The Circulation of Children Through Households in Yap and Kosrae

We describe a gendered pattern of circulation of youth through households in two Micronesian societies. Yapese girls andfemale teens shift into vertically extended households, where elders reside, between ages 11 and 17. The opposite pattern pertains in Kosrae, wherefemales shift between ages 11 and 14 into two-generation households. The opposing patterns are explained by opposite divisions of labor, with Yap having afe-malefarming system and Kosrae having a malefarming system. In both instances the residence shifts of female youth between middle childhood and the late teens involve movements into the kinds of households where they do the most work, per capita. We explain the patternfirst in terms of the desire to teachyoung people socially valued work roles, and, secondly, in terms of the need for their labor. These analyses demonstrate the need for processual modes of understanding household composition and its effects on the formation of gender roles and transmission of knowledge. _C ~ross-cultural studies of socialization have emphasized the effect on children of cultural practices and social relationships within households. Household patterns vary within societies and individual households change over time. Hence, studies of socialization must consider synchronic and diachronic variability in household form. In our research on household processes in four Micronesian societies in the early 1990s we collected four waves of census data for each of nearly 200 households. Through the process of coding

and analyzing these census data we encountered a high rate of move individuals and kinship groups among households. In this article on one consequence of these movements for the children of tw ties-because the circulation of children varies by gender, boys tend to live in different kinds of households over time. We analyze group and gender patterns in relation to food production and p tasks.
The Whiting model for the study of socialization emphasizes fects of the material system upon the child's learning environme ing 1973). The term ecocultural niche is now used to descri culturally based context of development (Super and Harkness 19 ner 1984). Important components of that niche include subsisten the division of labor by age and sex, children's work, and t younger children by older children (Weisner 1984, Whiting and1988). Many of these processes are affected by the structure of hold.
The Whitings' seminal work, Children of Six Cultures (1975 lighted the contrast between extended family households and nu ily households. In many societies grandparents and other elder important role in child care and the transmission of cultural v practices.1 Elders also may have lower levels of formal education commitment to tradition, greater involvement in subsistence p and less experience with wage labor. In many societies, moreov women have greater equality with men than do younger wome 1985). In the matrilineal societies of Micronesia senior women considerable economic power, often exercised through their rol ing inheritance. Hence, compared to residence in nuclear ho residence in a three-generation household may present children ferent cultural content, different social practices, and a differe of gender roles and of the life course of women. Traditional ped Micronesian societies stresses learning through observation and tion (Metzgar 1991;Rubinstein 1979), and the shifting of childr ferent age groups and sexes through a range of linked househol be understood as the deployment of youth not only to help with wo but also as a way to maximize their chances of learning vital s knowledge as well as appropriate social behavior, and to streng individual's vital kin linkages.
Micronesian households often vary considerably within a com with some children living in small nuclear family households, a living in complex multigeneration households (Burton et al. 200 the concept of ecocultural niche must allow for variability acr holds. This concept must also allow for variability over time in course of the individual. In many societies children live in several di households while growing up. Weisner (1984:346) lists a variety of reasons for these residence shifts including care of elders, repayment of debts, apprenticeship, formation of political alliances, and educational advantage. According to Weisner these residence shifts often begin in middle childhood (ages 6-12) and occur frequently in East and West Africa, and in Polynesia. In societies where these changes of residence occur, an understanding of the ecocultural niche of a child's development must incorporate that child's individual experience with visiting, fostering, and adoption-circumstances that may cause children to experience a variety of households in different sequences.
In our research in two Micronesian societies, Yap and Kosrae, we find that individual experiences with households vary by gender and by age group. We discuss this finding using ethnographic and survey data. In the ensuing sections of this article we describe how children circulate through linked households. We examine how the varying household experience of young males and females appears to be linked to children's work responsibilities.

LINKED HOUSEHOLDS AND THE CIRCULATION OF CHILDREN
People move frequently among Pacific Island households. Individuals normally have land rights in several locations and they may move among locations for a variety of reasons. Children sometimes move among households as part of their larger family unit. They also move as individuals.
Adoption adds social relationships rather than canceling them as is often the case in the United States. Adopted children typically retain their social ties to natal parents. Thus it is common for them to move between the households of their natal and adopted parents. Children may also be fostered for long periods of time with a variety of close or distant relatives, being sent to help care for grandparents or young children, or to stay with grandparents or other relatives on the home islands while their parents work in an urban center in the islands or in the United States. There are two consequences for children of the frequent movement of individuals and groups among households. First, the child's household experience may change over time because the child moves among households, either as an individual or as a member of a larger social unit. And, second, even if the child does not move, the household in which the child lives is likely to change its composition frequently, as some family members depart and others arrive.
Circulation of children has received the most attention from anthropologists who work in Africa (Bledsoe and Isiugo-Abanihe 1989;Castle 1995;Goody 1970Goody , 1982. Page (1989) reported data on the percentage of children not residing with their mothers, by the sex of the number of regions of Africa, and found no systematic sex differ rican societies share with Pacific Island societies a pattern of l households within larger extended family structures. We think pattern of linked households encourages the movement of indiv cluding children, among Micronesian households. The impli these patterns for the cross-cultural study of human developm clear. A set of snapshots of a child's activities taken at one poin will lead to underestimates of the degree of variation of childre hold experience over time. Static comparisons of nuclear famil holds with extended family households may be misleading children experience both kinds of households sequentially. Diffe dren experience different households at different points in tim we show below for Kosrae and Yap, the aggregate household co socialization varies significantly by gender. As a result, one cann the formation of gender roles and identities within the shared f of a single type of household, even if it is the statistically pre form.

KOSRAEAN AND YAPESE HISTORIES
Yap is in Western Micronesia between Guam and Palau. Kosrae is in Eastern Micronesia between Pohnpei and the Marshall Islands. They share an Austronesian linguistic heritage. They also shared a long colonial history under Spanish, German, Japanese, and American administrations and now constitute two of the four states of the Federated States of Micronesia (FSM).2 Hence, they are good candidates for controlled comparisons. Kosrae State is populated by a single ethnic group the Kosraeans. Yap State is populated by the Yapese, who live in the main island group, called "Yap Proper" and by the Carolinean Islanders, who live in a chain of coral atolls east of the Yap islands. The focus of this article is on "Yap Proper." In 1994 "Yap Proper" had a population of 6,919 and Kosrae had a population of 7,317 (Federated States of Micronesia 1996).
Yap and Kosrae are primarily rural societies that have maintained their subsistence systems based on farming and fishing but also now have high levels of education and wage labor. They have different gender divisions of labor. Whereas Yap has a female farming system based on cultivation of taro and other root crops, Kosrae has a male farming system that places more emphasis on a tree crop, breadfruit. Men do the fishing in Yap, but women make a substantial contribution to fishing on Kosrae. This gender contrast in the adult division of labor carries over to the work allocated to children and adolescents, and thus allows us to examine the effects of children's work upon the movement of children through households.
The populations and social systems of Yap and Kosrae were greatly affected by the colonial experience. During the mid-to late-19th century they both suffered depopulation and rapid social change. Severe population losses in Kosrae due to European diseases reduced the population to fewer than 300 individuals (Gorenflo 1993;Ritter 1981). As a result of this depopulation and the stresses that new illnesses brought, the Kosraeans abandoned their chiefly system of governance and traditional religion, and converted to the Congregational Church, which continues today to play a strong role in Kosraean social life (Peoples 1985). The Yapese retained more of their traditional social system. Councils of hereditary chiefs today have considerable influence in Yap State politics, including specific duties outlined in the Yap State Constitution.
Following World War I, Yap and Kosrae were under Japanese rule. Following the American victory in the Pacific during World War II, these Micronesian islands were administered first by the U.S. Navy and then by the United Nations Trust Territory of the Pacific, a trust that was held by the United States and administered by the Department of the Interior. Populations began to recover in Yap, Kosrae, and the rest of Micronesia, and by the time of our study both societies were experiencing population growth. Kosrae had the higher fertility, a TFR of 3.93 in 1994 versus 3.26 in Yap; and a correspondingly lower median age of 18.8, versus 20.9 in Yap (Federated States of Micronesia 1996).3 The Federated States of Micronesia was formed in 1979. Full independence came in 1986 with the signing of a Compact of Free Associat with the United States. Under the terms of that agreement the Unit States will continue its transfer payments to the FSM until 2001, an possible renewal of the Compact is now being negotiated.
The political and economic relationships of the past half century h had significant effects on families and households. During the 1960s United States greatly increased its spending on education and governm services. There were great increases in the levels of Micronesian education; in the number of wage jobs, mostly in the government sector, and in consumption of imported food. These processes continue today. The economic activities of young parents may be quite different than those of the grandparent generation. Where wage labor was once rare and most were in the subsistence economy, we find today that most households include wage laborers, and many younger people are less involved in farming and fishing. Many Micronesians are concerned about the potential for loss of cultural knowledge and of the food exchange systems that are at the core of social life in much of the Pacific. Contact of children with grandparents is seen as very important to preserving tradition, a theme that emerged in our focus group discussions when we formulated our project.
We collected data in Kosrae and Yap as part of a larger com study of household processes in four Micronesian societies. Lead each society were closely involved in the research design. Th graphic research was done by James Egan in Yap and by Karen Kosrae. Data for the present study were collected as part of four household surveys that were conducted during 1992 and 1993.4

METHODOLOGY FOR STUDYING HOUSEHOLDS
We define households as spatially contiguous units whose me share cooking and food preparation. Micronesian households oft prise several houses but utilize a single cookhouse. Often a nu related households are located in close proximity. Elders may occu own households within these linked household networks. In those cases most of their foods are prepared and/or delivered by their children grandchildren. Within the larger households, there may be smaller nucl family units that act independently in many ways; nonetheless, many family resources and decisions are shared across the various units comprisi the larger multigenerational extended family. Certain decisions affect a number of households may be made at the level of an apical househ including elders. For this reason our research design included examples what we have called "linked households." We ensured that our sample included households at several levels of the linkages, from households t were newly formed to mature households of elders.
We collected census data at four intervals within a year. In this arti we classify households into three basic types, based on the kinds of ext sions that occur within the household. Nuclear households usually cons of a married couple and their children. For the purpose of this study also included single-parent households in the nuclear category, as wel households that included a married couple, their children, and one more other relatives, such as a nephew or niece, provided that the hou hold did not fall into one of the following extended family categories.
We distinguished two kinds of extended family households. Vertica extended households consist primarily of an older person (or married c ple), one or more of their children, some or all of the spouses of the ch dren, and grandchildren or great-grandchildren. They also may inclu other relatives, such as a sibling, nephew, or niece of one of the elders. included in this category one household that consists of a grandparent grandchild, with the intermediate generation absent. Laterally exten households are extended through siblings, and usually consist of two more siblings, their spouses, and their children or other descendants These households could be formed by the death of the senior member a vertically extended household, or by the independent decision of two more siblings to form a household. We include in this category households that include persons linked through a sibling of the head of household in the case where that sibling is no longer present. While laterally extended households could span three or more generations,5 in Kosrae and Yap they tend to span only two generations.
After extended analyses it is important to note that the single most important distinguishing characteristic among these broad household types is the presence or absence of elders, who are often the grandparents of the children in our sample. In both Yap and Kosrae nuclear households are more involved with wage labor and vertical households are more involved in farming and fishing, however both economic forms are widespread within both kinds of households, for monetized and subsistence economic practices are highly interconnected within Micronesian societies.
We turn now to the data from Yap.

YAP
Yap is a small complex of high islands within a single fringing reef that enjoys high but seasonal rainfall. The Yapese economy employs a mix of wage labor and local food production. Men and boys fish within the reef or in the open sea while women and girls do most of the agricultural work. Taro, the staple cultigen, is grown in irrigated gardens. Yams, sweet potatoes, bananas, cucumbers, and other crops are grown in swidden gardens. Tree crops such as breadfruit, Tahitian chestnut, papaya, mangoes, and other fruits supplement garden produce. Yapese household resources usually include pigs and chickens, and both male and female children participate heavily in their care.
The fundamental sociopolitical unit in Yap is the estate-the tabinaw-which consists of a group of land parcels and the people who hold rights to them. An individual tabinaw may consist of a single household in which reside a man, his wife, and their children. More frequently, members of several linked households share tabinaw lands. The land parcels of the tabinaw, though widely dispersed throughout a village, are all symbolically linked to the stone platform (def) upon which the primary house rests. The def has both a personal name and a social rank with respect to other def both within the village and throughout the rest of Yap. The extension of tabinaw lands to several households usually begins with the marriage of sons. Residence is virilocal. When they marry, adult sons typically build homes upon separate parcels of the estate land, where each resides with his own wife and children. The sons and their families continue to identify with the def of their father and act as a political unit within the village. All have access to special resources that belong to the tabinaw, such as garden land, groves of coconut and betel trees, fishing rights. Hence, these households of the sons and father m tively comprise a single tabinaw. At the same time, each of th independent households maintains other resources of its own, pigs, chickens, and rights to garden produce. Each keeps a sepa count of income and expenditures, its own kitchen or cookhouse it prepares food, and its own refrigerators, automobiles, and oth Because so many decisions regarding resources and daily househo tenance are made at the level of these individual homesteads, w these as households for the purposes of the present analysis.
The nuclear household is the most frequently occurring ho form in Yap. Twenty-two of the 35 households in our sample w fied as nuclear. Next in order of frequency, vertically extended h constitute ten of our 35 households. These are created by one of cesses. First, adult sons and daughters with children may reside w parents. A young married son may reside with his parents wh amassing sufficient resources to build a house of his own. Or, a daughter with young children may relocate to the household of ents. While young children may accompany their mother after the d older children are more likely to remain with their father since rights will be in their father's estate.
A second process that creates a vertically extended househo when elderly parents leave their own homes to live with middleor daughters, who thus demonstrate their worthiness through he for aged parents, and thereby earn tabinaw leadership roles.
Laterally extended households are least frequent-three o households in our sample. These households are extended throug of the household, most commonly to incorporate children of a brother. When a man dies, his oldest surviving brother assumes t father to his children, and it is not unusual for a child or youn reside at some point with this new father. Less frequently the the household head or his wife may join the household, along w children.

CHILDHOOD IN YAP
Regardless of which type of household children find themselves in, each child will need to make connections to the land. Yapese children belong to their mother's matrilineal clan, not the clan of their father, on whose land they reside. Over time, a woman's labors on behalf of her husband's estate will earn both her and her children full membership in his tabinaw (Egan 1998;Labby 1976 the tabinaw def. From this point on, a male child's position within the Yapese landed hierarchy is primarily a matter of the social rank and networks of political relations associated with this def, although his position within the wage economy can provide additional political influence and prestige. Girls also maintain a lifelong association with the def from which they received their names. Although they will one day marry and join the tabinaw of their husbands, they retain membership rights in their natal estate. In fact, authority within the tabinaw is divided between the eldest son and eldest daughter, and each is charged with specific responsibilities in tabinaw affairs. While the son will see to the external affairs of the estate, the daughter protects the rights of her brother and oversees all internal estate matters. It is she who gives names to children born to the tabinaw; it is she who holds authority over her brothers' children; and it is she who will one day play a central role in allocating the estate taro patches among brothers' sons' wives. The complementary division of authority in the tabinaw and the gendered division of labor in production structure the contexts in which Yapese children are raised. The mother and older female siblings keep a watchful eye on young children. Close relatives often live next door, and young children roam the area playing in safety with young siblings and peers. Brothers and sisters may join each other in mixed-sexed playgroups when young, but increasingly avoid interaction as they approach puberty. Brother-sister avoidance after puberty shows respect for the central role of the cross-sibling dyad in structuring tabinaw affairs. These same relations manifest themselves in sleeping arrangements. Should the central building of the homestead contain only one room, parents and children sleep there together. Increasing numbers of homes today include a separate internal bedroom for the husband and wife, although nursing mothers may sleep with infants apart from their husbands. Adolescent boys, however, usually sleep in an entirely separate building of the homestead compound and thus maintain distance from their sisters.
Just as separation by gender comes to structure play and sleeping arrangements as children get older, so too does it organize daily chores in the household. Children of both sexes do few chores before the age of ten, but thereafter their workloads increase. Both boys and girls are frequently charged with the daily feeding of pigs and chickens. Both boys and girls also collect land crabs, help others to gather ripened coconuts from tabinaw coconut groves, and sometimes lend a hand in clearing land for new yam gardens. Boys alone are likely to acquire fishing skills, although we did find rare instances of older girls employing nets near the shoreline. Boys also join older brothers and young male relatives on excursions that involve hunting fruit bats and "wild" chickens with air guns. Boys are more likely than girls to harvest betel, breadfruit, and other tree crops. Girls, on the other hand, are more likely to be assigned care of younger sib be expected to help with food preparation and washing dishes, and with the routine care and harvesting of garden produce and taro.
Children are likely to spend significant periods of time in the of other relatives, notably those of maternal grandparents. Moth wage jobs often leave children in the care of grandparents or other re during working hours. Although children usually live in the same as paternal grandparents (whom they see frequently), this is no the case. Many Yapese reside in the vicinity of Colonia-Yap's por and sole urban area. Children whose parents live in Colonia may weeks or even months at a time with grandparents and other relatives in the village.
Yapese frequently send their children to live with other relatives for extended periods of time in order to broaden each child's network of close personal ties and to maintain and strengthen the parents' ties to relatives and to landed units of which the parents are members. Sending children to other households also gives the children an opportunity to observe and participate in activities that may not be occurring in their own households or home villages. This is especially so with respect to local food production. Young mothers rarely have opportunities to tend gardens and taro patches. On the other hand, middle-aged and older women who are no longer burdened with the care of young children are usually very productive, and they often take older girl visitors with them to care for agricultural resources. Similarly, young boys who live in households in Colonia may not have opportunities to fish. Spending significant periods of time with relatives in other villages allows them to learn and develop these skills. Children are also expected to participate in village dances. Yapese take great pride in traditional dance, and dances are a regular part of elementary school graduation ceremonies, competitive inter-village exchange events, the annual Yap Day celebrations, and other village and island complexwide festivities. Children who spend a summer or any extended period of time with grandparents or other relatives in a distant village will join the other children of the village for dance practice and performances if a dance is scheduled over the period of their stay.
Each of Yap's ten municipalities includes a public elementary school that is within walking distance of the municipality's villages.

LINKAGES BETWEEN YAPESE HOUSEHOLDS
An example will help to elucidate some of the processes that create the household types in Yap, especially with regard to movements of people between households. Figure 1 shows two linked households. Household 1 is a vertically extended household consisting primarily of an older couple, their divorced daughter, and her four children. This household is located in a village belonging to a lower social stratum of the Yapese sociopolitical hierarchy. Although the household is without electricity, it has abundant agricultural resources as well as easy access to the marine resources of the lagoon. None of the individuals in this household have wage employment.
By returning to live with her parents, the daughter of Household 1 has exercised an option widely used by divorced women in Yap. More unusual is the fact that she has brought her children with her. Children receive personal names that link them to the land of their father's estate. The linkage so provided may even specify portions of a son's inheritance. Still, a divorced woman may wish to have her children with her. She knows that should her husband remarry and have more children, her children will have a weaker voice in their father's estate than those of his new wife. More importantly, if her husband is himself the member of a low-ranking estate or is without land of his own, she may believe that her children's prospects Barring any protestation from her former husband's s hold authority over these children, she then may bring them the estate of her parents. In the case at hand, the divorced d former husband is of a low-ranking village.
The second household consists of a surviving son of the old his wife, and their seven children. One of the sons of the olde deceased, and he was divorced before his death. Following Yapese options for children's residence after a divorce, the dec three children had lived with him after the divorce, and they between these two households of their father's family. One d living in the vertically extended household, giving it a prepon girls (three to two). One son and one daughter are living in t household. As the eldest brother of the deceased, the head of hold is considered father to this brother's children. The husband and wife both hold steady wage jobs in Yap's public sector. The eldest daughter holds a part-time wage job and has many childcare and basic household maintenance responsibilities. In addition to wage employment, this household has its own agricultural resources and its members fish frequently. This second household, with its nuclear structure and greater involvement in the wage economy, has a preponderance of boys (six to three).
These two linked households are separated by little more than a ten minute walk. Since the second household is derived from the parental household (Household 1), both are constituent parts of a single tabinaw. Both act with "one voice" (lung) in any of the various village, municipality and broader Yap-wide political functions, and the members of the two households share collective resources. At the same time, each household maintains other resources of its own, prepares and consumes food independently, has its own sources of income, and makes its own decisions about resources and household maintenance.
We now examine the substantial cumulative effects of these kinds of decisions upon Yapese households and the socialization experiences of children who move through the various household forms.

YAPESE DEMOGRAPHIC DATA
We began by tabulating the residence patterns of Yapese children and adolescents from birth to age 17. Table 1 shows the total number of boys and girls who appeared in four waves of household census data, by household type. Here we have categorized Yapese households into two categories. The first category-vertically extended-includes three generations. The second category includes two kinds of two-generation householdsnuclear and laterally extended. We refer to these two types together as two-generation househo twice as many young fem households, while 30 perc in the two-generation ho young Yapese females res less than one-third (30.6 tended households. The pa This pattern of gender d by age (Figure 2). Gender to ten with a shift of boy generation households. This tion households into verti from 11 to 14. There is l before age six or after ag the movement of childre shifts in Yap begin in mi cence, and end by late ad

RESIDENCE
As Weisner and others have noted, there are many reasons f circulation of children among households. Several of these reason differ by gender, including apprenticeship, formal schooling, eld and subsistence labor. Here we focus on just one factor, the role o done by children and adolescents in food production and food prep We hypothesize that the circulation of children and adolescents is partly by the desire that children learn certain tasks as well as by the for household labor. Cross-culturally, the adult division of labor children's work. Girls tend to do more household labor than boy their work tends to be supportive of adult women's work (Bradley Given that nuclear and vertical households differ with respect t economic activity, we hypothesize strong relationships between th done by girls and female adolescents and their residence patterns. ticipate an intensified use of female labor in the vertically extende holds, where female gardening work is more prevalent. We tested these hypotheses using measures of total work con tions of children and adolescents from our food survey. We survey of 35 Yapese households during one week during four seasons of th collecting data on daily food production, consumption, exchange food-related work by both regular residents and guests. Most of th holds (27 out of 35) had young people between age six and 17. Th waves of data collection provided information on 468 separate in of food-related work by Yapese between ages six and 17. These t curred in 24 of the 35 households.  loads. Hence, the variance to be exp in which children's work is import where it is not especially important Within the age range six to 17, gi cent). Farming is the most important is a mainly female task. This is con bor-Yap is a strongly female farmin more important, especially in farmi assignment of girls to vertically ext valuable Yapese gardening knowledg If the labor of girls and female a circulation of children through Yap to find a relationship between house girls, with girls doing more work in v females account for less than half (4 butions by young people to two-gen cent) of the work contributions by households. Figures 3 and 4 show h Here we see that there is a great inc vertically extended households durin lates with the high concentration o extended households discussed ab workload of girls in these household numbers. To test whether girls work harder wh tended households, we computed worklo males in the three kinds of households. Here the unit is the week. Between ages six and 17 female work per capita averages 3.06 hours per week in vertically extended households, 1.68 hours per week in laterally extended households, and 1.53 hours per week in nuclear households. Male work follows a similar pattern, averaging 2.88 hours per week in vertically extended households, 1.73 hours per week in laterally extended households, and 1.07 hours per week in nuclear households. Both males and females work harder when they reside in vertically extended households. Across the three types of households girls have a higher average workload per capita-2.39 hours per week versus 1.67 hours per week for boys. Neither boys nor girls do much work on average. However, there is a high variance in work load across the sample, with some individual children having a much higher work load.
We used regression analysis to further refine these relationships. In these analyses the dependent variables are total amounts of work by male and female youth, per household. The independent variables are measures of the numbers of youth in each household, by age group (Table 3). The beta weights for these age groups estimate the net addition to total household work per additional young person. We see in Table 3 that there are significant effects for numbers of males in the age groups from six to 14, and for numbers of females in the age group 11 to 14. For each gender the strongest effects are in the 11 to 14 age group. After controlling for numbers, we find that both boys and girls have higher w vertically extended households. In sum, the pattern of gender differentiatio to move females into vertically extended househ highest workload, while simultaneously moving tended households.7 This finding supports the n of children is driven by the desire that girls lea these households.

THE ROLE OF WAGE LABOR
What role does wage labor play in household work patterns? Since th 1960s, transfer payments from the United States have fueled Yap's wa economy. By 1988-89, when our research began, 58.6 percent of Yap Sta households had income from wages (Taverner 1990).8 Most of the jo were in the public sector. Thirty-one of the 35 households in our Yape sample included at least one member who earned wages. Collectively, t 35 households included 38 male wage workers and 26 female wage wor ers, in keeping with a generally high rate of female employment in Yap.
The vertically extended households in our sample have an average 1.50 wage workers per households, versus 1.96 for all other household The vertically extended households are also larger, and they have just .1 wage workers per capita, versus .41 wage workers per capita in all oth households (p = .022). With fewer wage workers per capita, it seems like that the vertically extended households are more dependent upon subsistence food production. Be that as it may, wage work does not have a direct effect upon the children's work patterns described, once we control for numbers of children and household form.
We now turn to the data for Kosrae, where we will discuss a different pattern of relations among gender, household composition, and children's work that occurs in a male farming system.

KOSRAE
Kosrae is a high volcanic island with lush vegetation and high rainfall averaging more than 5000 millimeters (200 inches) per year, with little seasonal variation. Our Kosrae sample focused on Lelu, the city of the former chiefly rulers of the island. Lelu is still inhabited by the descendants of these chiefly families. Today Lelu is the main urban center of Kosrae and one of the state's four municipalities. Our Kosrae sample includes 30 households from Lelu and ten from each of the other three municipalities. Kosrae is a primarily horticultural society relying on tree crops, especially breadfruit and bananas cultivated by male farmers, augmented by its rich lagoon fish and shellfish resources harvested mainly by women, with off-shore pelagic species harvested by men (Wilson 1968).
Today's Kosraean families trace kinship cognatically, although some echoes of the importance of matrilineal ties can still be seen. Kosraean families are large, children are highly valued, and households tend to extend over several generations. Families live about equally in nuclear or vertically extended households, which sometimes include four generations in Kosrae; only two households are two-generation laterally extended. In most cases there is a strong nodal household of the eldest generation, surrounded by households of the children and grandchildren, each with separate cooking facilities.
Although many imported foods are incorporated into the diet, local foods continue to be important, in particular breadfruit, which is prepared in an earth oven (inuhm) every week. Once a part of every established household, the earth-oven cookhouse was a feature of three-quarters of all households in our sample; other households share foods from the ovens of their extended family members. Families gather every Sunday for a meal of Kosraean foods including baked breadfruit, Kosraean soup (made of coconut cream, breadfruit, and a protein such as fish or chicken), banana, fruits, and various specialty foods. Sharing of foods and household labor among houses is common, and individuals frequently shift between households, and sleep and share food at other households. Both individuals and small family units may move briefly among related households for reasons such as attending funerals and special occasions. Young girls usually shift within a close circle of related family households, helping to care for new infants and elders, and helping with household tasks. Slightly older boys shift within the related households to help with and learn skills related to farming and tree horticulture.
Participation in wage labor is high, and is heavily concentrated in the public sector. Males are more involved in wage labor. The 1993 census counted 565 men and 158 women in government employment, and 436 persons (gender not specified) in private employment (Bureau of Planning and Statistics 1993). A third of those 15 years and older (total 4,089) worked in 1993 (48 percent of men, and 19 percent of women). Although women's wage labor is highly valued, Kosraeans attempt to provide employment for at least one family member, with an emphasis on providing such employment for males. In 1993, 81 percent of all families had at least one wage earner. Fifty-five of our 60 sample households had at least one wage earner, with a mean of 1.6 wage earners per household. Most (38) of these 55 households also participated in farming and/or fishing.

KOSRAEAN CHILDHOOD AND ADOLESCENCE
Kosrae has a high birth rate and children of both sexes are highly valued. A mother is expected to breast-feed her infant at least six months, and breast-feeding often extends to a year or more. If a new mother is married the young couple generally moves in with the husband's family. Less frequently they stay with the wife's family. Married couples establish their own households and cooking area as they are able, often adding a house within a compound on family lands. The residence choice depends on circumstances such as work-related housing requirements to land. If a mother is not married she will stay with her parents adopt the baby or help the mother to raise her children. The either stay with their mother when she eventually marries or sta parents. In the case of divorce the father has the strongest ri children. However, the children, especially girls, may accom mother if she returns to her parents' household. It is expect families will have children. Childless couples or individual adopt one or more children, and both natural and adoptive par ties with the child, who may during her or his life live alternatel set of parents. Of our sample population of 296 individuals un thirty-two (11 percent) were identified as having been adopte two-thirds were girls and the vast majority of those adopted were extended-family households.
Although mothers and teens have general responsibilities fo of infants and young children, men and boys of all ages also p caring for children, and it is common to see men holding an infants. Kinship terms follow the Hawaiian pattern, with the wor (ngatik) extending to all children of the descending generati extended kin group. Infants are held during much of their wa and soon learn to hold on as they are carried on the hip. Kosr holds generally have one large room, and may have additio rooms for married couples. Infants sleep with their parents. After or after the birth of a sibling, a child sleeps on mats in the large siblings, cousins, and older relatives. Relatively large numbers and intensive child care practices result in a high domestic wo women and young girls.
Generalized respect behavior toward parents and grandpar cludes special eating protocols by which the elders and honore served first, with family members and children served last. Sm assist in running errands and doing small gender-specific tas morning, for example, young children pick up leaves and cle area. Young girls assist in washing dishes and clothes; young bo farming tasks, and take care of pigs and other livestock. Fish trolling outside the reef) is mainly done by women, and girls pany their mother. Both boys and girls accompany their par farmlands to work and relax on Saturdays. Young girls especia caring for infants and small children. Around puberty both bo take increasing responsibility for completing household and work. At this life-stage, girls stay much closer to home than boys, who may move out to stay with other relatives. Saturda days, as the inuhm is prepared and breadfruit and other food in preparation for Sunday when there should be no work.
Sundays are set aside to attend church and Sunday school classes. This is also a day for visiting relatives and for the larger extended families to come together for meals that are prepared on Saturday. Religious instruction is very important, and with the exception of the small Seventh Day Adventist School is handled directly by the churches and their local ministers, elders, deacons, and teachers. Kosraeans attend church more frequently than most Micronesians. Children learn bible stories, hymns and song, on Sundays and perhaps one or more evenings of the week.
Work and play flow into one another, nowadays interrupted by school. Children swim and collect shells in the reef flats, swing and play in the sand, and play group games on the beach strand and within the family compounds. Among younger children the groups are generally of mixed sex. A few families have televisions and videos (during our research in 1992 there was no television station on the island); it is common for ten to 30 children to gather together to watch a video. In addition to rented videos, the Kosrae Historic Preservation Office has videotaped many cultural and historical events, and these videos are available for families to see. A few families have a computer at home, on which children might at times play a game. All households have radios, tuned to the Kosraean radio station.
Kosraeans are highly educated, with 42.6 percent of the adult popu tion having a high school degree or higher, and many individuals hold h positions on Pohnpei within the national government of the Federate States of Micronesia. Schooling is highly valued by Kosraeans, who se several hundred children overseas for higher education every year, wh they are noted high achievers. Older children are fluent in English, wh is the official national language. Children begin classes in Kosraean bu soon are taught in bilingual programs with English gradually introduced the lower levels. Many Kosraean households have members who have liv overseas or who travel back and forth to the national capital in Pohn In these houses English is often spoken part of the time with young c dren.
There are no public puberty ceremonies for either boys or girls. adolescence there is a partial segregation by gender, as girls tend to st with their parents while boys are moving into linked-extended househo As we discussed above, this kind of gender separation in adolescence i common in Pacific Island societies.

KOSRAEAN HOUSEHOLD DEMOGRAPHY
The Kosrae samples includes 60 households-30 from the town of Lelu and ten from each of the other three municipalities. Our sample included 28 vertically extended households, 28 nuclear households, and four laterally extended households. Virtually all (56) of the househo young residents under the age of 18. Among those households th of young people ranges from one to 13, with a mean value of 4.7 Table 4 shows the number of children in different household gender and age category, along with sex ratios. In this table we s after age ten. Among younger boys and girls (birth to age ten) we f boys living in two-generation households and more girls living in ve extended households. From ages 11 to 17, however, the girls sh two-generation households, while the number of boys in verti tended households slightly expands.9 This pattern is shown in which plots the percentage of Kosraean males and females living cally extended households across three age ranges.10 This pattern the opposite of the pattern in Yap. Can the different pattern be by the fact that Kosrae has a division of labor opposite from tha As noted earlier, Kosrae has a male farming system. Previous on the gender division of labor has shown that male farming is with an intensification of female domestic labor, such as food pr food processing, and weaving (Burton and White 1984). If more labor is required in nuclear households and more farm labor re vertically extended households, then the shift of pubescent girls households would make sense, as would the preponderance of bo vertically extended households with male elders.
As with the case of Yap, our work data are based on four weekly of a survey. In our Kosrae sample 55 out of households had one young people ages six to 17 during at least one wave of the surv these 55 households recorded some children's work, with a ran low of one hour during the four weeks to a high of 323 hours, or 80 per week. Table 5 shows the average hours of work per household for the three age categories of boys and girls. As with the Yap data, these are total hours in a four-week period.ll Kosraean girls do 61 percent of the children's work. Kosraean boys make their highest household work contributions between ages 11 and 14, whereas Kosraean girls make their highest con-  Figure 5. Gender composition of Kosraean households by age group and house tributions between ages 15 and 17, but also with high contribu tween 11 and 14. Kosraean girls' work is much more heavily con in domestic tasks than boys' work, especially in food preparation ping, while boys work is concentrated in livestock care and farm We computed work hours per capita and per week, as with data. These numbers are higher than in Yap, but we must remem they are based on a larger number of tasks, including kitchen shopping. Kosraean boys, age six to 17 average 3.9 hours of wee per capita, versus 5.4 hours for girls. Boys do about the same a work in each of the three kinds of households-4.20 hours in nuclear households, 3.58 hours in vertically extended households, and 3.95 ho in laterally extended households. Kosraean girls' work, however, vari greatly across household type-7.59 weekly hours per capita in nuc households, 4.31 hours in vertically extended households, and only 2. hours per capita in laterally extended households. Table 6 summarizes regressions of total hours of boys' and girls' wo on numbers of boys and girls of the three age groups, and on a dum variable for residence in a nuclear family household. Whereas Yapese g do more work when living in vertically extended households and do m subsistence production, Kosraean girls do more work when living in clear households, where they assist their mothers in domestic tasks. A their intensified workload occurs between ages 11 and 17, when they m out of vertically extended households. Hence, as in Yap, the adult divis of labor affects children's work in a way that can account for gender diffe

DISCUSSION
We have demonstrated systematic re of labor, children's work, household t male and female children. In both Yap residence emerge in middle childhood ages 11 and 14, and disappear by the la graphic differences in the experiences are strongly correlated with the patte especially female work. Girls and fem reside in the kinds of households whe farm and garden work is done by women ration is less intensive. There girls and farm and garden work, and are heavily households, where that work is most in NOTES