3 THE PREMI HOUSE

Ritual and Relatedness

Understanding religious practice in Bustling Township is impossible without examining the ways people conceive of social relationships. Ancestor cults are inextricably intertwined with notions of kinship or “relatedness”—to use a recently introduced alternative term.¹ Care and respect for lineage members continue after their deaths, while the souls of the deceased continue to interfere with the lives of their surviving descendants. In order to understand religious practice and cosmological beliefs in Bustling Township, it is therefore necessary to understand how relatedness is socially constructed.

There are two clearly prominent principles of social organization in Bustling Township. The first is the exogamous unit created through patrilineal descent, called jhü, probably best rendered as “clan.” The second principle, which is particularly central in the structuring of daily life, is a residential unit called dzèⁿ, or “house.”² Traditional anthropological concepts like those of “kinship” and “household” are not very helpful in explaining these empirically encountered forms of social organization. To a certain extent, the dzèⁿ are coterminous with the anthropological concept of households. They constitute domestic groups that share tasks of production and consumption, and they are tied together by descent and alliance (Carter 1984; Netting, Wilk, and Arnould 1984). But as several anthropologists have noticed, household members are not necessarily coresidents (Harrell 1997);³ moreover, one house may contain several households. The Premi in Bustling Township also have a word for the domestic group, the “single, bounded, multi-activity group” found in the villages (ibid.: 50): k’a (in Walnut Grove) or ga (in North Village). But while dzèⁿ clearly refers to the built structure, the term is fully interchangeable with k’a when it comes to designating the inhabitants of a house. Keeping to the indigenous concept of dzèⁿ in this analysis of Bustling Township society avoids the chicken-or-egg controversy of whether residence is a generative factor for the creation of the domestic group or co-residence is the result when such a unit exists (Levine 1988: 128; Carter 1984: 77).⁴ But Premi dzèⁿ not only constitute a local category for the “conceptualization and practice of social relations”; they also make up the primary centers of ritual practice and symbolic meanings (Thomas 1996: 281). Having names and being ascribed with the magical power to provide its inhabitants with prosperous lives, the Premi dzèⁿ furthermore possess animistic qualities.

FIG. 3.1. Cluster of houses in Bustling Township. The roofs are made of planks kept in place with stones. While these houses are made partially of stone, in many areas of Bustling Township, houses are made entirely of logs.

Houses conceived as basic units ordering social life are not unique to Bustling Township Premi. Ethnographic data on communities within the larger region of the Himalayas and their foothills actually confirm that patrilineality as a structuring principle often is combined with residentially based principles of social organization, or houses. The Kachin word htinggaw literally means “people under one roof” (Leach 1970). It can refer to one household and, interestingly, also to several households belonging to the same patrilineage. According to Barbara Aziz, in conceptualizing social relations in Tibetan D’ing-ri, the idea of residence is more important than that of descent (1978: 117–24). Elisabeth Hsu finds in the concept of “the house” an alternative framework for understanding kinship among the Naxi and the Na, both close neighbors of the Premi. These two ethnic groups are closely related culturally, and their kinship systems have been described as patrilineal for the Naxi and matrilineal for the Na.⁵ While the Naxi are virilocal or neolocal, the Na are traditionally duolocal or natolocal. Comparing the ethnographic studies of McKhann (1992) on the Naxi and Shih (1993) on the Na, Hsu discerns two coexisting local ideologies: a hearth-oriented ideology and an alliance-oriented ideology. The hearth-oriented ideology is concerned with harmony and consequently stresses the prominence of the house as the basic unit for all kinds of social relations (Shih 1993: 135). Alliance-oriented ideology is concerned with honor, rank, and hierarchy and is expressed in the distinction between raka (bone), which stands for the exogamous patrilineage, and na or she (flesh), which stands for the lineages of the affines. Making a clear distinction between ideologies and practices, Hsu sees the interplay between these two ideologies as creating the conditions for the flexible implementation of kinship practices (1998: 90).

In the anthropological literature in many other areas of the world, however, “houses” also turn up as central concepts in the study of social organization. Claude Lévi-Strauss was the first to put forward “the house” as a specific form of social organization.⁶ He proposed the theoretical concept of “house-based societies,” which several writers subsequently elaborated on and criticized.⁷ Relatedness is not necessarily the sole result of fixed descent rules; it is continuously “under construction” through shared participation in everyday activities by, for example, the inhabitants of a common residence (Carsten 2000: 18). Everyday activities in a Premi house involve ritual offerings of food to the ancestors, thereby constructing and strengthening ties between the co-residents.

MARRIAGE AND RELATEDNESS IN BUSTLING TOWNSHIP

The traditional age for marriage in Bustling Township, as in many other Premi areas, is thirteen. At that age, children also enter into the world of adults. In Bustling Township, boys and girls go through a ceremony called “wearing trousers” and “wearing skirts,” respectively. When Premi girls become thirteen, they start wearing the traditional pleated skirts as well as braiding their hair and interweaving it with black thread in the typical and recognizable style of Bustling Township. The current age for marrying is now slightly higher, around seventeen on average. I registered marriage ages between fifteen and twenty-one. Although the official age of marriage is eighteen for women and twenty for men, people get around the limits by waiting to register their marriages or by not registering them at all. In cases of brothers or sisters marrying one spouse, they would register only the oldest sibling. Often, when I asked about how partners were chosen, people gave the standard answer that marriages were based on free choice. In Walnut Grove, where I got to know people well enough to make such political correctness unnecessary, several people told me that parents were in charge of choosing marriage partners for their children. They usually made these arrangements at the ceremony held when the children turned thirteen. There were only three well-known cases of couples in Walnut Grove who had married after choosing each other. Although I did not specifically pursue the matter, premarital and extramarital sexual relations do also occur, as testified by a few unmarried mothers and persons with uncertain clan identity.

There is no word in Premi for “marriage.” This does not mean that there is no concept of a legitimated sexual union in Premi society. When expressing the fact that one is married, Premi people say, “A pri getyèⁿsaⁿ,” which literally means “I have drunk pri” (pri or tri is a locally brewed beer). When saying, for example, that one got married at the age of eighteen, one would say, “A ga-hsüè-go k’o-nyè pri getyèⁿsaⁿ,” which means “In the year that I turned eighteen, I drank pri.” The explanation for this way of describing a formal union between two or more partners is that pri was consumed only rarely in the old days, and the drinking of pri was a major part of the marriage ceremony, which celebrated not only the relationship between the partners but also the establishment or strengthening of relationships between two families and two jhü, or exogamous clans. Asked to describe how they were planning to marry off the last of their four daughters in the coming spring, the parents in my host family explained that they would invite an anji to preside over the ceremony and perform several rituals to bless the union. The whole celebration would last six days: three days at their house and three days at the groom’s house. On her journey to the house of her future husband, their daughter would be escorted by her brother and ceremoniously handed over to her new house and new clan. They would not have to pay any real dowry, but their daughter would not be entitled to any share of the land. The families would exchange only a few symbolic gifts: arje, or homemade distilled spirits, dried pork meat, and homespun hemp cloth. If they managed to save some money in the coming months, they might give her a bit as a dowry, but they would never pay more than ¥100, since this was not the custom in the village. After marriage, their daughter would move in with her new husband, either in his parental home or in a new house when they found some land to build on. In Lévi-Strauss’s scheme of elementary structures of marriage, Premi marriages are definitely of the symmetric exchange type in which clans exchange women but no material goods. The major advantage of establishing affinal ties is that it facilitates future exchanges of women. Once families in two neighboring villages—whether they are Premi, Na, Naxi, or Rek’ua—have established their first marriage relationship, they will try to continue the exchange of women, and cousin marriages constitute the most obvious opportunities for such exchanges.

There is at least an ideological preference for cross-cousin marriages, which is also reflected in kinship terminology that makes the distinction between parallel and cross cousins. On the male side, the term apoⁿ is used for father, father’s brother, father’s sister’s husband, and mother’s sister’s husband and also designates the father of one’s wife, but one’s mother’s brother is called agu. On the female side, the women in one’s mother’s generation are all called ama— or ma for one’s mother—except for one’s father’s sister, who is called ane.⁸ Such distinctions do not exist in one’s own generation, in which there is no distinction between cousins and siblings: pep’ei is used for one’s older brother, older sister, and older male and female cousin, while gwèⁿgwèn is used for the cousins and siblings younger than oneself, also without regard to sex (see fig. 3.2). In practice, cross-cousin marriages are not all that frequent. Among forty-six marriages in Walnut Grove, I registered only five such marriages. With only two registered cases, there were even fewer marriages between parallel cousins.⁹

Polygamous marriages were frequent in Bustling Township. These were mainly instances of fraternal or adelphic polyandry, in which two or more brothers take one wife, with a few cases of sororal polygyny, in which sisters marry one husband. A prevalence of more than one-fourth of all marriages in the southern villages—one of which is Walnut Grove—suggests that fraternal polyandry is a preferred form of marriage in this part of the township. This is also the most ethnically diverse part of the township, and polygamy was as prevalent in Na, Rek’ua, and Naxi villages as in Premi villages (see table 3.1). All four instances of sororal polygyny I recorded were from the northern villages, two of them in one family from Hill Village: one of the two anji in the village had two mothers who were sisters and was married to two sisters himself.

While I have no data substantiating that this form of marriage has been practiced among the Premi in the neighboring province of Yunnan, Muli and the adjacent Litang (Lithang) area have long been known for the practice of polyandry. According to a survey by Wu Wen in 1956, the prevalence of brothers or sisters marrying a single spouse was more than 20–30 percent in Muli,¹⁰ where the practice is prevalent among Premi, Na, Rek’ua, Gami, and Naxi. In a 1982 marriage survey of the Naxi of the township of Eya in the southwestern corner of Muli, Liu Longchu has registered 46 households of a total of 130, or 35.4 percent, that practice fraternal polyandry (1986). Polyandry has also been widely documented throughout Tibet and among Tibetan communities in adjacent areas such as Nepal (see, e.g., Peter 1963; Goldstein 1971, 1976; Aziz 1978; Levine 1988; Ma Rong 2001). In recent years, the practice has seen a revival in some rural areas of central Tibet (Benjor 2001; Fjeld 2007).

FIG. 3.2. Kinship chart showing Premi terms of relatedness.

TABLE 3.1 Number of Polyandrous Marriages in Southern Villages in 1998

Most studies of polyandry in Tibetan areas base their explanations for this rather rare form of marriage on economic and social causes. Fraternal polyandry is seen as a peculiar adaptation to specific ecological circumstances. In the highlands of Tibet, the area suitable for agriculture is strictly limited and cannot sustain an increasing population. All the sons in a family are entitled to inherit a share of the family land. In order to avoid having to partition this land into smaller and smaller shares, and because theynare unable to open up new arable land, Tibetans practiced what Goldstein termed the “monomarital” marriage principle, by which there can be only one marriage that produces heirs for each generation (1978: 208). This would in most cases be fraternal polyandry, but there are also other possibilities. For example, when there are no sons in the family, daughters can inherit by marrying uxorilocally. If there are several unmarried daughters, they can marry one husband in sororal polygyny. Such factors must be seen in combination with other factors such as the tax system and prevailing political structures, according to Goldstein. He also shows that on a larger scale, fraternal polyandry significantly decreases fertility and functions as a mechanism for reducing population growth (1976).

In her study of the Nyinba in Nepal, who are of Tibetan origin, Levine argues that polyandry cannot be explained by economic and demographic factors alone. To a certain extent, her findings support Goldstein, and she demonstrates how fraternal polyandry makes it possible for Nyinba households to engage in both trade and agriculture and therefore be more prosperous then their Nepali neighbors, who do not practice polyandry. At the same time, Levine also shows that culture is involved as an explanatory factor, and indeed, if socioeconomic factors were the only determinant, we would expect to find more communities around the globe choosing this special form of marriage. There are other strategies available for dealing with the socioeconomic constraints. For the Nyinba, kinship is seen as “providing the most reliable basis for trust,” and in this way it is an important notion in structuring social relations, including marriages (Levine 1988: 278). The result is the cultural validation of brothers staying together, even at the cost of renouncing sexual exclusivity.

In her rationalization of polyandry among the D’ing-ri Tibetans living in exile in Nepal, Aziz is in line with both Goldstein and Levine: she points to the economic strength of the households based on polyandrous marriages (1978: 134–45). Dividing a household diminishes its prosperity, while well-managed polygamous unions increase prosperity. She also provides an interesting suggestion of how polyandry might have come about, related to the definition of the exogamous group, or pün. D’ingri Tibetans cannot have sexual relations with or marry anyone to whom they can trace bilateral descent. This excludes a large section of the local population and restricts the choice of marriage and sexual partners. At the same time, it makes anyone who is not kin a potential sexual partner, and casual sexual unions in which one’s kin are permitted to share one’s spouse are condoned and, according to Aziz, are the first step to polygamous marriages. An ideal that brothers should not be divided further fosters such unions.

The Bustling Township data do not unequivocally underscore this factor of limited access to marriage partners. Among the Premi in Bustling Township, the exogamous unit is established not bilaterally but patrilineally. Clan membership signifies, first of all, a factor of exclusion, determining which categories of people one cannot marry or have sexual relations with. Since exogamy is required only unilineally, and since most villages are made up of several clans, in principle, this provides villagers with enough potential marriage partners. But in reality, several factors limit choice. In one village there is one very large clan, and the other clans are very small. In another village, there are only two clans. Furthermore, Bustling Township marriage data indicate a preference for finding partners close to home, which considerably limits the choices. Of the forty-six marriages I registered in Walnut Grove, fourteen were among the three different clans within the village, and the rest of the marriages were between men from the village and women from outside. Twenty-two of those were with women from the three surrounding villages—one Premi, one Rek’ua, and one Naxi. The other ten women came from nine different Premi villages, all no more than a day’s walk away. The majority of this last category of marriages were within three families in Walnut Grove who could not marry locally because they were thought to be possessed by demons (see chapter 4). All of the registered marriages in Uphill and Downhill were between local men and women from nearby villages. Village clusters can therefore be considered basically endogamous. These combined factors might have given rise to an ideology that supports marriages in which same-sex siblings share a single spouse.

The emic argument given for polyandry in Bustling Township is of a materialist nature and tallies with the socioeconomic explanations already mentioned. Asked about the reason for choosing polyandrous marriages for their children, parents explained that when brothers stay together, the land need not to be divided, and polyandrous marriages are a good strategy for maintaining and increasing household wealth. The presence of several men in the house makes it possible for the family to be involved full-time in agriculture as well as to transport goods and find jobs in the market towns of Yongning and Wachang or the county-town. While there were a few older polyandrous and polygynous couples established before the Cultural Revolution, a substantial number were contracted within the last ten years. This is a remarkable new trend in view of the fact that polygamy is forbidden according to Chinese marriage laws, as discussed in chapter 2. Nevertheless, the trend is apparently not limited to Bustling Township. Recent field studies from rural Tibet and adjacent areas confirm the continued occurrence of fraternal polyandry and sororal polygyny.¹¹ In a recent study of a rural area in Shigatse Prefecture in central Tibet, the Tibetan anthropologist Benjor describes a strong revival of traditional polyandrous marriages (2001). He demonstrates that the de-collectivization of the early 1980s, which again made households the basic units of rural production, was the catalyst for this revival. Although households did not own the land, they received long-term contracts allowing them to use it.¹² However, since land could be neither sold nor bought and no new land could be opened up, there were no means of providing enough land for an expanding population. The traditional practice of polyandry then became a compensatory strategy for keeping the family land together and facilitating the sending of household members to engage in nonagricultural wage labor (ibid.: 195). Why people would choose precisely these strategies instead of other possibilities, Benjor leaves unanswered, overlooking the workings of cultural variables in his socioeconomic analysis.

Heidi Fjeld, who conducted fieldwork in Panam in rural central Tibet, confirms the findings of Benjor and also notes a strong increase in polyandric marriages, making it a preferred form of marriage among the studied population. But unlike Benjor, Fjeld goes beyond one-dimensional economic and ecological perspectives by placing her results in a larger context of sociocultural change. The practice of polyandry was traditionally limited to the leading classes of local landholders and government representatives. Fjeld suggests that polyandry today has become part of a process of social transformation and upward social mobility for common farmers: families view the adoption of the customs of the old upper classes as a means to both social and economic success (2007).

Many of Benjor’s findings are directly applicable to the Bustling Township situation. After de-collectivization, Bustling Township households were given land according to a formula of 1.1–1.3 mu per person in the household. Although, because of the lower altitude and more humid climate, Bustling Township does not face the same constraints on agriculture as does central Tibet, there is little level land available for an expanding population. Those mountain slopes that have not yet been brought into production are very steep and difficult to access. Constructing terraces and providing adequate irrigation would be very labor-intensive tasks in themselves. However, the major constraint is government policies on land use. First of all, many of the areas that could be opened up are forest-covered and cannot be exploited because of the Tianbao policy of 1998. Second, on the local level, the township government carries out a very strict land use policy aimed at upgrading the environment in order to avoid further erosion of formerly deforested areas. Land that is deemed at risk of being washed away by the summer rains can no longer be used for agriculture and must be planted with fruit trees or reforested. Consequently, there are several socioeconomic incentives for keeping the population as constant as possible and avoiding division of the land allocated to the family. The traditional practice of polyandry answers both needs; however, this explanation does not fully answer the question of why polyandry has become one of the acceptable remedies. There are other possible solutions that would enable villagers to keep the land in one piece, used by some of the neighboring ethnic groups of the Premi, including, for example, the custom of primogeniture, by which the firstborn son inherits all the land.

As with monogamous marriages, polyandrous marriages in Bustling Township are always virilocal or neolocal, and an in-marrying woman becomes part of her husbands’ clan. This implies that her children can marry the children of her brothers and of her sisters, if the sisters married into a different clan. Uxorilocality exists in a few exceptional cases, when a family has no male descendants: after the other daughters have married out to other houses, the husband of the remaining daughter will move into her house but will continue his own patriline. Extramarital sexual relations also strictly observe the rules of clan exogamy. Therefore, when a child is born out of wedlock and when the mother does not want to disclose the father, her child should preferably marry someone from the mother’s paternal clan, since this is the only clan that the child’s father could not belong to. In this way, the rules of clan exogamy continue to be observed.

In Uphill and Downhill, the thirty-five houses (out of forty) for which I registered jhü-mè, or clan names, were divided into seven different exogamous clans; the forty-three houses of Walnut Grove belonged to three different clans; and in one of the neighboring villages of Walnut Grove, Five Nut Trees, the fourteen families were divided into only two clans (see table 3.2).

Clan genealogies are orally transmitted, but those of the larger clans are also written down by the anji, using Tibetan script. The oldest living member of the Ak’ua clan in Walnut Grove could easily recite from memory the agnatic ancestors of twenty generations, not only of his own patriline but also of related lineages. The longest genealogies go back thirty-seven generations to Jiuwu Jiudaⁿhsi, the common ancestor of the different ethnic groups, followed in the thirty-sixth generation by the first Premi, Hsidiⁿ.

Several of the Premi clans trace their local origins across ethnic borders. Although the Ak’ua clan had a remembered history of more than twenty generations, its history in Walnut Grove began only six generations ago, when a young man from the neighboring Naxi village moved to Walnut Grove and married two Premi women, one from the Mesé clan in Walnut Grove and one from a third village farther away. Over the years, the Ak’ua clan became culturally assimilated into the Premi, but its Naxi origins were remembered and solidified through its existence as a separate clan in the village. The Ak’ua and Mesé were each the other’s most important exogamous group: of the forty-six registered marriages in Walnut Grove, nine marriages, or about 20 percent, were between these two clans. The third clan in the village, the Bot’a clan, also had Naxi origins, and its history in Walnut Grove started just four generations ago. The four houses of this clan were much less culturally assimilated into the Premi than was the Ak’ua: people still spoke Naxi at home, there was a larger degree of intermarriage between the Bot’a men and women from the Naxi village, their genealogies were kept by the Naxi dtô-mbà religious expert from that village, and this expert, rather than one of the anji from Walnut Grove, also performed the ceremonies at the New Year celebration.

TABLE 3.2 Clan Affiliation of Houses in Three Villages in Bustling Township

One of the two clans in the village of Five Nut Trees also had a nonPremi origin. The village was established only three generations ago, when Nadjoⁿ arrived from Yongning and built a house on a small stretch of sloping land on the mountainside facing Walnut Grove. He married a Premi woman from the Shoⁿ clan. This became the first house of the Gonga clan. The Gonga obtained their clan name because their ancestors originated from Shuiluo, an area in the west of Muli where the famous Gonga Mountain (Gongaling) is situated. They were of Gami origin, a local designation for Kham-speaking Tibetans in Muli. The story of how Nadjoⁿ ended up in what became the village of Five Nut Trees contains a theme common to Tibetan and Tibeto-Burman stories that are told to explain village or kin origins when brothers leave their ancestral home in search of a new home (Levine 1988: 30; Wellens 1998: 28). Nadjoⁿ’s nephew, Gonga Anna, told me the story:

Seven generations ago, there were five brothers in Shuiluo. As was customary among the Gami, the oldest brother inherited everything, and consequently the four other brothers had no place to live after their father died. As a result, they left Shuiluo and spread to different areas. One of those four was Dadre P’intsu, the ancestor of Gonga Anna. In the beginning, the brothers came back to Shuiluo once a year at the time of the New Year. But after some time, Dadre P’intsu stopped going, so it is not known what happened to the other brothers. Dadre P’intsu moved to Yongning in Yunnan. Five generations later, Nadjoⁿ, Dadre P’intsu’s descendant and Gonga Anna’s uncle, moved to Five Nut Trees.

It probably did not take long before Nadjoⁿ’s descendants were neatly fitted into the Premi system of exogamous clans in Five Nut Trees. Women from the Gonga clan married into the different clans of Walnut Tree and villages in the surrounding area, and several women from Walnut Tree married into the Gonga clan of Five Nut Trees.

There is some degree of segmentation, and segments cannot intermarry. The Mesé clan is actually a localized segment in Walnut Grove of the larger Ts’uop’i clan. The segmentation of the Mesé out of the Ts’uop’i is, at least according to local stories, closely tied to the origins of the village of Walnut Grove. The word me-sé literally means “people-kill,” and this connotation of “murderer” is explained in a story that not many people in Walnut Grove were happy to recount. In fact, people in the first houses I visited in Walnut Grove told me that they belonged to the Ts’uop’i clan; only after my local assistant made it clear that I knew the story of Mesé origins did they tell me that they belonged to the Mesé clan. According to the story, it all started in the village of Jiaze in Yunnan, not far from Ts’uop’i Mountain, the sacred mountain of the Ts’uop’i clan:

A long time ago, there were two brothers in Jiaze belonging to the Ts’uop’i clan. After partitioning the house, the youngest brother, Moⁿmoⁿ, did very well and had many animals. The oldest one was not that smart and was therefore very poor. Every night, he went out to steal some animals from his younger brother. Moⁿmoⁿ could not find out who the thief was, so one night he hid himself among his sheep, carrying a big knife. When the thief finally came, it was already pitch-dark, and therefore Moⁿmoⁿ did not see whom he killed. When the father found out Moⁿmoⁿ had killed his own brother, he became so angry that he chased his son out of the house and out of the clan. That is how Moⁿmoⁿ came here and started the first house in Walnut Grove, and out of shame, he kept the clan name Mesé.

The story does not correspond entirely with the genealogies I recorded, according to which the break between Ts’uop’i and Mesé happened nineteen generations ago, whereas the first house in Walnut Grove was established only six generations ago. There are also variations of the story, in which Ts’uop’i and Mesé are the names not of clans but of two brothers. Whether myth or historical fact, the story functions today as a rationalization of why two clans with different names cannot intermarry.

In addition to acting as a marker of the exogamous group, clans become visible as ritual units when one of their members dies. In Uphill and Downhill, the Ts’uop’i clan has its own place for cremation, called bu-dzu, while the other clans share another spot near the village. In Walnut Grove, each of the three clans has its own cremation place. There was no fixed place for cremating the body in North Village; instead, a ritual specialist had to help find a spot for each occasion. The cremation ceremony lasts up to three days and is the major ritual event in Bustling Township. As many ritual specialists as the family can afford will recite almost uninterruptedly and perform numerous rituals. Finally, after the corpse has been cremated, the ashes are collected in an urn together with pieces of certain bones: a piece of the skull, the hand, the hipbone, and the foot. This action of collecting the bones is an important element of Premi ontology in that it has a direct connection with understandings of relatedness and clan identity. Raka, or bones, is the substance that is transmitted through the male line, in contrast to na or she, or flesh, the other constituting part of a human being, which comes from the female side. Small parts of the raka, the substance of the male line, are all that is left in recognizable condition after cremation and will be cared for by those who also share this substance in their bodies, that is, the surviving male clan members. On several occasions, Premi living in the vicinity of Nuosu people (who also cremate their deceased) stressed that the Nuosu did not collect bone remains. It was viewed as a sign of incomprehensible lack of respect for the ancestors and an overall marker of a less developed ethnic group. After the pieces of bone have been placed in the urn with the ashes, the older members of the clan carry the urn to the mountain cave belonging to the clan.¹³ If the deceased was a woman, seven female clan members accompany the urn; nine men perform this task if the deceased was a man.¹⁴

Death is one of the few events—apart from choosing marriage partners for one’s children—when clan membership becomes salient, and this is reflected in ritual practices. According to local sayings, when a person dies, the bones become stones, the hair becomes trees, the flesh becomes earth, but the me-drö, or soul, remains. In addition to taking care of the tangible remains of the deceased after cremation, someone must look after this intangible substance, the me-drö. The me-drö may be understood as the soul of the deceased (memeans “person,” and drö was rendered into Chinese as lingwu, corresponding to “soul”). This me-drö has to be guided by the anji to a place where all the souls of the clan’s ancestors reside. This place, which is called Jewopöjedaⁿ, is said to be an actual place far to the north of Muli where the Premi originated. A ritual expert, the anji or the yèma, or what might be termed a Buddhist lay priest, recites the hsip’u text for the “opening of the road” ceremony, or drwashu, for the me-drö; like a real guidebook, it contains place-names and descriptions of the route to be followed.¹⁵

This whole process during which the me-drö has to find its proper place among the clan’s other ancestral souls is a risky time for all the living clan members. Correct rituals have to be performed so that the me-drö will not turn into a wandering spirit or, in the worst case, an evil wandering spirit, a shep’a. Furthermore, from this point onward, surviving kin residing in the house where the deceased lived will have to bring offerings every day and on special ritual occasions to his or her me-drö, in order to prevent it from turning into a shep’a who will cause afflictions among the survivors. It is shep’a who cause people to die, and one of the tasks of the anji is to figure out what kind of shep’a has been responsible for the death. By carrying out the appropriate ritual, the anji can chase the shep’a away and prevent it from taking the me-drö with it. He does this by invoking powerful deities, such as Hladaⁿ Soⁿma (see next chapter). But more must be done: in all houses belonging to the clan of the deceased, a small thread-cross, a nungk’e, is placed somewhere high on the walls (see fig. 3.3).¹⁶ It is used to prevent the me-drö of the deceased from enticing the me-drö of its surviving clan members into joining it on its journey to the place of the ancestors. The nungk’e stays on the wall for one or two years.

The idea that deceased kin interfere with surviving descendants who do not behave appropriately is found all over the globe. The villagers of Bustling Township conceive of the me-drö’s effect on their lives in a way that is strikingly similar to the African Ndembu’s view of the influence of their ancestral spirits. Turner conceptualized these spirits by borrowing the term “shades” from Monica Wilson’s study of African societies,¹⁷ expressing both the immaterial character of these beings and their relationship to the more tangible manifestation of the ancestors as human beings (Turner 1977: 11). The shades of the Ndembu are “the uneasy inhabitants of the ‘unquiet grave,’” and they leave the grave to plague their kinsfolk because they have been “forgotten” or their kinsfolk have acted in a way of which the shades disapprove, such as neglecting to make offerings, quarreling with other members of the kin group, disobeying a wish expressed by the deceased in his or her lifetime, or moving away from the village of the deceased. The afflictions brought on by the shades include causing diseases and infertility or spoiling the hunting (Turner 1967: 9–11). Although me-drö seem to be more easily handled by correct ritual behavior than are the shades of the Ndembu, as explained later in this chapter, the mechanism whereby fear and respect for the ancestral spirits influence social behavior and thereby are part of the process of establishing social morality is similar in both cultures.

While showing respect to the me-drö is—other than at the time of the cremation—mainly a matter for close descendants of the deceased, there are a few ritual activities for worshipping the common ancestors of the whole clan. Today, only the larger clans such as the T’aba and the Ts’uop’i hold yearly clan meetings on the fifteenth of the seventh month of the lunar calendar. Ideally, one male representative of each house of the clan attends the meetings, which take place near the mountain cave with the urns containing the ashes of deceased clan members. In the case of the Ts’uop’i, a large clan that originated in Yunnan, these yearly meetings are sizable events with more than fifty people participating and are reported on by the local newspaper. The meeting takes place in Yunnan at Cuopidian, near Chabulang in Labai Township. I have been told that during these meetings, ritual specialists hold ceremonies to worship and propitiate the different lines of the clan—up to the earliest common ancestors—and ask all the ancestors to protect their living descendants and bring prosperity to them.

In daily life, however, and in the organization of social life, the role played by clan membership, or descent in general, is not immediately obvious. In Walnut Grove, an often-used alternative expression for jhü, the standard word for clan, is arè dazeze, “we of the same descent.” This points to an awareness of common descent as a factor of common identity, but this is not reflected in social organization. Except for when a person dies, there are very few ritual occasions centered on the clan. Neither is the clan a corporate unit with economic or status connotations: the clan does not manage any property, and, at least in Bustling Township, descent does not seem to have any correlation with class. In Muli, there was one ruling clan, the Bar clan, which provided the ruling elite and kept slaves at its estate in southeastern Muli. But no Bar lived in the Bustling Township area, and there was no significant internal stratification within the majority of Premi commoners. In fact, there are a number of folk sayings expressing the idea of social equality, such as “Just as all eggs are equal in size, so are all Premi equally high” (Yin 1989: 28). Traditional systems dividing society into descent-based classes did exist among several of the neighboring peoples of the Premi. To the west, the Tibetans discerned four classes: aristocrats, priests, commoners, and outcasts.¹⁸ The extent to which these classes are related to the four Hindu castes is a topic of discussion within the field of Tibetan studies (Aziz 1978: 52). But among populations even farther away from India, like the Na of Yongning or the Nuosu of Liangshan, sharp social divisions vested in descent were also prevalent at the time of the Communist takeover. According to Yan Ruxian, the Premi living among the stratified Na of Yongning under the rule of a tusi, the hereditary native chieftain, were all considered commoners. While Na society was divided into sipei (aristocracy), zeka (commoners), and e (slaves), the Premi were all considered to be outer-zeka, signaling their position as commoners not to be enslaved as well as distancing them from the Na class system. The Premi enjoyed a privileged position under the tusi system in Yongning and had a special right to rise in protest against the tusi (Yan and Chen 1986: 16–18). One of the stories told among the Premi explaining their privileged position is related to the Ts’uop’i clan, according to which a member of the Ts’uop’i clan from Tuodian, Ts’uop’i Nyima Gyatse, once saved the tusi of Yongning during a battle against the Naxi Mu tusi from Lijiang by hiding him in a leather bag. Out of gratitude, the tusi made Nyima Gyatse an uncle. From then on, there has been a close tie between the Ts’uop’i clan and the tusi, as expressed in the participation of the Ts’uop’i in special rituals held at each succession of the tusi in Yongning (see also Ninglang Yizu zizhixian zhi 1993: 223; Yan and Chen 1986: 16).¹⁹

While there are clearly differences regarding traditional social stratification between the Na of Yongning and the Premi of both Yongning and Muli, the system of matrilineal descent groups among the Na, called sizi, bears some resemblance, at least on the functional level, to the Premi patrilineal clans. Shih Chuan-kang describes the sizi of the Na as comprising several yidu, or domestic groups, formed by descendants of an ancestress three generations removed from the eldest living member (1993: 116). Shih notes that this sizi is not corporate in an economic sense since it does not have any common property or economic activity (117). The corporate character of the sizi is apparent only in terms of its members’ collective obligation to worship the common ancestors and in defining the outer limit of the incest taboo. When Shih questioned his informants on the meaning of the sizi, their answer provides a poignant illustration: “to conduct funerals together, nothing more” (119). The clan system of the patrilineal Bustling Township Premi seems to be more elaborate than that of the matrilineal Na of Yongning: Premi genealogies are traced back much longer than three generations, and the Premi hold yearly clan meetings. But this discrepancy may only be a more recent development: Shih’s older informants mention annual worship ceremonies for common ancestors for the whole sizi (119).

When larger ritual activities take place in Bustling Township, the whole village is involved, not just one clan. For example, during the festival that occurs in the second month of the lunar calendar, the whole village spends two days out in the open, and everyone participates in propitiating the water and mountain deities. Larger rituals are also held occasionally in times of severe drought, when the anji leads the village in praying and making offerings to the water deities, the lwéjabu, in the hope that they will make it rain. Furthermore, the larger nonritual activities involving people from several houses, such as construction of a new house or maintenance and repair of the common irrigation and water supply systems, are never organized on the basis of clan membership. Group mobilization for such activities is often based on temporary recruitment of people with specific skills, regardless of kin relationship to the mobilizing individual, household, or village. Otherwise, the kindred of the head of a house (that is, those people considered to be closely related through both descent and alliance) are called upon to assist. As several anthropologists have pointed out (see, e.g., Holy 1996), such assemblies of close kin and affines are temporary, and, in Bustling Township, they do not form any stable culturally recognized units. They constitute, to use Bourdieu’s distinction, practical kin as opposed to official kin (1977: 33–38).

The basic entity of ritual activity and at the same time the most visible unit of social organization is then neither the clan nor the practical kinship unit but a unit based, first of all, on common residence, namely, all the inhabitants of a single house. Societies as diverse as the Zulu in southern Africa or the Kayapo of central Brazil have been found to have “houses” as a central principle for structuring societies. In such a context, the term “house” does not refer to the architectural space in itself but to an analytical category for understanding social relations in a given society or community. Rather than framing empirical data collected in Bustling Township within any master theory of “houses,” taking a brief look into some of the discussions of houses in anthropological writing can assist in conceptualizing Premi dzèⁿ and the role they play in the organization of Bustling Township society and the performance of ritual practice.

DZÈ^N, THE PREMI HOUSE

As in the case of the D’ing-ri and Dechen Tibetans, Premi houses have names, dzèⁿ-mè (the more commonly used term) or k’a-mè. Mè means “name,” and dzèⁿ refers to the built structure. The word k’a (or ga) specifically designates the people living in a house, the domestic unit. For example, during the fieldwork for this study, I frequently asked the question: “How many people are there in the village?” (Ni-er jiù me chedze re?). Often, the interviewees did not know the exact number and would give replies such as “Our village consists of twenty-five domestic units” (Er jiù nóno-gwa k’a re).

The compound words dzèⁿ-mè and k’a-mè are used completely synonymously. When asking a person about his house name, one could ask either: “You (suffix) what domestic unit name?” (Niè bi miè k’a-mè?) or “You (suffix) what house name?” (Niè bi miè dzèⁿ-mè?). This is not without importance: the house equals the domestic unit as a major marker of a person’s belonging in society. House names are an integral part of the names of all the people living in the house.²⁰ Within a village in Bustling Township, a person is addressed first by his house name, then by his personal name: Jiaⁿhsi Dudjits’er and Jiaⁿhsi Druma are man and wife and live in the Jiaⁿhsi house. Dudjits’er and Druma are rather common names in Bustling Township, but as Premi have no family names, one of the functions of house names is to make a further distinction. The origin of a house name is usually the name of the one who built the house or, if there has been a house on the same location before, the one who built the original house. Few houses in Bustling Township are more than thirty years old; therefore, this name most often refers to a close agnatic ancestor.

When it comes to “splitting the house,” or dzèⁿ-p’o—namely, when one of the sons moves out of the house to start a neolocal residence with his wife and children (if any)—a son can either use his own first name for the new house or keep the name of his ancestral home and add dzèⁿ-hsi or ga-shi (new house), as in the case of the Azha Dzèⁿhsi house. The name of the ancestral home is often such an integral part of a person’s name that a new house might carry both the former house name (the name of the builder) and the son’s personal name. For example, Galoⁿ’s grandfather, Yowji, built a house carrying his name. Galoⁿ’s two youngest sons, Dadre and Drala, shared one wife and kept living in the Yowji house. They had two daughters and five sons in all. One daughter married into another house in a neighboring village; the other daughter married into a house in the village. The youngest son, Pema, stayed at his paternal house after marriage. The other four, including Galoⁿ, moved out and established new houses in the village: Yowji Galoⁿ, Yowji Yongjong, Yowji Ashi, and Yowji Yangtsu. In the other four Yowji houses, the ancestral house was then often referred to as Yowji Pema.

House names are thus straightforward references to the person who built the house, the personal name of either a man in the house or one of his agnatic ancestors. They might in addition carry a reference to an agnatic ancestor, by combining the name of the ancestor with the name of the builder²¹ or by adding the word for “new house.”

Property is inherited in the male line. The land is divided when a brother leaves the house, and the house itself is usually taken over by the brother who marries last, often the youngest brother. Women do not normally inherit, except when a household has no male heirs; in that case, the unmarried daughter takes over the house and the land rights, and when she marries, the husband can move into the house of his wife or wives (most cases of sororal polygyny involve two sisters taking in a husband). Although he continues his own patriline and the children are part of the clan of their father, the house and its inhabitants continue to carry the original house name. This special case makes clear the principle that the named residential unit is not considered a symbol of the patrilineage but is instead an independent category. Moreover, the house name—which all people in a house share, whether or not they are linked by descent or alliance—is not shared with close patrilineal kin living in other houses. Therefore, from an emic point of view, the dzèⁿmè links people together and expresses relatedness on the basis of common residence first, even though co-residence is most likely the result of common descent and alliance.

There are also traces of another type of house name. Several villages in Bustling Township have house names that hint at a more complex way of conceptualizing houses. The houses of three brothers all start their names with the word besé, which means “village leader.” Before the Communist takeover, besé was the name of the house of a village leader. The rhetoric of the Communist campaigns and the attendant political indoctrinations not only effectively erased this institution but, unfortunately, also make it difficult to investigate. It is possible that besé might be a local pronunciation of an institution that is rendered in Chinese as maise. The maise was an administrator, appointed by the office of the tusi or head lama, responsible for a population that ranged in size from a few households to several tens of households. His job consisted of transmitting orders and pressing tenants to pay rent and provide horses and corvée labor. Unlike the position of village chief, which was hereditary, or that of rongban—the dispatched monk bureaucrat working at the village level—which was limited to three years, there was no fixed period of office for the maise. If the administration of the head lama found it convenient to do so, it could extend the term. It is unclear whether the position of village leader could become hereditary, and several informers maintained that individuals were chosen for such a position because of what could be termed their moral prestige and social competence, such as a demonstrated ability to lead and mediate. But different villages offered different explanations. According to Besé Nima of Walnut Grove, village leaders were appointed by the boⁿ (T: pön[po])—or the head lama, as he was also called locally—and if they did a good job, the position could become hereditary. This was the case with the family of Besé Nima, who had been village leaders for many generations.

Besé did not receive any payment from the Muli monastic government and supposedly had no special privileges. One of their major tasks was to divide the burden among the villagers when representatives from the government visited the region, either just traveling through or collecting taxes. It proved impossible to verify whether “village democracy” was merely the origin of an institution that later developed into a hereditary position, or whether this was a way of presenting a pre-Communist institution in a more acceptable light. The fact that the besé were only mildly criticized and did not suffer much under the Cultural Revolution points to the former explanation.

There are not one but three houses in Walnut Grove that carry the word besé in their names. After the traditional village leader system was abolished during the Democratic Reforms campaign in 1956, the original meaning of the name Besé lost its politico-legal connotations and today functions like any other house name. When the sons moved out of their parental house, they incorporated the Besé name into their new house names. The original Besé house does not differ from the other houses in architectural style, but there are other potential traces of what might have been a local elite. The Besé family had produced several anji and is now the house of Besé Nima, or Nima Anji, one of the most knowledgeable anji of the village and a very influential figure, instrumental in the flowering of anji practice in Bustling Township and beyond. Nima Anji had been educated as a health worker, and his brother was a teacher. Interestingly, one of the points Lévi-Strauss made about the “house” institution is its intermediate position between a kin- and a class-based society and how it “naturalizes” rank differences through the terminology of kinship (1983: 187). It is unclear whether the besé institution is reminiscent of the traditional chiefs of the Kachin, or whether the Besé house name is comparable to what Aziz has called “quasi-housenames.” These are found among families in D’ing-ri who are in the process of moving up in the social hierarchy and are derived from the profession or rank of a household head, such as the above-mentioned house name U-lag, meaning “Master Craftsman” (Aziz 1978: 123). Unlike the stratified Na of Yongning, Premi society in Muli is not very stratified. As mentioned previously, the head lama and many other leading religious-political figures come from the Bar clan. The manorial estate of the Bar clan is situated in southeastern Muli, and no member of this elite was present in Bustling Township, so this stratification presumably did not affect the local ranking of houses. Perhaps ethnographic data collected on other Premi villages in the future can shed light on the institution of village leaders and their houses.

The inhabitants of a Premi house in Bustling Township are usually of two or three agnatically linked generations: an older married couple with their youngest son and his wife and their children. In case of polyandrous marriages, the household might count two or more brothers married to a single wife. Sometimes an unmarried sister of the father or the son might live there as well. Each house has a “housemaster,” or (dzèⁿ) dap’u, usually the oldest man, if he is still capable of doing all of the work that is expected of him, or he might have relinquished the post to his married son. When there is a polyandrous union in the house, the oldest brother has the role of dap’u. He makes all major decisions relating to the house and its inhabitants, and he represents the house in its interactions with the outside world, for example, at clan meetings and cremation ceremonies. Also, some of the ritual practice in the house can be performed only by the dap’u. But the dap’u is not alone in making decisions. At his side is the damu, the female leader of the house, usually his wife or, when she is deceased, his daughter or daughter-in-law. While the dap’u decides on the “male” aspects of the house, often related to relations with the outside, the damu decides on the “female” side, which includes all internal matters of the house, such as disposition of some of the house income, marriage agreements, or whether to send daughters to school. In some cases, after the dap’u dies, the damu, not the dap’u’s son, takes over the role.

When splitting the house, the oldest son is usually the one who moves out first, either when he marries or, if he is already married, when one of his younger brothers marries. People denied that there were rules about this but pointed to lack of living space resulting from a growing family as the reason for splitting the house. Although land was allocated to households in 1982, it was done on a per capita basis. When sons want to leave the parental house to set up their own, they are entitled to their share of the land, but the average of 1.2 mu per person is not enough to sustain a new family. Sons may obtain more if the parental house can do without some land, or they may, for example, take other family members with them. In a few houses, one of the parents had moved into the departing son’s house. When daughters leave the house to get married, they do not receive any land, even though in 1982 their presence entitled the household to a share for them as well. It is unclear as yet how land will be divided now that a generation of men is coming of age who were born after the original allocation took place. Since 1982, there have been no major rounds of readjustments, as is the practice in many areas of rural China.

As in traditional Tibetan societies (Levine 1988: 121), splitting the house is culturally condemned, and it is preferable for brothers to stay together. The recorded data do not make it possible to conclude on the possible role of this ideology and how it interacts with the practice of polyandry. In Walnut Grove, brothers who were married to unrelated wives rarely lived in the same house, and it was often jokingly stated that two wives of the same generation could not live under the same roof. Although polyandry is practiced extensively in the surrounding villages and is generally recommended in the village as an ideal means of keeping the land together and protecting the prosperity of the house, at the time I conducted the fieldwork for this study, there were only three recently established polyandrous households in the village. The average number of people per house was 5.5, and many of the houses had been built within the last ten years. So in spite of a positive assessment of polyandry (many of the women born in the village had been married out into polyandrous marriages in the neighboring villages), polyandry was not practiced very much in Walnut Grove. At least four of the recently split houses were brothers who had moved out of polyandrous marriages, married again monogamously, and established their own houses. One reason for this might be the sudden return of men from the village who had been working in lumber-related jobs until 1999, some of whom had engaged in polyandrous marriages. On their return, they were not able to glide into well-established and, in practice, functioning monogamous marriages and, having brought some capital with them, preferred to take their chances and split the house. The situation was markedly different in Uphill and Downhill, where houses tended to be larger and could more easily accommodate an extended household; the average number of people per household is 7.1. No new polyandrous marriages have been established in recent years, but it is much more socially accepted in Uphill and Downhill that brothers married to different wives live together in the same house. Premi “houses” are then manifestations of what Elisabeth Hsu has called a “hearth-oriented” ideology with a focus on household harmony (1998: 71, 90).

The house is also a corporate unit in the sense that it “owns” property and makes up an economic unit: the land is divided between the different houses, and the inhabitants share the tasks of working it. Many of the tasks are strictly divided according to gender and age. Only men can plow the land, and even when no man is available, women are not permitted to use the plow. Killing animals, hunting, and playing the flute are also exclusively male domains. The women perform most of the tasks at home and feed the domestic animals, while herding is done mostly by the smaller children or the younger unmarried girls. Everyone in the house helps with the harvest. The products of the land as well as any other income generated by members of the house belong to all the inhabitants. The head of the household makes all major decisions regarding the house, concerning, for example, selling and purchasing domestic animals, renovating the house, finding suitable marriage partners for the children, and so on. Chinese authorities register and tax based on “single households” (hu), and since this measurement largely corresponds to the inhabitants of a house, the system does not interfere with Premi social organization.

HOUSES AND RITUAL PRACTICE

The most important contribution of the notion of the “house” is that it makes it possible to link the architectural, social, and symbolic aspects of a single institution and treat them as being closely related. The house as a principle of social organization cannot be seen detached from its role in religious practice. The Premi of Bustling Township have no temples in which to worship their deities or their common clan ancestors, but every house constitutes its own sacred space. It enables the residents to establish and maintain relations with their ancestors and deities as well as among themselves through the practice of rituals.

The house’s centrality as a cornerstone of Premi society is expressed through its prominence in ritual activity in Bustling Township. Ritual in this context is not limited to special sacral activities that are markedly different from everyday actions such as funerals or large communal offerings to the gods. In order to pinpoint some of the dimensions of ritual activity, Catherine Bell looks at how cultures ritualize human activity. She discerns different attributes of “ritual-like” action, such as formalism, traditionalism, disciplined invariance, rule-governance, sacral symbolism, and performance (1997: 138). Several daily, often recurring, activities that take place in the Premi house are highly symbolic acts. They are tightly connected to the way Premi in Bustling Township interact with ancestral souls, deities, and evil spirits. Nevertheless, they are only to a limited extent rule-governed or formal. Regular rituals vary from house to house and are subject to the creativity of the performers in the house.

In daily life as well as on special occasions, the hearth and fireplace with its iron tripod, or hsin-drwè,²² make up the locus of worshipping the “ancestors” (bap’u) and divine beings of the mountains, water, wind, heaven, and earth. Close ancestors are worshipped over large areas of Asia, but the available ethnographies of Tibetan communities mention few such cases. In many areas of Tibet, people worship territorial deities that are recognized as ancestors of local communities and apply kin terminology to them.²³ Hildegard Diemberger mentions rituals (lha-bsangs) in southern Tibet centered on mountain and clan deities. According to her study, among border communities that have retained a certain degree of autonomy from the state, such rituals constitute a vital part of the sacred relation between the territory and the community’s self-definition (1994: 144). Being focused mainly on close agnatic ancestors of the people living in the house, Premi ancestral worship does not have this extended aspect of larger communal self-definition. The persistence of some concepts of ancestral worship combined with the dearth of reported instances of close-ancestor worship in present-day Tibetan communities could be linked to the adoption of Buddhist teachings on rebirth and reincarnation. These conceptual notions do seem incompatible with a belief that ancestral souls exist eternally as immaterial spirits. But then again, it is unclear whether Tibetans ever had a well-developed system of worshipping close ancestors. Matthew Kapstein argues that the relative ease with which Tibetans adopted the concepts of rebirth and reincarnation might point to the fact that rebirth was considered possible under certain conditions even in pre-Buddhist Tibetan beliefs (2000: 43–44). In this context, he refers to an article by Anne-Marie Blondeau on the rebirth of children who died at a very young age (Blondeau 1997).

Some studies have tried to establish a link between kinship organization and ancestor worship. In a comparative study of 114 societies, Dean Sheils finds a clear correlation between descent type and ancestor worship: ancestor worship is absent in most cognatic societies, while a clear majority of unilineally organized societies are supportive of it. Relying on the Human Relations Area Files, Sheils classifies Tibetans as cognatic and without ancestor worship (1975: 434).

Such generalizations of both the category of Tibetans and Tibetan religious practice lack a sound empirical basis. To complicate matters even further, ethnographic evidence clearly points to the existence of worship of close lineage ancestors among at least some Tibetan communities. Eva Dargyay describes how—among the Tibetans of Zanskar—members of the same patriline, or rus-pa, worship a particular god that protects the family and their property. Such a family god is called pha-lha, or “god of our fathers” (1988: 127). Worship of the pha-lha has been integrated locally into Buddhism in different ways: one patriline, after its ancestral shrine was destroyed, started to worship its pha-lha in the local Buddhist temple, while another patriline identified its pha-lha as the six-armed Mahākāla (ibid.: 129–31).

The combination of Buddhist concepts such as reincarnation and rebirth with ancestor worship has also been described among the Shuhiⁿ people of Shuiluo in Muli. According to Swiss ethnobotanist Caroline Weckerle, the Shuhiⁿ people make offerings to the souls of their ancestors, believing that such a soul is immortal and that, after the person has died, it spends some time in a “world of souls” (Welt der Ahnen) before it finally reincarnates. An important component of Shuhiⁿ funerary ritual focuses on driving the souls of the dead out of the house and into this “world of souls” so that they will not cause harm to the surviving family members (1997: 36).

In traditional Chinese society, ancestor worship is the most important aspect of religious practice and is inherently connected to the centrality of the patrilineal family. The family makes offerings to close ancestors in the house as well as in clan temples that the larger community has dedicated to more distant common clan ancestors. Certain ancestral spirits of famous persons or other humans sometimes gain even wider recognition and become veritable deities worshipped throughout a larger region. Ultimately, Chinese ancestor worship and the hierarchical clan system are part of the system legitimizing the rule of the emperor and the state (see, e.g., Paper 1994: 78–80; Faure 2007; Feuchtwang 2001).

But while Chinese ancestor worship is strongly rooted in the existence of a common clan and a focus on the patrilineal family, as exemplified in the importance attached to family names, Premi ancestor worship is largely limited to the house, as expressed in the status of the house name as the most important indicator of a person’s belonging, and through the daily offerings made to ancestors who died in the house rather than to those of the whole patriline.

The Buddhist influence on Premi society can be seen in the zaⁿbala, a Tibetan Buddhist–inspired picture or relief placed in the corner of many houses where Buddhist deities are worshipped. In addition, the whole process of establishing a new house—starting with the ritual of splitting the house, then finding a suitable place to built a new house, constructing it, and finally consecrating the hearth—is steered by highly ritualized practices. With the exception of life-cycle ceremonies, these are the most ritualized practices in Premi society, and many involve religious ceremonies performed by anji or yèma. The head of the house-to-be must also carefully carry out several ritual tasks during the house-building process. Because of the sacral dimension of the house, small mistakes in the process can ruin prospects for prosperity and jeopardize the lives of members of the new house unit.

Before construction can begin, a suitable place must be found. In order to find such a spot, the man who plans to establish a new house collects several stones from different locations on the land the state has given his parental house the right to use, which will be divided when the new house is ready. The stones must be strongly connected to the location and the spirits who reside there. In order to avoid taking stones that people have placed there or that have been deposited recently by natural means like water, the stones must be dug up from at least thirty centimeters underground. They will then be shown to a person presumed to have a good sense of the invisible dimension of the natural world. Such a person need not be the anji or yèma but may also be a locally recognized—Buddhist-inspired—reincarnation or a soma, a medium through whom spirits and divinities communicate with the human world. These matters are not taken lightly, since choosing the wrong location could have grave consequences, as my host family in Walnut Grove experienced. Forty-seven-year-old Galoⁿ explains:

This house is situated on the location of my grandfather’s house. This is quite a steep place though, and when the house had to be renovated, my father decided to build a new house farther down the slope on a place that was more level and closer to our fields. This turned out to be a very bad idea: my father, my mother, and my uncle, one after another, all died of disease at the new house. So I decided to rebuild the house on the location of my grandfather’s house, which had proved to be a good location. Since the moment we moved in, we have not been struck by disaster again.

After the best spot has been selected, an anji or yèma uses traditional Tibetan almanacs to divine the most beneficial dates for performing important parts of the house construction. The first date the ritual specialist must fix is for cutting down the trees; the anji or yèma also participates in picking the right trees. The next date is for the start of digging to make the ground wall and the floor. The porch and the main door are constructed on a separate date. The final two dates are for moving in and for starting to use the iron tripod.

Even at the height of the Maoist period, people chose to risk being criticized for “old thinking” rather than face the dangers of not following the traditional procedures for constructing and consecrating a new house. Campaigns to wipe out “superstition” were more likely to drive these newly vilified practices underground than to produce a heartfelt acceptance of the official ideology. K’enbuzo of the Naxi Bot’a clan of Walnut Grove, who was sixty years old in 2001, still lived in one of the few houses built during the Cultural Revolution:

This house is one of the older ones here. Because we are a poor family, we have not been able to built a new house like many of the other families here [Walnut Grove]. The house was built in 1970, entirely according to the rules. But during the Democratic Reform period and the Cultural Revolution, everything had to be done in secret, otherwise we would be criticized. At that time we also were expected to go to the doctor when we were sick, but we would first secretly ask the anji to find out whether it was evil spirits who caused the disease or whether is was something the doctor might be able to cure.

The person who builds the house—the future head of the house—mobilizes his close kindred. For a few tasks that demand specialized competence, such as woodcarving or smithery, he calls in craftsmen and compensates them with arje or dried pork. A house-building team usually consists of five to eight people, and it will take about a month to build the house. The ritual specialists who together spend a great deal of time reciting and performing rituals also receive gifts such as dried pork.

FIG. 3.3. Part of the gujhi-jhataⁿ, or central pillar, with a small nungk’e, or thread-cross.

In Walnut Grove, all main doors must face east, and the main roof beams must run east to west. East is the location of the river that cut the steep slope on which the village and its surrounding fields are situated. The village of Five Nut Trees is on the opposite slope. Here the roof beams follow the same east-west direction, but the main doors look out toward the west, facing the river. The gujhi-jhataⁿ, the central pillar or post in the house, must be constructed and placed very carefully, so that it follows the direction of growth of the tree it was cut from. Placed in this way, it says that the house is living and growing. During my fieldwork, I heard terrible stories about what happened to houses in which the gujhi-jhataⁿ was placed upside down by mistake: this misalignment causes misfortune and death.²⁴ It is said, furthermore, that the gujhi-jhataⁿ connects the earth with the house and, further, with heaven and the deities.

Such beliefs are not limited to Bustling Township. Claes Corlin writes that in Gyelthang, a two to three days’ walk southwest of Bustling Township, the central pillar of a house, or bekha, is a symbolic representation of the world-tree, “the centre of the universe, and the communication channel between the middle world of men, the upper world of gods and beings, and the subterranean world of the kLu erpent spirits” (1980: 87). In Bustling Township, the central pillar forms one of the major supports for the platform, called drèⁿ in Walnut Grove, and represents the connection of the inhabitants of the house with the earth and heaven. Although the style may differ slightly from village to village, every house in Bustling Township contained such a platform, including the houses of the Naxi and Rek’ua (figure 3.4 illustrates the main room of a house in Walnut Grove). The platform is elevated about fifty centimeters above the earthen floor and covers about a third of a Premi living room or jima. There are basically two styles of platform. The elevated version shown in figure 3.4 is found in the south of Bustling Township and is called drèⁿ. In the northern villages, drèⁿ means “bed”; here platforms are called goli, are only about ten centimeters above the floor, are rectangular in shape, and fill roughly half the main room. The main hearth is also on this platform, which is where Premi spend most of their time in the house; it is where they eat, talk, entertain guests, perform rituals, and—at least for some family members—sleep. Therefore the platform has to be connected to the gujhi-jhataⁿ.

When people are not working in the field or herding their animals, they spend almost all of their time on the platform. Being so important, the platform is also one of the places where social and ritual positions are expressed: in Walnut Grove and Uphill and Downhill, the oldest men sit closest to the fireplace, while women sit on the perimeter because they are considered ritually polluted. Children are farthest away from the fireplace. In some of the villages, the men sit on the right side of the hearth, and the women on the left, with the oldest people or guests closest to the hearth. The most important place is always the place on the left, closest to the hearth. In the middle of the hearth, or hualip’e, is the iron tripod, and in one corner of the hearth is the offering stone, or drwama. Both iron tripod and offering stone are the most important loci of worship and offering in Premi houses. Before each meal or before drinking, people place offerings of food or wine, che-drö (lit., “food for the souls”), on top of the iron tripod, specifically on the three places where its feet join the upper ring. Food and drink offerings are also placed on the offering stone. When an animal has been slaughtered, its head is placed on the offering stone during the meal. While some people told me this was an offering to the ancestors, I was also told on several occasions—note the clear Buddhist influence—that the divine powers of the offering stone would cause the soul of the animal to be reborn in another animal far away, so that it would not be killed again. The offering stone links the house, the ancestors, and the mountain, water, and other deities to the inhabitants of the house.

The offering to the me-drö, or souls, of the ancestors, is strongly related to the house and the patriline. Food is offered to the me-drö of male ancestors who have died in the house, not to the entire patriline. It invites them to join the family in eating and drinking. Anyone in the family can make such offerings, which are part of the ritual reaffirmation of the house as a unit, linking present co-residents to previous residents. As mentioned before, the Kachin term for “house,” htinggaw, has the sense of “the people who worship the same set of household spirits” (Leach 1970: 126), and although the Premi word dzèⁿ does not explicitly carry this meaning, it certainly has this connotation. Ancestor worship and the belief that ancestors, or their souls, influence the world of the living are the most central aspects of religious beliefs and cosmological understanding of the Bustling Township Premi. Besides strengthening the relations of those who share a common residence, ancestor worship also provides a powerful basis for morality in Premi society. The ancestors will protect their living descendants, if these descendants behave in an appropriate way. This of course involves performing all the required ceremonies at the time of death as well as making the regular ritual offerings in the house. But the ancestors are also keeping an eye on the general behavior of their descendants and can punish them if the harmony of the house is disturbed, for example, by a matrimonial dispute.

The offering stone and iron tripod are also involved when members of the house make prayers and offerings to the divine beings: to the mountain deities, or rèdzeng rèda; the water deities, or lwéjabu; Lama Yidaⁿ, the deity of heaven; and several others (see next chapter). The head of the house asks these divinities and the ancestors to protect the residents of the house. In Walnut Grove, this is done every morning in every house by burning some branches of a bush called sanske or dzajhi. It is a very informal form of prayer and can be improvised to fit special needs or the personal style of the one who prays. As one head of a house explained: “Praying can be about many different things: protecting the people of the house, paying respects to the ancestors who died in the house, and so on. The real words are not that important; it’s the meaning or the thoughts that count. You can make it up yourself and say something like ‘Oh, this or that hla [deity], I ask you: please protect the members of this house!’”

In Uphill and Downhill, praying is not done regularly by all families and is more common on special occasions, such as before someone from the house sets out on a journey or the day a pig will be slaughtered. Before the prayer, it is not permitted to pollute the air by smoking or drinking. When the head of the house is away, another male household member conducts the morning prayer, but women are excluded because they are considered ritually polluted.

During the construction of a new house, making the offering stone is one of the important tasks of its future head. He must cut it from a large stone that comes from a ritually cleansed place in the mountains. On the underside of the offering stone is a small cavity where the builder places some silver or gold (depending on the financial situation of the builder) as well as a few grains of maize, highland barley, wheat, and rice. Only these grains can be placed there. This is to assure the future prosperity of the house. People deny that this is an offering to any divinity or ancestors, which implies that the house, in addition to being a medium for worshipping ancestors and divinities, possesses an animistic quality in itself. It can provide its inhabitants with good harvests and material wealth. This aspect is very important during times when maintaining and increasing prosperity seem increasingly unpredictable and when villagers can no longer count on outside support in case of adversities. Even families who are less “religious” and, for example, do not normally invite anji when-someone gets sick will have an offering stone that conforms to the rules.

In the corner of the platform, behind the hearth, is the sarra, a small bench, with an incense burner. Behind it is the set’u, a cupboard or chest, usually the most beautifully carved piece of furniture in the house. Both are dedicated to the Buddhist deities worshipped by the house’s residents as well as to the ancestors. Here, on fixed dates, and especially during the New Year period, the family presents offerings such as fruit or spirits. Pictures of famous lamas, such as the Panchen Lama or Dalai Lama, as well as highly revered heroes of the Communist Party, including Mao Zedong or Zhou Enlai, are often placed in this corner. On the wall behind the cupboard is a painting or a relief with a motif taken from Tibetan Buddhist iconography, most often the “precious jewel,” or norbu rimpoche. This is the zaⁿbala, and it is found in the houses of many Tibetans, Na, Naxi, Premi, and other ethnic groups in this corner of Southwest China. Zambala (or Dzambala) is the wealth-granting or wealth-restoring deity of Tibetan Buddhism. Few zaⁿbala actually depict this deity, and even fewer people understand the origins of their zaⁿbala, but it likely started as an altar for worshipping the deity of wealth (in the hope of bringing prosperity to the house), not unlike the practice of putting gold or silver and grain into the offering stone.

FIG. 3.4. Main room or jima as found in Walnut Grove, with fixed sitting places for men (M) and women (W).

FIG. 3.5. The family burns pine branches as an offering in the morning. The iron tripod with offering stone to the right of it are in the foreground, and the gujhi-jhataⁿ, or central pillar, is visible in the back.

In many houses, the zaⁿbala, the set’u, and the sarra constitute an alternative means of worshipping the same ancestors and deities worshipped by means of the iron tripod and the offering stone, rather than being a shrine to a different divine being. In some houses in Uphill and Downhill with a stronger Buddhist connection—such as having, or having had, a monk in the family—the zaⁿbala forms a more prominent presence in the house and is considered the central place of religious activity. Several of the families who seldom pray and make offerings in the morning regularly burn incense on the sarra. Because of the clear association with Buddhist religion, most of the zaⁿbala were destroyed during the Cultural Revolution, and not all houses have replaced or rebuilt theirs, although many plan to do so. The placement of the set’u, sarra, and zaⁿbala in the corner of the platform testifies to its important position in Premi households: it is close to the offering stone and the hearth with the iron tripod, and the imaginary line from the set’u and sarra to the gujhi-jhataⁿ, or central pillar, creates a sacred alignment that cannot be transgressed. In Walnut Grove, this prohibition is strictly observed, and no one would ever pass behind the hearth to get to another part of the platform, no matter how inconvenient the resulting detour.

FIG. 3.6. Sarra and set’u, the small bench and chest, with incense burners and other paraphernalia used in ancestor worship. The inhabitants of this house had not yet repaired their zaⁿbala, the Buddhist picture or relief, which was destroyed during the Cultural Revolution. Until then, Chairman Mao would have to suffice.

Many of the houses of the northern villages have a room that is used exclusively for worshipping deities and in which monks may stay, called ch’akoⁿ or hlijeⁿ. This room is always the most luxurious place in the house, sumptuously adorned with Buddhist iconography, and is most often found in the houses of families with strong Buddhist connections, such as those with a son at the monastery. The ch’akoⁿ is the place where the son sleeps when visiting his family. In North Village, the biggest house had a very nice ch’akoⁿ especially prepared for the visits of an important lama who lived in the next township. In Walnut Grove, every house with an anji also had a ch’akoⁿ. The head of the house must put two more items in place before the house is finished. On the flat part of the roof there must be a small offering oven, which is used once a year on the ninth day of the first month, when pine branches are burned in all the ovens as an offering to the mountain deities, the rèdzeng rèda. Finally, on top of the roof, an iron trident must be fixed. This is the kadraⁿ, and its function is to protect the houses from the zyè, or wind demons. The master of the house must place the kadraⁿ, but it must be consecrated by an anji or yèma, who touches the kadraⁿ with a rooster, the zyè-ro, or wind-demon rooster. The zyè-ro is viewed as a temporary help lent by the mountain gods. As such, it does not belong to humans and can never be killed; when it dies a natural death, it has to be taken into the mountains and left there so that it is returned to the mountain deity.

When construction is finished, an auspicious day must be chosen for starting to use the iron tripod. First, the head of the household makes an offering of a sheep, to ask for the protection of the different deities. Then, either an anji, yèma, or other person with special powers, such as one of the living deities, recites prayers and, most important, touches the iron tripod to consecrate it. Until it is touched by such a person, it remains just an ordinary piece of iron. The ceremony, which lasts half a day, is called jalasoⁿ. In all the Premi, Rek’ua, and Naxi households in Bustling Township I visited, the iron tripod had been consecrated. In the Naxi and Rek’ua villages, this was done either by an anji or a dtô-mbà, but the custom was said to be Premi in origin. The fire can be lit for the first time only on the date after it has been consecrated, but this is ritually regulated: first the pot is put on the iron tripod, next clean water is poured into the pot, and finally firewood is placed under the pot; only then is the fire lit. Some houses wait until the Premi New Year to use the iron tripod for the first time, because this is an auspicious period in general. It is also the only period when the iron tripod may be cleaned.