Goddess on the Frontier
Religion, Ethnicity, and Gender in Southwest China
Megan Bryson



Religion, Ethnicity, and Gender in Dali

August 14, 2009, marked the beginning of the Torch Festival, and on that day a huge crowd gathered on the hilltop of the City of Virtue’s Source to celebrate the grand opening of the new temple to Baijie Shengfei (Holy Consort of White Purity).1 A red banner hanging across the temple doors read, “Ceremony Celebrating Eryuan County’s Renovation of the Holy Consort of White Purity Temple.” To the west of the temple, celebrants circumambulated a giant torch that would be ignited at sundown. Festooned with colorful streamers, flags, and pom-poms, the torch bore its own messages of celebration: red banners exhorted the crowd to “joyfully observe the ethnic Torch Festival” and proclaimed that the torch was “erected by the Eryuan County People’s Government.” Dance troupes wearing brightly colored costumes performed routines while waiting for the official ceremony to begin.

Everyone on the hilltop that day would have recognized Baijie as an eighth-century widow martyr who committed suicide rather than marry the man who killed her husband. According to Baijie’s legend, her husband ruled a small kingdom that had its capital in the City of Virtue’s Source. His fiery death at the hands of a rival ruler was the origin story for the Torch Festival. The celebrants interpreted Baijie’s title “Holy Consort” as a reference to her relationship with her husband. However, it originally referred to Baijie’s relationship to someone else: the wrathful god Mahākāla. In the earliest writings on Baijie Shengfei, from the twelfth century, she was not a widow martyr at all but a Buddhist dragon maiden. This book tells her story.

The Goddesses Called Baijie

Baijie Shengfei first appeared in Buddhist ritual texts and art of the Dali kingdom (937–1253) as the consort of Mahākāla, a form of the Indian god Śiva “converted” to protect the Buddhist teachings. By the fifteenth century, the name Baijie also referred to the mother of Duan Siping, founder of the Dali kingdom. According to her legend, she conceived the future ruler when a dragon disguised as a piece of wood floated into her foot while she bathed. This Baijie was enshrined in temples near the ancestral home of the Dali kingdom’s founder. The eighth-century widow martyr only came to be called Baijie during the Qing dynasty (1644–1911). She, too, was enshrined in temples and worshipped as a goddess and moral exemplar. Today the Baijie revered as a tutelary village deity (known as benzhu in Dali) contains elements of all three of these forms.

The name Baijie was initially written differently for the different figures that bore it: the Buddhist Baijie Shengfei means “Holy Consort White Sister”; Duan Siping’s mother was Baijie Amei, “Little White Sister”; and the widow martyr was known as Baijie Furen, “Lady Cypress Chastity.” As the three figures became intertwined, their names were also combined, and so today one commonly finds the name of the Buddhist Baijie attached to a figure identified as the widow martyr by villagers. I refer to these three figures collectively as Baijie because of the historical continuities between them and their eventual commingling. Despite the differences in names and forms, two elements of Baijie’s identity remain consistent: she always appears as a feminine figure, and she only appears in the Dali region of southwest China’s Yunnan Province.

Dali is a frontier zone where different cultures meet. Though now part of the People’s Republic of China, Dali has long bordered Southeast Asia, India, and Tibetan regions. From the seventh to thirteenth centuries it was the capital of two independent kingdoms, Nanzhao (649–903) and Dali, and after the Mongol conquest of 1253 it was claimed by the Yuan, Ming, and Qing dynasties centered in Beijing. Dali offers a case study of how people craft local identities out of multiple possibilities and how these local identities transform over time.

Baijie’s transformations from the twelfth century to the present have echoed and shaped Dali’s local identity and how it has been gendered. As Dali changed from an independent political center to a peripheral region in a vast empire, Baijie similarly changed from a dragon maiden and consort of Mahākāla called Baijie Shengfei to a chaste widow martyr known as Baijie Furen. The dynamic aspects of Baijie’s identity mirrored the dynamism of local identity. Baijie, like her worshippers, could not be reduced to either her local or her gendered identity: like them, her identity changed in connection with the broader context; and like them, her local identity did not emerge in a vacuum—it developed through encounters between local and translocal forces.

Translocal forces entered Dali from multiple directions, but Dali elites did not draw equally from each source.2 Baijie’s transformations primarily show how Dali elites represented gendered local identity in relation to Chinese culture and the Chinese state, which even before the Mongol conquest of 1253 constituted the main translocal presence in the region. However, Dali’s location in a frontier zone (also known as Zomia) meant that its elites could draw on a broad repertoire of gendered symbols when positioning themselves in relation to China. Baijie survived because she allowed Dali elites to invoke local identity while maintaining an image of feminine propriety that signified civilization in the semiotic system of Chineseness.

MAP I.1. Current PRC
MAP I.2. Modern Yunnan Province

Deities and Society

Doing justice to Baijie’s identities requires examining closely the complex connections between divine symbols and social roles. Baijie’s transformations are tied to changes in gendered local identities in Dali, but Baijie does not just mirror her worshippers. Scholars of religion in China have long observed the correlations between human society and the spirit world (especially their bureaucratic character), but as Meir Shahar and Robert Weller note, “the Chinese supernatural is neither a mere tool of China’s political system nor a simple reification of its social hierarchy.”3 Deities lack single, fixed identities. They do not have to be either constructive or destructive; they can be (and usually are) both. As polysemic symbols, deities still relate to human society, but not as one-to-one reflections.

Deities’ complex roles are especially important to recognize with respect to gender. Gender is a common focus in studies of goddesses, and for good reason: when sources by and about women are scarce, goddesses can provide an entry point for discussions of femininity and gendered symbolism. As Caroline Walker Bynum observes, divine symbols emerge from the experiences of gendered subjects and are therefore themselves gendered in some way, even if not explicitly.4 However, Bynum also notes that the connection between gender and religious symbols is not straightforward: “Gender-related symbols, in their full complexity, may refer to gender in ways that affirm or reverse it, support or question it; or they may, in their basic meaning, have little at all to do with male and female roles.”5 Even though Baijie’s various identities correspond to some of women’s social roles—consort, mother, and widow—this does not mean that she merely reflects or reinforces those roles in Dali society. Baijie has been a dynamic symbol that people throughout Dali’s history have encountered and used in different ways.


Goddesses are more than mirror images of women’s social roles, and they do more than either oppress or liberate women. Earlier studies of goddesses, fueled by Mary Daly’s criticisms of monotheistic religions centered around masculine deities, presented goddesses as empowering symbols that could establish gender equality.6 However, powerful female deities do not automatically give women economic, political, or social power. They can in fact reinforce patriarchal structures by embodying virtues tied to women’s subordination or by demarcating an unbridgeable gap between goddesses and women. Steven Sangren argues that the Chinese goddesses Guanyin, Birthless Eternal Mother (Wusheng laomu), and the Queen Mother of the West (Xiwangmu) embody an idealized version of femininity that is both maternal and chaste, something that cannot be achieved by women.7 Brigitte Baptandier similarly concludes in her study of the Chinese goddess Lady by the Water (Linshui furen) that women only enjoy equality with men in the symbolic imaginary, that is, as goddesses. While the Lady by the Water defies social convention by postponing marriage to pursue Daoist training, the women who worship her seek fertility to fulfill their wifely obligations.8

Understanding how goddesses relate to women’s social roles requires going beyond the binary of oppression and liberation. Instead, we must examine how gendered symbols operate in context, and how people encounter and manipulate these symbols in different ways. In Dali, Baijie’s widow martyr role signifies different gendered values in the Ming and Qing dynasties than in the PRC. Even during the Ming and Qing dynasties, outside officials presented this form of Baijie as proof that the “barbarians” of Dali could be civilized, while local intellectuals used her to claim that their ancestors had long embraced the feminine virtues that represented civilization. In the PRC she may symbolize women’s oppression under feudalism or inspire women with her refinement and virtue.

In the same vein, there is no single women’s experience of goddesses. Feminist theorists have argued for the need to address diversity in the category “woman” by considering gender alongside other forms of identity such as class, age, and ethnicity. Modern Western conceptions of gender, sex, sexuality, and women do not necessarily fit premodern periods and different cultures.9 Tani Barlow, for instance, challenges the idea that the category “woman” even existed in premodern China, noting that one’s role within the family was a more important identity than the idea of a shared womanhood uniting daughters, mothers-in-law, and maids.10 The use of the male-female (nannü) binary in premodern Dali justifies collectively labeling Baijie’s different roles as feminine, but it would be a mistake to elide the distinctions between consort, mother, and widow. Baijie’s female worshippers (as observed in contemporary Dali) likewise do not belong to an undifferentiated mass of women but align with other forms of identification, such as age, village, or religiosity.

Nor does gender only apply when considering the relationship between goddesses and women: masculine deities are gendered, and men worship goddesses, too. James L. Watson’s study of Mazu worship in Southeast China has shown that women speak of Mazu as a personal deity, while men (at least those from powerful lineages) speak of her as a “symbol of territorial hegemony.”11 The category “man,” like “woman,” is not monolithic, and premodern sources about Baijie come from male elites with access to education and other resources, whether the rulers and court officials of the Dali kingdom or degree-holders of the Ming and Qing. These sources may not represent understandings or uses of gendered symbols in other parts of the Dali population, and they do not explain women’s social status in Dali. Nonetheless, the male elites whose work has survived did not create these gendered symbols anew; they encountered the symbols as part of a larger cultural framework that female elites (and other segments of the population) might have shared. Similarly, their authorial intentions do not dictate how people interpreted their texts. However, in the absence of additional sources on how other segments of the Dali population would have encountered and interpreted these depictions of Baijie, I limit my conclusions to the male elites that produced the materials and claimed to represent the kingdom, clan, or ethnic group.


Baijie does not merely reflect her worshippers’ gendered worlds but is herself part of those worlds. Her devotees include diverse populations of men and women of different ages, time periods, education levels, and village affiliations who encounter and invoke Baijie in different ways. In addition to being a goddess, Baijie is also a local figure that has never been worshipped outside of Dali. Locality is a discourse of power that creates distinctions between what is rooted in a particular place and what is not, namely, the universal, unlocalizing, or utopian.12 But while the local-universal binary has heuristic value, it inevitably masks more complex interactions between localizing and unlocalizing forces.13 Great goddesses develop local or regional identities, or become identified with local or regional deities, just as local goddesses are not cut off from outside influences. For example, Baijie’s cult may be more geographically restricted than Guanyin’s, but Guanyin has different local identities and Baijie’s different forms incorporate translocal elements.

In Dali, the dominant unlocalizing or universalizing discourses have been Buddhism and Chineseness, both of which manifest in Baijie’s different forms, starting with her earliest appearance as a Buddhist dragon maiden. Buddhist narrative encounters with localizing forces from India through East Asia often involve monks converting or controlling chthonic serpent deities. As Richard Cohen and Bernard Faure observe, Buddhists could only claim universality by emplacing themselves locally, thereby intertwining localizing and unlocalizing forces.14 Faure further draws attention to the gendered dimension of these encounters, in which the Buddhist monks signify masculine universality while the serpents signify feminine locality.15 This gendered encounter occurs in Dali through Baijie’s first role as a Buddhist dragon maiden, which binds the masculine, translocal god Mahākāla to the feminine, local goddess Baijie and mutually reinforces the local power of Buddhism and the universal power of local tradition.

Chineseness, the other universalizing discourse in Dali, exists in tension with two kinds of centrifugal forces, those of localization and of “barbarism.” Localization threatens the notion of a cohesive Chinese culture; scholars of religion have long discussed the relationship between universalizing and localizing forces, sometimes framing them as official versus popular or classical versus vernacular. Maurice Freedman and Arthur Wolf famously debated this issue, with Freedman arguing for a common Chinese religion and Wolf positing the irreducibility of local traditions.16 More recently, Prasenjit Duara’s notion of superscription, Paul Katz’s concept of reverberation, and Kenneth Dean’s syncretic field have offered ways to theorize how these two forces interrelate rather than arguing that one force simply dominates the other.17

Narratives, records, and practices associated with deities are ways in which people play out the relationship between universalizing forces of Chineseness and the localizing forces that challenge attempts at standardization. Local deities can disrupt universalizing claims of Chineseness by not fitting ideals of uniformity or standardization. State officials selectively proscribed cults devoted to such disruptive figures, including the Five Emperors in Southeast China and fox spirits in the North.18 Other local deities, such as Mazu and Wenchang, gained official approval when their worship was seen as supporting the state.19

Attempts to standardize deities could meet resistance from their devotees, who in some cases strategically relabeled or recast gods to conform to official prescriptions. Representatives of the state did not form a monolithic group, either, but might decide to allow or suppress local cults depending on local conditions, top-down pressure, or individual proclivities. Baijie’s roles as Duan Siping’s mother and as widow martyr were modes through which people in Dali—especially local elites and outside officials—promoted localizing or universalizing discourses. Officials presented Baijie as a moral exemplar whose veneration fit into universalizing Ming and Qing civilizing projects, but gazetteers show that people worshipped her as an efficacious goddess rooted in the region’s history.

These two poles of moral sagacity and spiritual efficacy fit Dean’s concept of the syncretic field, a multidimensional space created by the tension between two poles, which Dean defines as the universalizing, hierarchical, Confucian sheng sagehood at one end and the territorial, localized ling of spiritual efficacy at the other. Sheng forces try to incorporate ling forces within the hierarchy, but tension remains between these extremes.20 A single deity’s cult will look different depending on one’s position within the syncretic field. Baijie can be simultaneously a paragon of wifely devotion and a goddess who ensures timely rainfall.

The syncretic field is a three-dimensional space created by intersecting planes. In Baijie’s case, the plane demarcated by universalizing and localizing forces intersects with the gendered plane demarcated by masculinity and femininity. Patterns govern how these two planes intersect: Chineseness, like Buddhism, contrasts its universalizing masculinity with localizing femininity. These discourses homologize masculine universality with texts, institutions, and hierarchies while homologizing feminine locality with oral traditions and looser sociopolitical organization. However, these homologies are part of the discourses, and it is important to attend to ways in which forms of social difference are neither analogous nor structurally similar.21 Even though Baijie remains a feminine figure in her different roles, her gendered symbolism can align with masculine universalizing discourses of Buddhist rulership, Confucian virtue, or (in the modern period) ethnicity. Alternatively, she can signify the feminine localizing forces characterized, in the manner of so-called little traditions, by their lack of names.


Dali’s position on the Chinese frontier means that its deities not only engage the tension between centripetal Chineseness and centrifugal locality, they also engage tensions between the universalizing discourses of Chineseness and barbarism. Like the binary of Chineseness and locality, the relationship between Chineseness and barbarism is gendered, but in ways that reflect the different dynamics linking these concepts. Whereas localizing discourses challenge the universality of Chineseness, the notion of barbarism reinforces Chineseness as a discrete and cohesive category by constituting the other against which it defines itself. In fact, barbarism is only the apparent opposite of Chineseness that conceals the latter’s true opposite, namely, localizing forces.

As forms of discourse, concepts of Chineseness and barbarism changed throughout history, covered broad semantic fields, and could be wielded strategically to support different positions. Chineseness, consistently denoted by hua, xia, or a compound thereof, arose during the Zhou dynasty as a label for the people and civilization of the Yellow River plain.22 Different groups received specific labels: the di in the north, yi in the east, man in the south, and rong in the west. Some of these became metonyms for uncultured “barbarians” in general.23 Criteria for being “Chinese” were not fixed: for some, Chineseness depended on lineage, or what in modern discourse might be called race or ethnicity; for others, Chineseness could be acquired through mastery of a particular cultural repertoire, so that barbarians could become Chinese. These criteria were usually combined, but people could strategically foreground one or the other.

The goals of the Chinese state in standardizing temples and deities in both Chinese and barbarian regions were the same: to display state power and suppress potential resistance. However, the rhetoric differed, as Chinese expansion into so-called barbarian territories—including Dali, following the Ming conquest—involved what Stevan Harrell has called “civilizing projects” aimed at bringing Chinese culture to those regions.24 C. Patterson Giersch notes that when the Qing empire expanded into southern Yunnan, the court promoted temples to city gods, the literary god Wenchang, and the martial god Guandi.25 Settlers from Chinese regions also established native place associations and temples to their local deities, who then took on more universalizing roles as Chinese gods. Promoting “Chinese” gods and their temples did not entail eradicating local cults but initiated a divine takeover of the region to complement the military and political encroachments.

As with state policies toward local deities, official attitudes toward barbarian deities and their cults varied considerably depending on the place, the dominant political strategy of the day, and the position of the official in question. State officials might allow such cults to continue as part of a strategy of accommodation, assimilate these cults to those of Chinese deities, or suppress these cults as a show of power. For example, Qing officials in western Hunan developed new explanations of the Heavenly Kings, originally deities worshipped by the Miao people, that turned the gods into historical Han figures.26 Representatives of the Ming and Qing courts attempted to standardize Baijie’s legends and her role as widow martyr to conform to Chinese models. Officials from outside Dali expressed astonishment that the widow martyr Baijie’s example could even survive in the land of “barbarian mists and miasmic rain” (manyan zhangyu).

Discourses of Chineseness and barbarism (like those of locality and universality) were not fixed but could be employed strategically. Most people were probably not concerned with whether deities were “Chinese” or “barbarian” but whether they were efficacious. Han settlers who started to worship the Miao Heavenly Kings in western Hunan likely would not have articulated their practices in terms of Chineseness or barbarism, just as people indigenous to Dali do not seem to have explicitly distinguished between Chinese and Bai deities. The selective use of these kinds of discourses continues to the present, when ethnic terminology is far more common among the educated, urban, and male than among the less educated, rural, and female.

The nineteenth century saw the rise of many new discourses in China, including those that frame this book, namely, gender, ethnicity, and religion. Ethnicity (minzu) superficially replaced the civilization-barbarism binary, which became a binary between the Han and minority nationalities. This new dyad reframed the civilized-barbarian divide in the model of sociohistorical evolution, in which the Han were seen as more advanced than ethnic minorities. Religion (zongjiao) found its opposites in superstition and science, which according to Prasenjit Duara constituted a more black-and-white binary than the previously dominant spectrum that ranged from correct (zheng) to perverse (xie) practices.27 When combined, ethnicity and religion create reified categories of “ethnic religion” that assume the existence of distinctive ethnic practices, beliefs, and deities. In modern Dali, local scholars and government officials present Baijie as a goddess of the Bai ethnicity even though most of her worshippers do not use ethnic language to describe her.

Frontier deities such as Baijie illustrate how Chinese-barbarian and Han-minority binaries are gendered. Those claiming the Chinese position have long used perceptions of female sexuality to mark a given population’s level of civilization. Barbarian women tend to be seen as sexually uninhibited and promiscuous, as in the Tang dynasty claim that when Nanzhao women married, their secret lovers came to see the women off.28 Xiaofei Kang’s study of fox spirits highlights the intersections of gender and Chineseness in showing how female foxes were associated with both dangerous sexuality and barbarism.29 Within this semiotic system Baijie’s role as a widow martyr who refuses to remarry after her husband’s death signifies that the people of Dali are civilized, because women’s sexual propriety is a marker of civilization. In modern China the correlation between sexuality and barbarism continues, though within the discourses of gender and ethnicity. Minority women are also still perceived as sexually uninhibited, especially in the Southwest, as seen in depictions of promiscuity among Mosuo women and in the Yunnan school of painting that specializes in nude images of minority women.30 People in modern Dali represent Baijie in ways that engage the semiotics of both gender and Bai ethnicity but in different ways, depending on their own positions in Dali society.

Baijie’s symbolism encompasses the complexity of deities’ relationships to human society. She can illuminate the oft-studied connection between goddesses and women and show as well that gender is only one dimension of identification, which intersects with others in dynamic ways. Baijie is a local deity that engages the universalizing discourses of Buddhism and Chineseness and the binaries of Chinese-barbarian and Han-minority. Dali, the frontier locality where she appears, is itself a dynamic place where such divisions have been constructed and challenged. Understanding Baijie’s significance requires locating Dali as a frontier region that has become part of China.

Locating Dali

Frontier zones are places where different groups encounter each other and different kinds of boundaries—political, cultural, economic—overlap.31 Located next to Chinese, Tibetan, Indian, and Southeast Asian regions, Dali is precisely such a zone. However, Dali’s proximity to these different areas does not mean that its people adopted ideas, technologies, and objects from its neighbors equally, nor that its elites maintained the same political ties with each neighboring regime. The cultural and political choices that people in Dali have made over the centuries shed light on networks and interactions in areas far from the better-known centers of “great civilizations.” Baijie’s transformations offer a lens for viewing these networks and interactions as Dali’s position changed from an independent political entity to the periphery of empires centered in the East to the border region of a nation-state. Locating Dali involves placing it within both the Zomia region and the Chinese zone into which it was eventually incorporated.


Dali lies within the mountainous Zomia region identified by Willem van Schendel and further theorized by James C. Scott. Zomia straddles the national borders of India, Burma, China, Vietnam, Cambodia, Laos, and Thailand.32 The term, derived from zomi (“highlander” in several Tibeto-Burman languages), offers a new approach to regional divisions that does not conform to nation-state boundaries or postwar area studies disciplines. Indeed, Zomia’s statelessness is a defining characteristic: governmental power cannot remain strong in places where transportation is so inconvenient and taxation so inefficient. Scott challenges the standard narrative that paints Zomia and similarly “inhospitable” areas as refuges for those who could not compete in the civilized realm. Instead, he argues that people actively moved to Zomia to avoid state control, in some cases willingly giving up markers of civilization such as literacy and fixed agriculture.

Yunnan, with its diverse population, mountainous terrain, and distance from the political centers of China, Tibet, India, Burma, and Vietnam, is located within Zomia. Both van Schendel and Scott describe the Nanzhao kingdom as an example of state formation within Zomia, and the Dali region falls well inside the boundary between Zomia and the direct reach of Chinese states. Dali also fits conceptually into van Schendel’s notion of Zomia as a region that suffers academic neglect because it lies outside traditional area studies disciplines that shape knowledge production.33

Zomia offers a geographical metaphor that has the benefit of not making Dali merely an appendage of one of its central neighbors. At the same time, simply locating Dali in Zomia does not explain Dali’s changes throughout the centuries. Dali displays gradations within Zomia: at an elevation of 6,585 feet it is a high-altitude region, but it also centers around the Er Lake (Erhai) plain. Elites in Dali, unlike those in other parts of Yunnan, have been using Sinitic script since at least the Nanzhao kingdom, and Ming rulers imposed direct central rule in Dali rather than continuing the “native chieftain” (tusi) system they used elsewhere in the southwest. Dali has been more accessible to Chinese states than other parts of Zomia, even those farther east. The work of Richard von Glahn and John Herman on the Chinese colonization of Sichuan and Guizhou, respectively, shows how proximity alone does not determine a state’s ability to impose direct central rule.34

Despite closer ties between Dali elites and Chinese culture, Dali still falls into the Zomia realm. In the first place, it served as a site of refuge and rebellion for people from Chinese territory, including Buddhists fleeing the Huichang persecution of 845, the Ming Yongli emperor (r. 1646–1662) in the wake of the Qing conquest, and Du Wenxiu’s Islamic sultanate that resulted from the Panthay Rebellion in the 1850s.35 Moreover, it has been a zone of interaction and exchange between different populations, including representatives of large neighboring states and smaller groups of the surrounding mountains. And finally, Dali’s population has rarely been classified as Chinese or Han and has in fact become a bone of contention in matters of ethnicity and language. H. R. Davies wrote in 1908, “Min-chia is undoubtedly the most puzzling language of Yün-nan to classify” because it contains elements of all four language families in the region.36

“Min-chia” (Minjia, “civilian household”) became a term for Dali’s indigenous population following the Ming conquest, when the households of Ming soldiers who remained in the region were known as junjia, “military households.” From the Ming dynasty through the Republican period, “Minjia” was used along with the ethnonym “Bai,” which first appears in a record from the Yuan dynasty. In the ethnic classification project of the 1950s, “Bai” became the official name of the minority nationality centered in Dali because it was seen as a more indigenous term. Some modern scholars apply the Bai ethnic label to Dali’s population going back to the Nanzhao and Dali kingdoms, but sources from Nanzhao and Dali do not use such ethnic autonyms. The only labels for Dali’s population at this time come from Chinese sources, which consistently refer to people from Dali as man barbarians. Some modern scholars also treat these Chinese labels as autonyms for Dali’s population despite their clear bias. Another problematic view of ethnicity in Dali is the misconception that the Nanzhao rulers were Thai, which persists despite having been convincingly refuted in the 1960s.37

Dali’s perceived eclecticism based on its location has meant that the Bai can be whoever a scholar wants them to be, whether Thai, Tibetan, or Chinese. Dali is a litmus test of sorts that reveals more about scholars’ preconceptions than about the region’s inhabitants themselves. For example, two students of Malinowski, Frances L. K. Hsu and C. P. Fitzgerald, both conducted ethnographic research in Dali during the 1930s, but Hsu treated the Bai inhabitants of West Town as prototypically Chinese, while Fitzgerald emphasized their distinctive ethnic practices.38 More recently, David Y. H. Wu has used the Bai to illustrate how the ethnic classification system of the PRC preserved ethnic differences even as cultural differences faded away.39 However, Beth Notar has criticized this on the grounds that Wu ignored important emic claims about cultural difference and conducted his research in eastern Yunnan, far from the Bai population center in Dali.40 These examples support Wang Mingke’s argument that Chinese identity is not demarcated on the basis of objective criteria such as shared language, religion, or dress but is instead expanded or contracted according to shared historical memory.41 For this reason I am less interested in determining whether Dali elite religion came from China, India, Southeast Asia, or Tibet and more in examining how Dali elites represented their religious traditions as part of their overall self-representation.

The difficulty of categorizing Dali’s population and language attests to the fluidity of self-representation that characterizes Zomia. Scott describes this fluidity as strategic, as it allows people to invoke different identities depending on the context, such that someone could claim Bai identity in interactions with other Dali locals or Chinese/Han identity when traveling outside of Dali.42 People can use ethnic discourse strategically to galvanize communities or to claim to represent a larger group. As Rogers Brubaker and other scholars of ethnicity have observed, ethnic discourse sometimes masks individual, class, or kinship interests.43 The vast majority of premodern sources from Dali come from the elite stratum, which includes the rulers and high officials in the Dali kingdom and degree-holders of the Ming and Qing. Their depictions of Dali identity must be understood in connection to this high status rather than as universal representations.

In sum, Baijie is a product of Zomia that embodies its hybridity and fluidity, as well as its tension with the larger state against which it is defined. Zomia’s mountainous statelessness is only meaningful in contrast to the polities in the plains. While transnational populations engage with more than one state, people in remote areas farther from national borders might deal mainly with one only. In the case of Dali’s elites, this larger state was Chinese.


An official Chinese state presence in what is now Yunnan dates back to at least the Han dynasty, when in 109 BCE Emperor Wu founded Yizhou Commandery in modern-day Kunming and then in 69 CE Emperor Ming founded Yongchang Commandery on the modern-day Burma-China border.44 Han power in Yunnan was limited and depended on cooperation with local authorities, who sometimes rebelled against the state. This pattern continued through subsequent Chinese dynasties until the Tang, when as Tibetan power grew the Dali region became more important. In fact, the competition between Tang and Tibet allowed the Nanzhao kingdom to expand and strengthen in the eighth century.

Nanzhao clearly engaged with several surrounding states in addition to the Tang. Its rulers established political and military alliances with Tibet, conquered the Pyu kingdom in what is now Burma, and competed with Tang troops over the Annam region in modern-day Vietnam. However, it appears that Nanzhao elites adopted more from the Tang than from these other neighbors, including Sinitic script, political structure, and technologies. Nanzhao used the kidnapped Tang official Zheng Hui as a royal tutor and in 829 raided Chengdu for skilled laborers. The ruling class’s focus on China remained consistent into the Dali kingdom. Though the Dali kingdom had less political and military contact with the Song state than did the Nanzhao kingdom with Tang China, records from the Dali kingdom and the Song dynasty attest to continued interactions in the form of trade.

The Mongol conquest of 1253 brought Dali under the direct control of a larger empire for the first time. Though Dali remained far from the capitals in Beijing and Nanjing and served as a haven for refugees and rebels, the Yuan, Ming, and Qing empires could still impose central rule on the region, if not on other parts of Yunnan. Du Wenxiu’s sultanate managed to control Dali from 1856 to 1872 because the Qing army was occupied with the Taiping Rebellion in the East, but after the Qing army turned its attention to Yunnan, it quickly conquered Du’s forces and massacred Dali’s Muslim population. In the twentieth century Yunnan’s borders became fixed, and Dali became part of the Chinese nation-state.

The history of Dali’s elites—the only history available—strongly suggests closer ties between Dali and the Chinese state than between Dali and other regions, such as Tibet, Burma, India, Laos, or Vietnam. Almost all written records from Dali are in Sinitic script, and the others are Sanskrit materials that also circulated in Chinese territory. This does not make Dali “Chinese,” but it shows that Dali elites engaged primarily with Chinese states and their representatives. Historical materials about Baijie’s different forms also come from the elite stratum of Dali society and show how Dali elites developed a gendered local self-representation in relation to Chinese culture and the Chinese state. It is only in the contemporary period that perspectives from women, rural areas, and less educated populations become accessible. To fully explore Baijie’s transformations I adopt a multidisciplinary methodology centered around semiotics and history that also incorporates historical-textual studies, art history, and ethnography.

Methodology and Structure

Semiotics (or semiology), the study of signs, examines how language creates meanings by focusing on the relationship between the signifier (e.g., the term “apple”) and signified (the actual apple). It is the second-order system of symbolism and mythology, whereby the sign “apple” could stand for temptation within a semiotic structure, that helps to make sense of deities such as Baijie. Ferdinand de Saussure’s structuralist approach—particularly his attention to paradigmatic and syntagmatic relations among symbols—offers a way to locate Baijie in relation to other deities, which is particularly important for the Dali kingdom, given the lack of contextual materials.45 Though this book focuses on Baijie, I remain mindful of Bernard Faure’s observation that gods are not distinct persons but “nodes in constantly changing networks.”46 Baijie’s significance stems from her position in divine constellations that look different depending on the viewer’s perspective and the historical context. Making sense of Baijie’s transformations requires attention to both the semiotic and historical aspects of her identities, an approach that Christian Wedemeyer adopted in his study of Indian tantric Buddhism.47 For example, the historical context of Ming Dali can explain why Duan Siping’s mother became an important figure but not why she came to be called Baijie; for that, we need to pay attention to how the Buddhist Baijie and Duan Siping’s mother are symbolically linked. Roland Barthes’s mythological analysis of the Paris-Match issue with a black soldier on the cover offers a model for this approach. In Barthes’s reading, on the mythological level this cover signifies “that France is a great Empire, that all her sons, without any colour discrimination, faithfully serve under her flag, and that there is no better answer to the detractors of an alleged colonialism than the zeal shown by this Negro in serving his so-called oppressors.”48 This reading engages a historically contingent semiotic system in a way that can similarly illuminate Baijie’s different forms.

This book traces Baijie’s different forms chronologically, starting in the Nanzhao kingdom and ending in the present. The first chapter examines religion and collective representation in the Nanzhao and Dali kingdoms to contextualize Baijie’s emergence in the Dali kingdom. The remaining four chapters cover Baijie’s four identities as Buddhist goddess, Duan Siping’s mother, widow martyr, and village deity. I begin each chapter with an overview of the historical changes of the period in question and then look at how Baijie’s new identity arises in relation to these changes.

The nature of available sources shapes my methodology in each chapter. Information about the Nanzhao kingdom comes primarily from Tang records and secondarily from a handful of artistic and textual materials from the Nanzhao court. Materials on the Dali kingdom are the inverse of those on the Nanzhao: Buddhist texts and art from the Dali court are relatively abundant, while Song records about Dali are few. Despite this asymmetry, sources on Nanzhao and Dali present a cohesive image of religion in the region during these periods, which provides a solid foundation for considering the specific sources about Baijie in Chapter 2.

The second chapter, which covers Baijie’s Buddhist identity in the Dali kingdom, relies on three Dali-era Buddhist ritual texts as well as the masterpiece of Dali art, the Roll of Buddhist Images (Fanxiang juan). Buddhist texts from the Dali kingdom were discovered more recently and have only been made widely available in the past few years.49 While most of these texts are Chinese translations or creations dating to the Tang-Song period, six have been found only in Dali. Of these six, three mention Baijie Shengfei or her double, Fude Longnü (Dragon Maiden of Good Fortune). The Dharma Assembly Ritual of Unrestricted Light and Food (Wuzhe dengshi fahui yi) and Bodhimaa Ritual of Unrestricted, Widespread Offerings (Guangshi wuzhe daochang yi) include Baijie Shengfei or Fude Longnü in lists of deities invited to the ritual arena. The Bodhimaa Ritual of the God Mahākāla (Dahei tianshen daochang yi) devotes a section to Baijie Shengfei, who is identified as the consort of Mahākāla, the ritual text’s central figure. These texts complement existing artistic sources from this period and allow us to reconstruct a more complete image of Buddhism in the Dali kingdom.

Baijie Shengfei’s gendered symbolism manifests itself in her relationship with Mahākāla, her identification with other goddesses, and her own textual and visual representations. Though she is paired with Mahākāla in materials with significant esoteric or tantric Buddhist content, their bond is never depicted in overtly sexual terms. Her appearance marries Indian and Southeast Asian nāgī (female serpent) iconography with the long robes of an elegant Song lady, and Dali kingdom texts constellate her with figures such as Śrī Lakmī, Hārītī, the nāgī in the Lotus Sūtra who attains buddhahood, and the Chinese moon goddess Chang’e. I read her gendered characteristics as exemplary of how Dali rulers positioned themselves between India and China. They claimed Indian origins for their Buddhist tradition but rejected sexually transgressive esoteric material.

Chapter 3 moves into the Yuan and Ming periods and shifts focus to claims of divine lineage. It centers on the legend of the Dali kingdom founder Duan Siping’s birth, which begins with Baijie Amei’s own miraculous birth from a plum. One day, while bathing, she is impregnated by a dragon disguised as a piece of wood and gives birth to the future founder of the kingdom and his twin. I examine why this legend developed in the wake of the Mongol and Ming conquests and how it replicates another legend from Yunnan as well as a legend from Burma. Finally, I consider the deification of Duan Siping’s mother and why she shares a name with the Buddhist Baijie.

The legend of Duan Siping’s birth appears in epitaphs, temple records, unofficial histories, and gazetteers. I also look at the legend of Baijie Amei in relation to similar legends from Burma, Yunnan, and Chinese antiquity dating back to the Han dynasty; these are recorded in official dynastic histories as well as local and regional histories. I argue that Baijie Amei’s own birth story violated Chinese historiographical standards that allowed male rulers, but not their mothers, to have miraculous births. Moreover, the sources for Baijie Amei’s legend circulated mainly among a branch of the Yang family in the Dali plain that actively promoted Bai language and group identity. Baijie Amei was a conduit that linked them to the illustrious Duan lineage of Dali’s independent past. In the chapter’s final section I examine sources for the continued worship of the Buddhist Baijie Shengfei. These consist of temple records contained in local gazetteers, most of which come from the Jianchuan region northwest of the Dali plain.

The fourth chapter examines the legend of the widow martyr Baijie Furen in the Ming and Qing. I look at how this form of Baijie fits into the chastity cult that became prominent in late imperial China and how her chaste image was used to construct and contest ethnicity in Dali. Baijie Furen embodied feminine virtues that were closely tied to discourses of civilization and ethnicity but which people could invoke in different ways depending on their position. I also show why this widow martyr acquired the name Baijie in the nineteenth century after being known previously as Cishan, “Kindness”: the Buddhist Baijie’s name and temples outlived the rest of her identity, so people in the late Qing began reading “White Sister” as “Cypress Chastity.”

The popularity of the widow martyr legend means that more writings about this Baijie have survived. Alongside unofficial histories, temple records, and gazetteer entries, we also find in gazetteers several poems commemorating the widow martyr Baijie (or Cishan). Most of these poems were written on the occasion of the Torch Festival, which had become connected to the legend of Baijie Furen by the Ming. Though male elites composed these poems and other writings on Baijie Furen, the legend of the widow martyr Baijie reached the general population through the celebration of the Torch Festival. This mode of popular transmission allows me to consider Baijie’s role as a moral exemplar for women in Dali, not just her representation by elite men.

Chapter 5 constitutes a methodological break from the previous chapters. Instead of relying primarily on textual (and some visual) sources, my study of Baijie’s contemporary worship comes from field research I conducted in Dali from 2006 to 2009, mainly in 2007–2008. I show how modern conceptions of religion and ethnicity shape Baijie’s current identity as a tutelary village deity, or benzhu, and how representations of Baijie differ along lines of gender, class, and age. This chapter begins by laying the foundation for understanding contemporary Baijie worship. I examine how modern discourses of religion and ethnicity took form in the Republican era and continued to develop and change in the first three decades of the PRC, with the classification of “nationalities” and the Cultural Revolution. The rest of the chapter is devoted to contemporary Baijie worship.

Gender still marks a population’s degree of civilization, though now this is framed as masculine modernity in contrast to feminine tradition. As a symbol of the Bai minzu, Baijie combines the allure of minority female sexuality with the widow martyr’s sexual propriety. Bai scholars, officials, and tourism-industry workers can use her as an attractive Bai woman to appeal to stereotypes about minority women while proclaiming her chastity. People in Dali who work outside the realm of minzu discourse generally foreground Baijie’s gendered virtue and do not invoke ethnicity in discussing her legends and efficacy.

Baijie is the only local deity in Dali’s history that has survived for so long and whose transformations provide so valuable a lens for understanding Dali’s changes over a millennium. From her emergence in Buddhist texts of the Dali kingdom to her recent appearances on websites, Baijie has represented both Dali’s distinctive religious traditions and its engagement with translocal, transregional, and transnational forces. Following Baijie’s transformations over time shows how people in Dali have used gendered religious symbols to represent local and regional identities within a large range of possibilities.


1. The City of Virtue’s Source (Deyuancheng), located in Dali Prefecture’s Eryuan County, is the site of the widow martyr Baijie’s legendary self-sacrifice.

2. I use the term “elite” as a broad category for those at the top of the social hierarchy who sponsored and created the sources I use. The identities of these elites change over time: in the Nanzhao and Dali kingdoms they are the ruling families and top officials (including Buddhist monks) of the court; after the Ming conquest they include local and nonlocal state officials, local degree-holders, and travelers; and for the contemporary period they are the scholars who produce works on Dali (specifically, Bai) culture and history. It is in the contemporary period that I can finally explore how non-elites represent Baijie in relation to gender and ethnicity.

3. Shahar and Weller, “Introduction,” 2.

4. Bynum, “Introduction,” 2.

5. Ibid.

6. Daly, Beyond God the Father.

7. Sangren, “Female Gender in Chinese Religious Symbols.”

8. Baptandier, The Lady of Linshui, 261–262.

9. Foucault’s argument that sexuality as a discrete part of human identity is a modern concept has informed the work of scholars such as Judith Butler and Thomas Laqueur, who argue against the idea that gender is to culture as sex is to nature. Instead, “scientific” ideas of sex, sexuality, and the body are expressed in language, which necessarily reflects the power structures of the culture that produces it.

10. Barlow, “Theorizing Woman.”

11. Watson, “Standardizing the Gods,” 297–298.

12. J. Smith, Map Is Not Territory, 101. As Talal Asad observes, this is misleading given that all religious phenomena appear in specific historical contexts. Asad, Genealogies of Religion, 7–8.

13. See Padma, Vicissitudes of the Goddess, on how in India the great goddesses Kālī and Durga retained some village characteristics even after brahmans incorporated them into universalizing discourses of śakti and prakti’s feminine power.

14. Cohen, “Naga, Yaksini, Buddha,” 374–380; Faure, “Space and Place in Chinese Religious Traditions,” 339.

15. Faure, The Power of Denial, 316.

16. Freedman, “On the Sociological Study of Chinese Religion,” 20; Wolf, “Introduction,” 17–18.

17. Duara, “Superscribing Symbols,” 779–780, 791; Katz, Demon Hordes and Burning Boats, 114–115; Dean, Lord of the Three in One, 58–60.

18. Kang, The Cult of the Fox, 159; Szonyi, “Making Claims about Standardization and Orthopraxy in Late Imperial China,” 49. Both Kang and Szonyi note that not all officials actively proscribed such cults, and those who tried were not always successful.

19. Watson, “Standardizing the Gods,” 276–277; Kleeman, A God’s Own Tale, 49–50.

20. Dean, Lord of the Three in One, 58–60.

21. Makley, The Violence of Liberation, 10; Butler, Bodies That Matter, 18.

22. Chen, “From Exclusive Xia to Inclusive Zhu-Xia,” 196–197.

23. This language of barbarism comes from Roman terminology, which does not necessarily apply to the Chinese world. Scholars such as Nicola di Cosmo and Lydia Liu have challenged translating terms such as yi and man as “barbarian” because (for di Cosmo) they elide the distinctions between those Chinese terms and (for Liu) they reinforce colonial power relations. See di Cosmo, Ancient China and Its Enemies, 100–104; Liu, The Clash of Empires, 31–39. While I attend to the specific terms used for Dali’s population, I still find that the larger contrast between Chinese and yi, man, di, or rong aligns closely enough to the civilized-barbarian binary to justify using the concept of “barbarian” to discuss Chinese views of uncivilized others. Moreover, the translation of these Chinese terms as “barbarian” may have gained currency in Western scholarship as the result of British officials’ strategies to impose unequal treaties on Qing China, as Liu argues, but their use within China also occurred within unequal power structures.

24. Harrell, “Introduction,” 4–7.

25. Giersch, Asian Borderlands, 146–149.

26. Sutton, “Myth Making on an Ethnic Frontier,” 458.

27. Duara, “Knowledge and Power in the Discourse of Modernity,” 76.

28. Man shu, 72.

29. Kang, The Cult of the Fox, 27–29. Male foxes are also associated with barbarism in ways that reinforce stereotypes of barbarian masculinity.

30. Harrell, “Introduction,” 11.

31. Elton, Frontiers of the Roman Empire, 4.

32. van Schendel, “Geographies of Knowing, Geographies of Ignorance”; Scott, The Art of Not Being Governed.

33. Ibid., 656.

34. von Glahn, The Country of Streams and Grottoes; Herman, Amid the Clouds and Mist.

35. For the effects of the Huichang persecution, see Sen, “Astronomical Tomb Paintings from Xuanhua,” 48; for the Yongli Emperor, see Struve, “The Southern Ming, 1644–1662,” 679–710; for the Panthay Rebellion, see Atwill, The Chinese Sultanate.

36. Davies, Yün-nan, 343–344.

37. See Blackmore, “The Ethnological Problems Connected with Nanchao”; and Backus, The Nan-chao Kingdom and T’ang China’s Southwestern Frontier. James C. Scott still presents the Nanzhao (as Nan Chao) population as Thai (Tai), though he notes that the Tai-ness of Nanzhao has been contested since the publication of Backus’s book. Scott, The Art of Not Being Governed, 141 and 368n32.

38. Hsu, Under the Ancestors’ Shadow; Fitzgerald, The Tower of Five Glories. Liang Yongjia offers a compelling defense of Hsu’s depiction of Xizhou society. Liang observes that the merchants of Xizhou probably did represent themselves as Chinese, largely because the distinctions between Bai and Han were not salient for them and the minzu system was not yet in place. See Liang Yongjia, “The ‘Ethnic Error’ in Under the Ancestors’ Shadow and Dali Society in the Period of the Nationalist Government,” 79–81.

39. Wu, “Culture Change and Ethnic Identity among Minorities in China.”

40. Notar, “Wild Histories,” 63.

41. Wang Mingke, Huaxia bianyuan, 410. Wang additionally argues that people invoke ethnic categories as part of competition for resources, expanding and contracting ethnic boundaries as necessary.

42. Scott, The Art of Not Being Governed, 241.

43. Brubaker, Ethnicity without Groups.

44. Hou Han shu 86:2846, 2849.

45. Saussure, Course in General Linguistics, 122–127. Paradigmatic associations refer to symbols that perform the same role, while syntagmatic associations refer to symbols that appear next to each other. Bernard Faure invokes these categories in his study of Bodhidharma as textual and religious paradigm. See Faure, “Bodhidharma as Textual and Religious Paradigm,” 191–195. As Jacques Derrida argues, Charles S. Peirce’s semiotics allows for the “endless play of signs” because it does not posit a binary relationship between the signifier and signified as found in Saussure. This poststructuralist approach may ultimately be more compelling when discussing semiotics in general, but Saussure’s theories have heuristic value in analyzing Baijie’s significance over time. See Peirce, Collected Papers of Charles Sanders Peirce, 2:227–242; Derrida, Of Grammatology, 47; Liu, The Clash of Empires, 7–11.

46. Faure, The Fluid Pantheon, 14.

47. Wedemeyer, Making Sense of Tantric Buddhism, 3–13.

48. Barthes, Mythologies, 116.

49. Manuscript reproductions of Dali’s Buddhist texts appear in the five-volume set Dali congshu. Dazangjing pian (hereafter abbreviated DZJP), which was published in 2008. I discuss this further in Chapter 1.