along gender lines" among the rural Nahua (Cline 1993:91). Cline's felicitous translation of the Quauhchichinollan and Huitzillan censuses show men working the fields, looking for cotton, delivering tribute goods, and laboring to pay tribute. Women are reported as spinning yarn, weaving cloth, grinding maize, and caring for children. Rarely does women's work overlap with men's, and when this happens, such as in the case of "tribute helpers" and certain kin, it seems to evoke comment to explain or justify the transgression. Preparing tortillas requires a lot of work, and in ancient Mexico only women performed this exhausting task. When additional women were incorporated into a household, such as widows or the abandoned, their roles in the household were spelled out and recompense noted (Cline 1993:91), as in household Q120 (p. 209): "Here is Acxotecatl's [widowed] mother-in-law, named Necahual. . . She just spins, and for that she is fed by him." The Morelos censuses point to a strict division of labor by gender, along the lines, it seems to me, developed by Rodriguez Shadow in her book, La mujer azteca (1991; Rodriguez-Shadow and Shadow, 1996).

Kellogg warns against applying modern notions of equality, subordination, and the like to Aztec society. Instead she proposes that Nahua gender relations formed a compound of parallel, symmetrical, complementary, and hierarchal elements (1997:132). The fact that kinship was neither patrilineal nor matrilineal, but cognatic alerts us that the structuring of gender took usual forms among the Nahua. From her study of female priests, merchants, property-holders, and the like gender roles may seem more complementary than hierarchal, but in the country-side the professions were few and they were all occuppied by men. Hierarchy was the rule. Age, gender, wealth, household headship, social differences, and political position served as markers for power, rank and status, both within the household and beyond. Among rural Nahua patriarchalism kept young men and women of all ages in their place, subordinate to older men. With the Christian invasion, a new, colonial form of patriarchalism slowly emerged, a blend of Hispanic, indigenous, and new autoctonous forms of family and gender relations.

By the beginning of the 17th century conjugal families were becoming the norm among the Nahau and, for newly married couples, neolocal residence the standard. The key to these changes was the spread of Christian marriage with its insistence on, first, voluntary consent of both males and females in deciding when and whom to marry and, second, a minimum marriage age of twelve for females and fourteen for males. As marriage age was forced up for females and downward for males (see Borah and Cook 1966 for a contrasting interpretation), the ancient foundations of Nahua households were unsettled. Gradually, parents, kin and community lost much of their influence over marital decisions of the young couple. Young men were liberated from a prolonged period of sexual abstinence and young girls gained a childhood, spending a few more years out of the rebozo before settling into a nuptial bed and setting up a household (McCaa 1996a:18-31; 1996b:43-44, 48-51). As the conjugal family became the basis for land allotments and tribute payments, the economic raison d'etre of complex families disappeared.

The pace of change varied from city to country-side and region to region, but by the end of the colonial era the transformation was complete almost everywhere (Kellogg 1995b:187-200). The extended family disappeared among the native peoples before Spanish hegemony ended, as shown by a study of the Indian village of Ozumba (today, in Morelos state), where in 1793 only 4.6% of the households had a complex form, compared with 16.8% for households headed by Spaniards or Mestizos (Vera 1993:50). In the village of Chilapa (Oaxaca) in 1777, the fraction stood at only 2.6%, and the mean age at marriage reached 18.1 years for females and 21.5 for males (Chena 1992:172-176--my calculations).

The final stage of the colonial transformation can be seen in the regional city of Antequera (Oaxaca) where we have an impeccable demographic study of living arrangements of 18,061 individuals by sex, "calidad" (race--see McCaa 1993), and age for the year 1777 (Rabell 1996). For the 4,350 Indians enumerated in Antequera in 1777, the nuclear family, not the extended, was the principal place of refuge. Instead of complex households with two or more conjugal families forming a great mass of kin as in rural Morelos in the 1540s, only 2% of the Indian population of Antequera lived in households where they were related to the head, but not a member of the head's conjugal family. In contrast, outsiders with no kin tie to the head constituted almost as large a fraction of the population as children of the head (27% versus 28%). Many of the outsiders were servants or criados, working in Spanish households (Rabell 1996:103). These living arrangements reflected economic conditions in the city, the availability of housing, a racialist ideology, and, I suspect, a weakening of kin interactions between households--although the census data are silent on this point. Indian living arrangements were not a matter of their imitating the city's dominant social group, the españoles (Spanish-speaking creoles with a sprinkling of peninsular born males). In proportional terms, more than four times as many Spaniards found a place with kin outside the conjugal family as did Indians (9% versus 2%).

Table 3. Ciudad de Antequera (Oaxaca)
Household Composition by Calidad and Sex, 1777

Source: data kindly supplied by Cecilia Rabell; see Rabell 1996.

Age, or what family historians refer to as the "life course," greatly influences living arrangements, and it must be taken into account if the trajectories of family and gender relations are to be understood (Hareven 1991:105; Anderson 1995:17-18). The life course approach has become de rigeur for family historians working on Europe or North America, but unfortunately it is little used in Latin America. This neglect is not due to a lack of data. Typically, census lists in Latin America are much more detailed than those in Europe. Instead, it seems to me that the problem lies with the influence of the Laslett school, which continues to propagate two grave methodological errors among Latin American historians of the family (Laslett 1993; for American contrasts, see Anderson 1985, 1988; Doenges 1991; Ruggles 1994:117). First, Laslett and his followers focus on the household as the unit of analysis rather than the individual, and thus lose sight of gender, the socially constructed differences between male and female. Second, they commit the error--or rather, heresy, in the minds of the demographically inclined--of ignoring age. The English have reveled in their religious and political heresies for centuries, and at least in the case of family history, the blasphemy is perhaps justifiable because their impoverished census listings rarely report age. The English censuses are often simply lists of names marked off by lines, with no other data at all! In Latin America, where age is commonly recorded in census lists (along with calidad or race, relationship to head, marital status, and less commonly occupation, communion status, and even height), the only explanation for the continued propagation of the Laslett heresy is cult-like imitation.

Table 4. Life Course of the Population of Ciudad de Antequera
by Calidad and Sex, 1777 (percent)

Source: data kindly supplied by Cecilia Rabell; see Rabell 1996.

A recent study by Cecilia Rabell is a model of what can be learned from life course analysis (1996:75-118). In a set of splendidly crafted figures, with kin ties grouped into seven categories--head, spouse, offspring, other kin, criado, agregado, and orphan--Rabell shows how in 1777 family residence varied over the life course (what she calls, "trayectoría de vida") by gender and for each calidad in the important regional center Ciudad de Antequera (Table 4, graph). She argues that gender historians often exaggerate the frequency and significance of female headed households (see also Grajales Porras 1991 and Kanter 1995 for rural, primarily Indian settlements). Female headed families were smaller than those headed by males, even after taking into account the fact that the latter always had a woman in them whereas the former rarely reported a male companion. While a substantial fraction of older adult women did become heads of households in Antequera, the fraction was relatively slight and chances were much greater for españolas, particularly widows, than for others (see also McCaa 1991). From age 50, half of all Spanish women, three-fourths of whom were widows, headed their own households, compared to one-fourth or so for women of other calidades. For most women, that is Indians and Castas, from age 30, about one-fourth were in service, rising to roughly one-half by the age of 55 or so. For Spanish women to the age of 50, kin continued to offer refuge, but very few other women found shelter with extended kin after age 25. Rabell's life course analysis uncovers what European family historians call "life cycle servants," although in Antequera this phenomenon was limited to Indians and Castas in their teens (1996:103). From the tutelage of parents, many non-Spanish males and females passed to the control of masters, then later married and formed their own households. Finally, upon the death of the spouse, females were likely to return to service, whereas males quickly remarried.

I have summarized Rabell's main findings to illustrate the insights that may be gained from a life course perspective. Her text takes into account the four main socio-racial groups and highlights the differences and similarities among them. Mestizos and Castas constituted sizeable fractions of Antequera's residents, 18 and 20 percent, respectively. Each had distinct living arrangements, and they also differed by sex. It is noticeable for example that the life course of female Mestizas was more like indias, whereas the pattern for male Mestizos was more like españoles. Rabell concludes by arguing that when the entire population is taken into account, including both sexes --not just males, as is often the case among historians working on the social structure of colonial Mexico--, "calidad" (which incorporates both socioracial and cultural elements) explains much of the variation in family living arrangements, although demographic, economic and cultural factors are also important (1996:115-7).

To gain perspective on changes in life course over the centuries, I compared Rabell's graphs for 1777 with similar figures for the city of Oaxaca in 1990 which I computed from individual-level national census samples. Individual-level census data for Oaxaca City survive for several years in the nineteenth century and, in this century, for 1930, but these fascinating sources must await computerization before they can be worked. Doing so would allow one to study family and gender changes and their correlates over shorter intervals of time.

Comparing Antequera of 1777 with Oaxaca City in 1990 reveals patterns that are strikingly different in rather surprising ways. First, for adult women, up to the age of 55, headship is much less common in 1990 than in 1777, but the fraction of females who are spouses of the household head more than offsets the difference. In 1990, the conjugal family is a reality for a much greater proportion of adult women and for a much longer period of their lives than in the distant past. Second, in 1990 the extended family offers much greater support than in 1777--without, however, approaching the 50% level characteristic of the ancient Nahua. The increased proportion of women living as extended kin is not a sign that the nuclear family is failing but rather that in recent times fewer Oaxaqueñas live in households as non-kin. Family now provides more succor for the very young and the elderly as extended kin than two centuries before, meaning that relatively fewer women live as non-kin dependents. Today migration is even greater than in the past, but now it occurs within a family context, either nuclear or extended, rather than the coming together of unrelated individuals and of different social groups, such as masters and servants, as in the past. The life-cycle servant pheonomenon has virtually disappeared, and what remains is pushed to the late teens and early twenties. From age 25 to 64, fewer than five percent of females live in the households of non-kin. The nuclear family also offers more sustenance for its offspring than in the past--just as the conjugal family does for adults. Childhood occurs almost entirely in the parental family and adolescence is prolonged, for many, into the 20s. Finally, elderly mothers are much more likely to be taken in by their own married children than two centuries ago.

Male life course changes in Ciudad de Antequera/Oaxaca parallel those for females, although for adult males heading of the household is the rule over a much longer period of the male lifespan, notwithstanding their later age at marriage. Even at age 65 and beyond more than four-fifths of Oaxaqueño males continue to head their own households. Today, few males experience the life-cycle servant phase. The nuclear family shelters males through adolescence and into the twenties. Although males are marrying much later than in the past, they are also more likely to continue to reside in the parental household until marriage. Then too, the extended family is much more likely to succor young males who leave the parental household without starting a family of their own. Beyond age 20 less than five percent of the male population stay in households where they are not related to the head.

Table 5. Life Course of the Population of Mexico by Sex, 1990
(percent)

Sources: my computations from database constructed from INEGI 1994 (one percent sample; total number of cases: 801981). Percentages do not sum to 100.0 due to truncation.

Family residence in Oaxaca City in 1990 mirrors patterns for the entire republic to a surprising extent. Since the national figures are of broader interest than those for Oaxaca city alone, I offer a brief discussion of the national life course figures in Table 5. In 1990, living arrangements are less strongly gendered for the national population as a whole than in Oaxaca City, particularly if one views headship and spouse of head as equivalent categories. Female headship rates for adults aged 20 to 44 are only one-fifth those of males. Thus, female headed households are even less common among the general population than in Oaxaca City, rising from 2.5% in the 20s to 25% aged 60 and over. On the other hand, childhood residence for males and females is quite similar, with the nuclear family occupying an ever stronger position at the national level. Nine of every ten children aged 0-14, whether male or female, live in the conjugal family of birth. (Note as well that gender differentials in educational opportunities to age fifteen have been reduced to near insignificance in 1990.) Of the young who are not related to the head (circa 10%), two-thirds to three-fourths dwell with other kin. In adulthood, nine of ten continue to live in a conjugal family either as head, or spouse or children of the head.

Life cycle servants are also relatively less common at the national level. The two highest rates of non-kin residence for women are 8.6% for the age group 20-24 and 16.2% at age 65 and over, but in both instances the percentage of women living as extended kin is greater than that for non-kin, 9.4 and 27.0%, respectively. For males the equivalent figures for non-kin are 6.7 and 9.9%, compared with 8.4 and 9.9% for extended kin. Extended kin relationships are less common at the national level than in Oaxaca City because overall migration rates are higher to the city, and, as we have seen, in the waning decades of the twentieth century migration tends to be kin based, unlike urban migrations in the last decades of the eighteenth century.

The broad brush strokes in this essay suggests that over the course of almost one-half millennium, family and gender relations changed dramatically. The complex joint family system of the rural Nahua contrasts starkly with the late colonial urban type which was based on a conjugal family unit enlarged with non-kin dependents. As the millennium ends, Mexico is undergoing another major family transition where residence patterns are greatly simplified from former times. The conjugal family is rapidly becoming the universal type, however at the same time it is also an important refuge for kin.

If we shift perspective from the household to the individual, patterns of gender and the life course are more readily engaged. Ancient Nahua families were highly stratified by gender. Males monopolized headship, but in doing so they were prevented from marrying until much later in life than females. Females were at the mercy of male relatives what with marriage before puberty, a large age gap between spouses, and a strong rule against widow remarriage. On the other hand, very few females depended upon non-kin. Family was the haven from the certainties of a capricious demographic system.

In late colonial times, residency was strongly influenced by calidad. The privileged life course of españoles of both sexes is fairly well known. Males became heads in their twenties and retained their position until death. Females became heads upon their husbands' death, because remarriage prospects for widows were almost nil. Yet their social position and resources were such that they could maintain a household without the presence of a male head. Rabell's analysis shows the strong dependency on non-kin that Indian and Casta females experienced, as children, and adolescents, and finally as widows. Male dependency for the same calidades was concentrated primarily in the teen years because, when widowed, males readily remarried.

In contrast, in the 1990s, gender seems to be relatively less important in structuring families than in earlier times and places studied here. Headship remains the male perogative. Males are slightly less likely than females to reside either as extended-kin or as non-kin, but these differences rarely amount to more than a percentage point at most ages. In old age women are much more likely to depend upon kin, while males continue to head households to advanced ages with seventy-five percent of those 65 and over in this position. Differences by social groups (data not presented here) are much less significant than in earlier times.

Peter Laslett enticed us to consider the larger canvas of family history. The Mexican picture is much more complex than the uniformitarian hypothesis of the nuclear family allows; indeed the thesis has been discarded in much of Europe and North America (Ruggles 1994; Anderson 1995). Here I have identified three prototypes in the history of Mexican families and gender. It is likely that with studies of other periods and places, the repertory of types will grow, along with our understanding of how different the past is from the present (for the United States, see Ruggles 1994 and Smith 1995).