The Goddess of the Flaming Mouth Between India and Tibet

Abstract

This article examines the evolution and potential cross-cultural adaptations of the “Goddess of the Flaming Mouth”, Jvālāmukhī (Skt.) or Kha ‘bar ma (Tib.), in Indic and Tibetan traditions. A minor figure in medieval Hindu Tantras, Jvālāmukhī is today best known through her tangible manifestation as natural flames in a West Himalayan temple complex in the valley of Kangra, Himachal Pradesh, India. The gap between her sparse portrayal in Tantric texts and her enduring presence at this local “seat of power” (śakti pīṭha) raises questions regarding her historical development and sectarian affiliations. To address these questions, we examine mentions of Jvālāmukhī’s Tibetan counterpart, Kha ‘bar ma, across a wide range of textual sources: canonical Buddhist texts, original Tibetan works of the Bön and Buddhist traditions, and texts on sacred geography. Regarded as a queen of ghost spirits (pretas) and field protector (kṣetrapāla) in Buddhist sources, her portrayal in Bön texts contain archaic motifs that hint at autochthonous and/or non-Buddhist origins. The assessment of Indic material in conjunction with Tibetan texts point to possible transformations of the goddess across these culturally proximate Himalayan settings. In presenting and contextualizing these transitions, this article contributes critical data to ongoing efforts to map the development, adaptation, and localization of Tantric deities along the Indo-Tibetan interface.

1. Introduction

This article examines the evolution and potential cross-cultural adaptations of the “Goddess of the Flaming Mouth”, Jvālāmukhī (Skt.) or Kha ‘bar ma (Tib.), across Indic and Tibetan traditions. A minor figure in the early Hindu Tantras of the first millennium CE, the deity is today best known for her fiery manifestation at a temple site in the Himalayan foothills of the Kangra Valley (Himachal Pradesh, India). The gap between the minor entity encountered in Indic texts and her prominence in a major pilgrimage site begs explanation. To explore this seeming transformation, we turn to the substantial corpus of Tibetan literature, both Buddhist and Bön, in which a goddess bearing the same name appears in diverse contexts. This textual convergence is strengthened by historical records attesting to the presence of Tibetan spiritual masters at the Kangra temple, especially significant in the 13th century (L. Shastri 2009). Together, these materials suggest a plausible diachronic connection between the Tibetan Kha ‘bar ma—a wrathful protective deity associated with fire and transformation—and her Indic counterpart, the Tibetan sources’ frequent portrayal of the deity as a preta queen and fierce guardian echoing themes encountered in Indic depictions.

Building on Alexander Zorin’s recent analysis (Zorin 2022) of a 13–14th century Tibetan manuscript from Khara-Khoto concerning the propitiation of Kha ’bar ma, we identify thematic and symbolic intersections between the Hindu and Tibetan figures. The existence of two divinities with identical names and similar characteristics along the Indo-Tibetan frontier is suggestive of religious and cultural exchanges, and consistent with numerous indications of potentially deep historical and cultural ties between Tibetans and their close neighbors, the “hill-men” (Pahāḍīs) of the Himalayas. These include an abundance of local gods, often with complex familial relationships, veneration of serpent deities, a tendency toward theocracy, divine embodiment in children, the prominent role of oracles, ransom rituals such as the rope-sliding ceremony, and even polyandry (Emerson n.d.; Thakur 1997).

Himalayan Studies scholars and anthropologists have, by nature of their intensive engagement with the terrain, been especially active in exploring and contextualizing such similarities.1 While this study shares these anthropological and area studies sensitivities, its primary disciplinary alignment is with the history of religions. It therefore departs from the latter fields in two critical ways. First, rather than relying heavily on ethnographic data, it privileges the philological and historical analysis of primary and secondary sources. Second, mindful of the pitfalls inherent in focusing on a single deity’s development in multiple textual traditions and diverse geographic spaces, it adopts a middle ground approach between the expansive definition of “religion” as an umbrella term for multiple syncretic traditions and its reduction to narrowly defined sectarian cults.

This approach is especially warranted when examining early Hindu–Buddhist interactions, where debates persist between models of mutual exclusivity and those positing a gradual divergence from shared ritual and cosmological frameworks. Arguing against the notion of a generalized cultural substratum, Alexis Sanderson and his students have advanced detailed case studies that evince inter-sectarian borrowings through meticulous philological analyses and the identification of precise cultic lineages (Sanderson 1990; Törzsök 1999; Kiss 2021). James Mallinson’s study of “Kālavañcana in the Konkan” (Mallinson 2019), for instance, demonstrates the assimilation of wrathful Vajrayāna deities into Kaula Śaivism and their subsequent appropriation by Nāth yogīs—an ascetic lineage that also features in this article. At the same time, we maintain that when employed judiciously, conjectural models may be useful for explaining long-term transformations. Studies such as Joel Bordeaux’s (2024) on the goddess Tārā, for example, illustrate how textual and iconographic shifts can chart plausible trajectories of cultic evolution across traditions.

Our study seeks to strike a balance between these two approaches, assessing detailed, lineage-specific data against broader historical contextualization. In doing so, it provides new channels with which to think about the historical transformation of Jvālāmukhī/Kha ‘bar ma, foregrounding Tibetan sources that have often been overlooked in studies of Indo-Himalayan goddess worship and their possible relation to developments in the rise and localization of Indic goddess cults.

The article is structured into two parts. The first part situates Jvālāmukhī within Indic textual traditions and contextualizes her presence in the history of the Kangra temple, a site associated with the ancient sacred geography of Jālandhara. It traces the development of the goddess from a marginal Tantric figure to a regionally prominent deity whose transgressive heritage is preserved with particular clarity in a locally authored Śākta text and the presence of Nāth yogīs connected with the site. The second part turns to the Tibetan reception of Kha ‘bar ma on the basis of Buddhist and Bön sources: canonical texts, original Tibetan works, and texts relating to sacred geography. It traces her transformation from a malevolent preta queen to a fierce guardian within Vajrayāna frameworks, highlighting continuity and innovation in the modes of her representation. The conclusion reflects on these findings to propose key trajectories in the goddess’s evolution, including a potential relation to regional fire cults and congruence with attested patterns of religious exchange between autochthonous, Indic, and Tibetan traditions in Himalayan sacred sites.

2. Indic Sources on Jvālāmukhī

This section explores the gap between the portrayal of Jvālāmukhī in Indic texts and her manifestation as the presiding deity of a central temple in the Himalayan valley of Kangra. While Śaiva and Śākta Tantras depict her as one of numerous minor deities (yoginīs, joginīs), the site of her temple has been held sacred for centuries, attracting spiritual masters and pilgrims from within and beyond the subcontinent. The investigation thus leads from medieval Tantric texts to the history of the temple and its custodians in the modern era—primarily under colonial rule—and is supplemented by recent ethnographic data pertaining to the site.

We begin with a review of Jvālāmukhī in medieval Hindu Tantras and her relation to the subcontinental network of “seats of [divine female] power” (śakti pīṭhas). Although a minor deity in the textual tradition, within the Jālandhara Pīṭha itself, the goddess and her temple are shown to have held a prominent—if not paramount—position in relation to the numerous other deities said to inhabit the region.2 The grounding of the goddess of Tantric texts in the Himalayan foothills is contextualized by noting broader patterns of localization during the second millennium CE, such as the integration of the Mahāvidyā tradition in the Śākta center in Kāmākhyā (Assam). Her adaptation to the social and ritual settings of West Himalayan societies is then further explained through specific examples from the hills. The second part of this section explores the history of the Jvālāmukhī temple and its attendants. It delves into the origin myths found in local publications and colonial sources to highlight the divergent interests of the multiple stakeholders involved with the site, from local kings to priestly attendants and different types of ascetics (Gosains, Nāths) linked with its shrines. While the goddess features prominently in all of these, her Tantric heritage is most clearly borne out in a nineteenth-century pilgrimage guide to the Jālandhara Pīṭha with a distinctly Śākta orientation that awards Jvālāmukhī a conspicuous position in the divine hierarchy.

2.1. Jvālāmukhī in Medieval Tantras and Regional Adaptations

The literal identification of Jvālāmukhī as the “Goddess of the Flaming Mouth” reflects an evolution in the symbolic role of fire as both medium and object of worship. In the broadest sense, the tradition of Hindu (Brahmanical) orthodoxy saw fire evolve from its early Vedic role as a mobile sacrificial medium to a fixed marker of divine feminine power with the development of sacred sites known as Śakti Pīṭhas or “seats of power”. Integral to Jvālāmukhī’s identity, this transformation is famously illustrated in the Purāṇic myth of Satī. According to the myth, Satī, the daughter of King Dakṣa and Śiva’s consort, threw herself into the sacrificial fire after her father had deliberately excluded her partner from a festive ceremony. Arriving too late to save her, the wailing Śiva carried her burning body and circled the skies in anguish. As Satī’s body was dismembered into (usually) 51 parts, the places where they touched the ground turned into “seats of power”, potent sites where divine female power (śakti) is believed to reside (Sircar [1948] 1973; Urban 2010, pp. 34–35).3 According to these traditions, the goddess’s flaming tongue (jvālāmukhī) landed in the Himalayan foothills alongside a cluster of her body parts, creating a collection of sanctified spaces that are collectively known as the Jālandhara Pīṭha.

One of the four earliest Śakti Pīṭhas on record, the focal point of this seat of power lies in Kangra Town, where Satī’s breasts are said to have fallen and from which emerged the goddess Vajreśvarī. While local publications (e.g., S. Sharma 1996) link Vajreśvarī’s association with the thunderbolt (vajra) to her stabilizing power, her name is, in fact, suggestive of Tantric (Vajrayāna) Buddhism, which appears to have thrived in this region in earlier times.4 Be that as it may, the casting of Vajreśvarī as an icon of stability contrasts with Jvālāmukhī’s fiery, transformative energy to facilitate a popular perception of the two deities as complementary aspects of Śakti, the former anchoring the Pīṭha and the latter animating its course (more on this in Section 2.2.3, below).

As Sircar ([1948] 1973) had established some time ago, the tradition of Śakti Pīṭhas developed through a dynamic layering of religious strata over centuries. This entailed a projection of earlier Vedic sacrificial motifs upon Purāṇic mythologies, which were then elaborated within the sacred geography of Tantric Śāktism. Originally, the Śakti Pīṭha scheme comprised four cardinal locations: Śārdā (Kashmir), Kāmarūpa/Kāmākhyā (Assam), Uḍḍiyāṇa (Deccan), and Jālandhara (Himachal Pradesh/Punjab).5 Over time, this network expanded in tandem with the development of different Tantric traditions, a phenomenon witnessed in texts such as the Kālikā Purāṇa (Sircar [1948] 1973, p. 14). The inclusion of Jvālāmukhī within the Jālandhara Pīṭha underscores the site’s relative antiquity within this sacred geography. This is corroborated by the philological studies of Alexis Sanderson and his students, which trace the primacy of goddess worship in Tantric traditions to the Śaiva corpus, from which it developed in multiple sectarian strands culminating with the esoteric school of Kaula Śaivism and its various subsects (e.g., the cults of Kubjikā, Tripurasundarī, and so forth).6

While Jvālāmukhī seldom appears in these texts—and when she does, it is as a minor deity—her association with the Śakti Pīṭha network and the goddesses central to these traditions are indicative of her sectarian orientation and functions. Within Kaula Śaivism, fierce goddesses, or yoginīs, are highly regarded for their bestowing of transformative knowledge through teachings. Closely linked to these figures are the Mātṛkās, an earlier goddess collective rooted in Purāṇic traditions.7 Known as the Seven (or Eight) Mothers, the Mātṛkās embody protective maternal power, but also carry esoteric associations that prefigure yoginīs. Śaiva texts such as the Brahmayāmala describe Mātṛkās as “clan mothers” whose “fragments” (aṃśas) subsequently manifest as yoginīs and other Śākta deities (Hatley 2012; also English 2002). Over time, these traditions integrated into mainstream Tantric practice, as fierce goddesses associated with fire and energy centers (cakras) were summoned to occupy positions in the practitioner’s body. Recently published Śākta texts highlight these connections.

Shaman Hatley (2012, 2019), for example, characterizes yoginīs as a diverse collection of divine female figures that cuts across Tantric traditions, be they Buddhist, Śaiva, or Vaiṣṇava. This malleable framework is instructive of Jvālāmukhī’s alignment with the yoginī archetype and suggestive of her adaptability to different sectarian contexts. An early reference that explicitly mentions Jvālāmukhī may be found in the aforementioned Brahmayāmalatantra, where she is described as a she-jackal with a “flaming mouth” (Kiss 2015, p. 318).8 This imagery links Jvālāmukhī’s elemental worship to the transgressive practices central to the Kaula traditions that would emerge from the Śaiva schools. In later texts, she is squarely presented as a yoginī. The thirteenth-century Matsyendrasaṃhitā thus mentions her as one of fifty yoginīs that are said to embody “Śakti” and whose names all end with “mouth” (mukhī); these include “Disfigured-mouthed” (vikṛtamukhī), “Meteor-mouthed” (ulkāmukhī), “Beauty-mouthed” (śrīmukhī), “Wisdom-mouthed” (vidyāmukhī), and, of course, “Fire-mouthed” (Kiss 2021, p. 551).9

As the Śākta traditions spread into broader publics and communities, subsects tied to specific geographical spaces came into being. This process of localization is clearly demonstrated in the Mahāvidyā (lit., “Great Knowledge [Goddesses]”) tradition, which consolidated the worship of fierce feminine archetypes in Śākta Tantras (Sanderson 1988; 2014, pp. 11–12). Texts like the Brahmāṇḍapurāṇa and the Rudrayāmalatantra identify ten cosmic goddesses (Mahāvidyās) as embodiments of the single great goddess (Mahādevī) and, by extension, of the laws and powers underlying saṃsāra. As Jae-Eun Shin explains (Shin 2018, pp. 18–32), these textual traditions became localized by stressing the prominence of their goddesses for worship in the ancient Śākta seat of Kāmākhyā, Assam. As the Mahāvidyā tradition gained fame (approximately from the fifteenth century CE), its elements spread across India, where they became transposed and were adapted in various ways.

In the West Himalayan foothills, Mahāvidyā goddesses—including Bagalāmukhī, Cāmuṇḍā (alternatively, Caṇḍamuṇḍā), and Vajreśvarī—form a sub-set within the Jālandhara Pīṭha, filling localized roles that are distinct from their characterization in the textual tradition. While Vajreśvarī may anchor the site as its primary deity and could therefore potentially be regarded as anterior to other deities in the pīṭha, other goddesses from the group of ten came to address regional protective needs by occupying specific sites in the valley.10 The extension of Vajreśvarī’s influence over the subsidiary power seats in the Kangra Valley, including the temples of Tārā Devī, Cāmuṇḍā,11 and Bagalāmukhī, reinforces the prominence of the Jālandhara Pīṭha as a hub of Śāktism. And although Jvālāmukhī is not officially part of the ensemble of goddesses encountered in the textual tradition, her identification with Dhūmāvatī, the “smoky”, widowed Mahāvidyā goddess (Chaturvedi n.d., p. 5), is clearly aimed at incorporating her into the latter.

Reflective of the conceptual framework of Śākta traditions, these parallels and homologies are elaborated in official publications compiled by temple priests and officers, providing further clues regarding the goddesses’ sectarian pasts. The government’s guide to the temple at Kangra, for example, the Śrī Vajreśvarī Devī Mandir: Itihās evam Paricay (S. Sharma 1996), explicitly states that the worship of Vajreśvarī follows protocols prescribed for Tripurasundarī. Reflective of long-term processes of cultural appropriation and change, this prescription conforms with the historical triumph of the Tripurasundarī cult over competing Kaula sects from the early modern era onwards (Mallinson 2016).12 Similarly, the prescription of ritual protocols associated with Dhūmāvatī for the worship of Jvālāmukhī (S. Sharma 1996, p. 34) hints at an early modern realignment of the goddess with the popular cult of the Mahāvidyās.

By way of contrast, the adaptations of Mahāvidyā goddesses to the higher reaches bordering the Tibetan plateau tend to minimize theological connections with pan-Indian cults, highlighting local conceptualizations and praxis instead. This is clearly borne out in the case of the goddess Mātaṅgī. Associated with ritually impure women in Indic sources (Aktor 2016), Mātaṅgī acquired decidedly local markers in the frontier region of Kinnaur (Himachal Pradesh, India). According to a firsthand account by a resident of Sangla Valley in inner Kinnaur dated to the 1920–30s, “matingos” were liminal deities that governed the wilderness beyond human settlements. As befitting of entities living beyond the pale of civilization, the matingos had to be propitiated through offerings of animal entrails that were cast into sacrificial pits near streams and at river crossings (Negi 2012). While these practices echo the emphasis on transgression that characterizes Mātaṅgī in the Mahāvidyā tradition, their audiences follow local mores with barely a hint of any link with the Sanskritic tradition.

The attested interaction between scholastic, regional and popular traditions outlined above reflects a general dynamic of Śākta localization, in which ritual beliefs and practices found in the Kaula, Mahāvidyā, and related traditions are continually being echoed and expanded to accommodate local settings. While the long-term nature of these processes of acculturation and adaptation may impede the identification of a precise trajectory for Jvālāmukhī’s worship, the outline of “Śaiva Literature” by Alexis Sanderson hints at one possible path. According to Sanderson, the Kashmirian brahmins who followed the “syncretistic, Tripurā-centred Śākta tradition of the Devīrahasya” worshipped “Śārikā, Śāradā, Rājñī, and Jvālāmukhī” as lineage goddesses (kuladevī) (Sanderson 2014, pp. 80–81).13

Supported by local tradition regarding Abhinavagupta’s guru,14 this would suggest a close connection between the developments in eastern Śākta traditions, Kashmirian brahmins, and Jālandhara, delineating a plausible trajectory for the importation of fierce goddesses associated with the Kaula/Mahāvidyā traditions into Kangra. This places the transposition and projection of yoginīs, Mātṛkās, and Mahāvidyās upon the Jālandhara Pīṭha landscape on a somewhat firmer footing and strengthens their alignment within the broader realm of Śākta cosmology. To gain a sense of how these adaptations might have played out, we may turn to the relatively better-documented history of the temple to Jvālāmukhī in modern Kangra.

2.2. Jvālāmukhī in Kangra Myth and History

The contrast between the relatively minor figure of Jvālāmukhī in Tantric texts and the goddess worshipped in the Kangra Valley is substantial. Manifesting in gaseous flames issuing from the rocks in a vast temple complex, the goddess occupies a nodal position within the subnetwork of sacred sites that constitute Jālandhara Pīṭha. As with the Śākta center at Kāmākhyā (site of the Purāṇic deity’s fallen womb), the temple seems to have developed in a region that was already steeped in goddess worship for centuries before it emerged as a site of supra-regional importance. While the center—and, by extension, holiest part—of the Jālandhara Pīṭha is widely recognized as the temple of Vajreśvarī in Kangra Town, the temple at Jvālāmukhī gradually came to rival the latter.15

By the early modern era, the reputation of Jvālāmukhī as a sacred site of significance had sufficiently grown to attract key historical figures. Sultan Firoz Shah Tughlaq, for example, is explicitly mentioned as visiting the temple in 1361, passing at the site en route to punish the ruler of Kangra (contrast with Mahmud of Ghazni, whose raid of Kangra Town in 1009 CE makes no mention of Jvālāmukhī).16 The ties between the temple and its custodians with the subcontinent seems to have radically increased with the integration of the Punjab Hill States into the Mughal Empire. Beginning with Akbar’s campaigns of the 1580s and culminating with the conquest of Kangra Fort in 1620, the temple was part of Jahangir’s celebratory tour of hills that year. The first—and apparently only—Mughal Emperor to have visited the temple, Jahangir, lamented the flocking of Hindu and Muslim pilgrims to the site in a misguided belief that it would grant them the goddess’s blessings (Jahangir [1621] 1914, pp. 223–25). Indicative of the temple’s growing popularity with publics from different religious backgrounds, these remarks underline its function as a site of pan-Indian appeal. By the close of the nineteenth century, the temple of Jvālāmukhī had earned an “even greater reputation than that of Kangra” (Punjab Government 1884, p. 254).17

The images worshipped in the temple today consist of “nine flames” (navajyotis), whose actual number tends to fluctuate between three and fourteen depending on the intensity of geothermal activity. The central flame, surrounded by a gilt silver frame, is identified as the great goddess Mahākālī. The remaining flames manifest female deities associated with other Sanskritic goddesses, whose blessings are sought to tackle specific challenges.18 As with other temples in the hills, custodianship of these shrines seems to have changed hands between various groups from within and outside of the hills (M. Sharma 2022). By the mid-nineteenth century, its main keepers consisted of Brahmanical Bhojkis, householder Śaiva Gosains, and ascetic yogīs of the Nāth sampradāya. Mindful of the cultural orientation of the Jālandhara Pīṭha within Tantric tradition, this section examines the relationship of these divergent groups towards the goddess and her temple.

We begin with an examination of the origin myths told about the temple’s founding, which highlight the enmeshment of religion and politics in local history. This is followed by a review of the main custodian groups and their networks, illustrating the diversity of agents involved in maintaining the temple in the modern era. We conclude with a review of the Jālandhara Pīṭha’s sacred geography in a locally authored pilgrimage guide from the 1860s, which both affirms Jvālāmukhī’s standing as a leading deity in the seat of power and illustrates the continued relevance of Tantric Śākta heritage to regional history.

2.2.1. Origin Myths

The story of the discovery and founding of the temple has several versions, all of which relate to the Purāṇic narrative about the origin of Śakti Pīṭhas (Section 2.1, above). Thus, while the creation of the site is unequivocally linked to Satī’s dismemberment, the tales concerning who had discovered the flames and what they did afterwards vary. The version most commonly told today stresses the connection between the goddess and the valley’s erstwhile kings, the Katoch Rajputs of Kangra. Recurring in popular and official publications from the colonial era to the present, it may be summarized as follows:19

When Śiva circled the skies with Satī’s burning corpse, her fiery tongue fell and disappeared in the thick of the jungle. Since it could not be discovered, a temple was founded in Nagarkot (Kangra) to honor the goddess instead. This temple is the place where Vajreśvarī Mata now resides. Years later, a herder who used to tend cattle in the forests of Trigart (the ancient name of Kangra) stumbled upon a flame burning miraculously from a rock in the depth of a forest. Astonished by the sight, the herdsman rushed back to inform the ruler of Trigart about his discovery. Rājā Bhumi Chand, the founder of the Katoch Rajput lineage, then traveled to inspect the site. Being a direct offspring of the great goddess—he was born to Mahādevī during her battle with Jālandhara, emerging fully armed when a drop of her perspiration hit the ground [bhūmi]—he immediately recognized its sanctity and began worshipping it.20 He then ordered a temple to be built around the flame and imparted the legacy to his descendants. Thousands of years later, the Pāṇḍavas would find the temple in ruins and reconstruct it.

By welding popular belief in the goddess to the dynastic account of the valley’s rulers, this narrative successfully underlines the link between religious and secular authorities in Kangra. Early colonial administrators were more circumspect in reading this myth and quite plausibly proposed that it was based on an earlier Śaiva version, in which the flames referred to were not the goddess’s tongue, but rather the steaming mouth of Jālandhara, a demon (daitya) defeated by Śiva. It was thus Śiva—rather than Mahādevī—who purged the region of demonic powers and rendered it habitable by pinning Jālandhara’s humongous body down (Punjab Government 1884, pp. 9–10). The popular version of this tale, however, replaced the victorious Śiva with the goddess and located her defeated enemy’s body with the Kangra Valley heartland: the demon’s head was in Baijnath (east), the navel in Kangra (center), the shoulders in Jvālāmukhī (south) and Triloknath (north), and its feet in Guler (west) (Punjab Government 1884, p. 248, fn 3).

Other versions are even more explicit about the connection between the goddess and the flames. In a variant telling from the same region (Vashishta 2004, pp. 8–9), the herder is said to have noticed that one of his cows had stopped giving milk despite grazing freely all day long. Puzzled by its emptied udders, he decided to follow the bovine into the woods. As the cow meandered into a clearing, a beautiful girl appeared from between the trees with her tongue hanging out, eagerly approaching the animal for milk. The herder then initiated contact, only to see the girl dissolve into a great flashing light, demonstrating her divine status.21 As in the commonly told version, the herder informed the ruler, who ordered the construction of a temple on site. The monarch also invited two “Bhojak” Brahmins from Śākadvīpa—paṇḍits Śrī Dhar and Kamlā Pati—to serve as her attendants, their progeny supposedly filling the same roles to date (Vashishta 2004, p. 9). In yet another variant, the herder is cast as a Brahmin who migrated “from the south” following the goddess’s summons in a dream, circa 1200 CE. Evidently aimed at gaining credit with orthodox Hindus, the story concludes with the Brahmin discovering the “tongues of fire” and earning the patronage of the local ruler (Punjab Government 1884, p. 255).

Saturated with popular tropes about the goddess—manifesting as a virgin-child (kanyā), communication in dreams, etc.22—these versions emphasize the link between the deity’s attendants and Rajput patrons. Reflecting sociopolitical conditions in nineteenth-century Kangra, they present the “Bhojak” priests and the Katoch rulers as local variants of the dharmic pact between councilor-priests and warrior-chiefs promoted in orthodox Brahmanical literature. Emulated by generations of headmen, the patronage of the temple and its custodians extended beyond the Katoch clique of kings to numerous political leaders, transforming Jvālāmukhī into an arena for showcasing wealth, discussing politics, and, in certain cases, signing treatises that would alter the course of history.23

2.2.2. Custodians

While the tales regarding the temple’s foundation successfully cast kings as patrons of religion, their sanitized depiction of Śāktism belies the diverse—and at times, distinctly heterodox—practices associated with Jvālāmukhī’s custodians. Consider the Bhojkis, the primary ritual officiates (pūjarīs) who tended the central shrine. In the preceding narrative myth, the Bhojkis are cast as Brahmins from a mythic island (Śākadvīpa) whose office is sanctioned by the Katoch kings. To an extent, this view is consistent with the historical data: eighteenth century documents clearly note the Bhojkis’ taxation of visitors to their temples being sanctioned by the Kangra kings (Goswamy and Malhotra 1973). However, the British conquest of Kangra (in 1845) upended the alliance between Rajputs and priests due to the negative bias of colonial officials towards religious orders in the hills.

Reporting on Jvālāmukhī, Kangra’s first settlement officer explained that the Bhojkis formed part of a large class of hereditary officials who served in the Kangra Valley’s Śākta temples.24 And while the temple priests claimed to be Sārasvatī Brahmins and “all wear the sacred thread” to support this, they were, in fact, “not Bráhmans” at all, but rather “a debauched and profligate set”; they would “intermarry among themselves alone, eat flesh, drink wine”, their men were “constantly in the Courts involved in litigation, and the women […] notorious for their loose morality” (Rose 1919, vol. 1, p. 107). Colonial officialdom thus appears to have been less forgiving of divergences from Brahmanical orthodoxy than the rulers of Kangra had been. And while the consumption of meat, anathema to orthodox Brahmins but common in the hills, would have labelled them heterodox in conservative circles, the remainder of the shortcomings listed seem to derive directly from negative biases of the colonial regime. The “loose morality” attributed to Bhojki women thus agrees with the laxer attitudes towards conjugal relations in the hills, while the “litigious” character of Bhojki men was likely prompted by the diminishing of the privileges that they had formerly been granted by local kings.25

By the turn of the twentieth century, the official assumption regarding the Bhojkis was that they were “mere Jogís who have obtained a reflected sanctity from the goddess whose service they have entered” (Rose 1919, vol. 1, p. 107). Taking a long view of history, this reading does not seem entirely unfounded. As Daniela Bevilacqua proposes (Bevilacqua 2024, p. 40), pilgrimage sites associated with Tantrism would have played a part in bridging the transgressive rites of early yogī recluses from the Kaula Schools with householder society through interactions between pilgrims and itinerant/resident ascetics. In this respect, the casting of Bhojkis as bona fide Brahmins in local narratives belies a more complex social history, in which Tantric and orthodox communities coexisted as ritual service providers within the same temple complex. The recorded presence of a sizeable community of Gosains in the township of Jvālāmukhī would support this reading.

Similar to the Bhojkis, the Gosains seem to have originated with ascetics who adopted the householder way of life. However, contrary to the former, who primarily served as temple custodians, the Gosains were only tangentially connected with temple shrines, their main earnings being connected with mercantile activity. In this respect, the Gosains agree with the common portrayal of ascetic groups as extraneous to settled society, but who were in fact deeply intertwined with the administrative and economic histories of the hill states.26 Comprising the principal portion of the town’s 2424 permanent residents during the 1880s, the Gosains of Jvālāmukhī were wholesale traders in opium and medicines with a turnover of 125,000 Rs per annum (Punjab Government 1884, pp. 254–56). Colonial officials who reported on these figures judged them to be a fraction of the income generated in earlier times and that was reflected in an abundance of sizeable, dilapidated buildings and havelis surrounding Jvālāmukhī Town, witnesses to this once bustling entrepot on the high road connecting Kangra Valley with the Punjab Plains.

These details shed new light on Jvālāmukhī that complements its envisioning in origin myths. Still a site of religious and political significance, the temple and the town that surrounded it were also a trade center that seems to have been at least partly independent of the regular administration of the state. And while the Gosains played a key part in these long-distance commercial activities, they also filled a role as custodians of sacred sites. The shrine of Arjun Siddh (alias Nāgarjuna Siddha), for example, accessible by a long set of stairs above the main temple, is tended to by Gosain families to this day. Set next to an impressive peepal tree, the precincts contain seven burial mounds of Gosain ascetics (samādhis), suggesting it had once indeed been the turf of ascetic priests. Today, however, ritual services on the site are provided by householder Gosains who perform their duty on a rotating basis, enabling members of the community to uphold their ascetic heritage while earning a living by pursuit of various professions (personal communication, Naresh Sharma, 20 September 2021).

The third group of custodians worth noting are the ascetics of the Nāth saṃpradāya. Occupying a large structure above the main sanctum that is said to have been founded by a disciple of Gorakhnāth, their shrine has flames issue from the cliffside above a naturally carved pool of water (a very similar setting in Mustang, Nepal, is explored below). The flames are unique in that they can be made to rise up dramatically, signaling divine presence. James Mallinson, who visited the site in April 2025, describes how this is performed: “In a small cave at the shrine there is a basin of water at the bottom of a niche in a wall, up through which flow bubbles of gas into the niche. Every few minutes a Nāth yogī ignites the gas that has accumulated in the niche with a taper, making a small explosion, and then splashes the four or five pilgrims who have crammed into the cave with water from the basin, both as a blessing and to show that despite the bubbles the water is cold” (email communication, 12 June 2025).

In the wider world of the Nāth saṃpradāya, Jvālāmukhī is famous as one of three especially sacred goddess sites in North India, alongside Hinglaj (Pakistan) and Devi Patan (near Tulsipur in Gonda District, Uttar Pradesh, India). Accordingly, it is visited by numerous ascetics of the order—i.e., the Nāth jamāt—during both the autumn and spring Navarātras.27 From a wider historical perspective, the link between the saṃpradāya and Jvālāmukhī is congruent with the political empowerment of ascetic orders in the early modern period. Thus, while Nāth claims to provenance with Kaula Śaivism agree with the general trajectory of East Indian Śākta systems spread into the region (Section 2.1, above), their prominence in the North Indian and Himalayan landscape is indicative of the extensive spread of the saṃpradāya network in the second millennium.

Administered from Bālnāth Ṭillā in the Punjab (the order’s headquarters till Partition, presently in Pakistan), the Nāths cultivated relations with Mughal Emperors (Akbar twice visited the site) and Rajput leaders (Nepal included), established subordinate “seats” (gaddīs, e.g., at Jakhbar), and entertained close ties with competing sects throughout the region (e.g., the Vaiṣṇavas of Pindori).28 In this respect, the presence of Nāth yogīs in Jvālāmukhī comprises a distinct early modern variant of the broad conceptualization of yoga as an expression of political power across South Asia (Kale and Novetzke 2024). The dynamic networks and exchange patterns sustaining this particular iteration of power were radically challenged with the transition to colonial rule. Overshadowed by the Rajput rulers’ alliance with the British, the custodians of the temple were forced to devise new ways to retain their hold on power. The composition of a pilgrimage manual to the Jālandhara Pīṭha some two decades after the British conquest is a case in point.

2.2.3. Sacred Geography

Authored in 1864 and republished with an extensive Hindi introduction in 1983, the Jālandhara Pīṭha Dīpikā (hereafter, JPD) seems to have been written by a Kaula Śākta ascetic with a view to explaining the merits of visiting this seat of power.29 The text projects established Tantric tropes onto the Kangra Valley landscape with marked effect, depicting Jvālāmukhī as both a Tantric protector deity and an accessible figure of popular devotion to be propitiated through acts of piety along the pilgrims’ route. The author of the text instructs readers to visit no less than 141 sacred sites over (a minimum of) 68 days. The majority of the shrines and temples on this itinerary are Śaiva and Śākta, and a handful are Vaiṣṇava institutions. The goddess Vajreśvarī is unequivocally presented as the supreme entity in the Pīṭha, being the cosmic focal point (bindu) in a network of interrelated sites dominated by four secondary female deities: Jvālāmukhī, Ambikā, Jayantī, and Añjanī.30

Although the text acknowledges Vajreśvarī as supreme, it subtly subverts this claim by prescribing a pilgrimage route and rewards that ultimately center on Jvālāmukhī. While the conclusion of the pilgrimage in the temple of Vajreśvarī in Kangra Town conforms with its historical status as the ultimate destination for pilgrims (tīrthayātrīs) and plunderers, the ultimate benefactor and granter of wishes is said to be Jvālāmukhī. Thus, while the latter is implicitly recognized as subordinate to the former, the text’s final verses effectively unsettle this hierarchy by directing pilgrims to return from Vajreśvarī’s temple to that of Jvālāmukhī, where they are to perform a series of rites for awakening kuṇḍalinī. The Nāths’ proverbial claim to Kaula knowledge grants weight to these instructions, which aim at uniting Śiva and Śakti as a means for salvation (mokṣa), the release from saṃsāra being further supported by the form and structure of the pilgrim’s route.

The text contains 569 ślokas that are divided into 11 chapters (dīpikās, literally, “lights”) with each chapter devoted to a different stage along the pilgrims’ path.31 Specific instructions are provided as regards to the shrines that need to be visited, the amount of time one should stay there (the more powerful the site, the longer the stay), the type of offerings required, and in key instances, the laudatory verses (stotras) to be recited. These localized concerns are linked with the Purāṇic tradition by framing the tale in an epic frame. Thus, in its opening verses, the legendary author of the Mahābhārata Vyāsa turns to the sage Vaśiṣṭha in his search for a “pure place” (pavitrapada) for poetic inspiration. The latter directs him to Jālandhara Pīṭha, which he describes as the holiest of all 64 Śakti Pīṭhas, where even outcastes, if cremated there, may gain salvation (P. Shastri 1983, p. 4). This exceptional situation, the sage explains, is due to the Kangra Valley’s constituting a physical manifestation of the triangular yantra of Mahādevī with the goddess Vajreśvarī (alias “Mahi Mai”, as per S. Sharma 1996) at its center. To receive the latter’s blessing, the pilgrim is prescribed an inwards spiraling path that leads from the edges of the “yantra” (the boundary points of the valley) into the focal point at Kangra (JPD, First Light).

The yantric design and kuṇḍalinī shaped-route—explicitly fashioned after the 3.5 folds of the serpent-goddess residing at the base of the pilgrim’s spine—are a modern projection of late Kaula beliefs (specifically, of rites associated with the cults of Kubjikā and Tripurasundarī; Sanderson 1988, pp. 702–3). They are also consistent with the appropriation of sacred sites and traditions by ascetic orders and constitute an important mechanism for the dissemination of esoteric knowledge to wider publics (e.g., Bouillier 2017). Combining a cosmological vision with practical instructions, the text encourages pilgrims to engage in both ascetic practices and devotional acts of sacrifice, including animal sacrifices (paśubali).32 In yet more extreme cases, the text calls for offering one’s tongue to Jvālāmukhī, implicitly acting on the popular belief that it will regenerate as per the practice that has long been associated with the cult of Vajreśvarī in Kangra.33

At the same time, the text calls for generous donations to Brahmins at the numerous shrines along its path regardless of sectarian affiliation. This apparent inclusivity constitutes a significant departure from the exclusivist lineage-based practices associated with esoteric traditions and a radical reorientation of religion towards the syncretic model of devotionalism that emerged from the early modern era.34 The mingling of popular devotion with transgressive practice is readily apparent in the story of Dhyānu Bhakto, the 16th century devotee who is said to have cut off his head out of devotion to the goddess, and whose images adorn the temple.35 By fusing esoteric practice with popular belief, the story overcomes the more gruesome custom of offering one’s own tongue, while reinforcing the goddess’s status as a deity who commands profound acts of surrender. By presenting a unified Śākta landscape where blending Tantric heritage with popular devotion, the author of the JPD devised a text that asserts Śākta authority while recognizing popular sentiments to overcome the challenges facing religious orders in British India.

Despite their limitations, these findings ultimately strengthen the attested openness of Tantric traditions to inter-sectarian and inter-doctrinal exchange. While the gap between the manifestations of the Goddess of the Flaming Mouth in Hindu Tantras and in the temple in Kangra remains wide, it is worth recalling the capacity of certain groups of tantrikas for adaptive assimilation. The marked flexibility of Kaula Śaivism in the Konkan region, for instance—which facilitated the absorption of Vajrayāna Buddhist elements, thereby contributing to the rise of fierce goddesses in Tantric and Kaula worship (Mallinson 2019)—may serve as an example of such inter-sectarian fertilization. And while further research on site—especially in iconography and paleography—is necessary before any such crossovers between Indic and Tibetan traditions concerning Jvālāmukhī may be ascertained, the idea of such beliefs and identities transitioning between the two realms seems to be far from implausible. To put this hypothesis to the test, we must take a closer look at the Tibetan material on hand.

3. Tibetan Sources on Kha ’bar ma

While Jvālāmukhī’s place in Hindu traditions does not appear to be clearly defined, she at least has a dedicated temple where she is worshipped and a broad textual tradition that is transposed upon her. Her Tibetan counterpart remains an even more vague and enigmatic figure. There seems to be no known center for her veneration, such as a temple or shrine, nor are we aware of her iconographic representations as a protective deity within the Buddhist pantheon. The depictions known to us—whether in Buddhist or Bön sources—present her solely as a demonic being.

At the same time, the Tibetan sources turn out to contain relatively rich material, enabling us to reconstruct, to a certain extent, her role in Buddhist history and, ultimately, to assess whether links to Tibetan-Himalayan cultural interrelations can be traced in Tibetan rituals and legends associated with Kha ’bar ma, including those tied to the Bön tradition. This section thus necessarily differs from the previous one insofar as it offers a summary of ritual and narrative details extracted from the texts available to us. Nevertheless, we will aim to build a cohesive narrative in its final paragraph with a view to using these details to trace possible links between the Indic and Tibetan traditions.

3.1. Buddhist Material

3.1.1. Texts Included in the Tibetan Buddhist Canon

The first part of the canon, the Kangyur, contains two versions of a dhāraṇīsūtra in which Buddha Śākyamuni provides his disciple—Nanda in one version and Ānanda in the other—with the sacred formula (dhāraṇī) and instructions on making the torma (ritual cake) offering to the realm of pretas, ghost spirits.36 This disciple is threatened by an ugly preta (male) with a flaming face called Kha nas me ’bar ba37 in the first text38 or an ugly pretī Kha ’bar ma (female) in the second.39 It is notable that the expression Kha nas me ’bar ba is also found in the Tibetan version of Suhṛllekha40 by Nāgārjuna when types of pretas are described.41 The hideous demonic being warns Nanda/Ānanda that he will die in a week (according to the first text) or in a day (according to the second text) and be reborn as a preta in that location unless he makes a massive offering of food and drinks to a colossal number of hungry ghosts and pleases one hundred thousand Brahmanical sages. The preta/pretī also says that if this offering is successfully made, he/she could die and be reborn in the realm of gods. The Buddha’s intervention apparently saves his disciple from death and offers other adherents a powerful method for handling similar threats in the future.

No Sanskrit originals of either version have been found. No Sanskrit titles are provided in the majority of the Kangyurs,42 no translators or editors named in the colophons. As Paul Hackett notices in the introduction to his English translation of the first text, both texts appear in Chinese versions within the Chinese Buddhist canon. The first was produced by Amoghavajra between 757 and 770 CE, the second, by Śikṣānanda, is dated to ca. 700 CE (Hackett 2023, i.1–2).43 Hackett remarks: “Given the close resemblance between the Tibetan and Chinese versions (Tôh 646 with Taishō 1313, and Tôh 647 with Taishō 1314), it is possible that both Tibetan texts were translated from Chinese” (ibid., i.5).44

We can be confident that both texts were translated into Tibetan during the “early spread” (snga dar) of Buddhism in the Land of Snows. The first text is found among the Dunhuang manuscripts in the Stein Collection (IOL Tib J 349),45 while the second is mentioned in the early 9th-century catalogue of translated texts known as the Denkarma (Ldan dkar ma).46

Kha ’bar ma, as a secondary figure, is mentioned in some of the so-called “old tantras”, associated with the Nyingma (“old”) school of Tibetan Buddhism, which were excluded from the canon by earlier compilers as inauthentic. A selected group of these texts was later added as a special section in the major editions of the Kangyur.47 The most significant of these texts is the Guhyagarbhatantra. In its third chapter, the tantra introduces the concept of the six munis (sages) (Tôh. 1647, f. 312a5), or buddhas of the six realms of saṃsāra.48 Each of these munis has a specific name, and Kha ’bar ma is the muni of the realm of pretas.49 The reference is to Lama Chönam, and it appears in the reference list under the first word of the name. However, during the editing process, the word Lama was omitted in the citation code. I would suggest restoring Lama to maintain consistency and proper attribution. This muni is also mentioned in several other old tantras not included in the Kangyur.50 Although the figure appears to be male in this context, one of these old tantras refers to Kha ’bar ma as the queen of pretas.51

One “old tantra”, featuring Buddha Vajrakumāra (also known as Vajrakīlaya) as the protagonist, presents a list of his emanations in the second chapter, intended to teach a more detailed classification of disciples: bodhisattvas of the ten stages, bodhisattvas of the eight stages, gods of the Brahma realm, other gods, asuras, planets, nāgas, infernal beings, hindrances, animals, and humans.52 The reference is to rKTs: Gpb025.009, and this entry is found in the reference list. The term “hindrances” (sgrib pa rnams) may correspond to pretas in this list, as the name of Vajrakumāra’s emanation intended for them is Kha ’bar ma. The fifteenth chapter of this text, composed in verses, tells the following story relating to the time of the taming of Rudra (a name of Śiva), “the fierce ignorant one”:53

When the great glorious Vajrakumāra/“Liberated” the great Rudra,/The Lion of Speech bestowed empowerment,/And [his] tongue juice became the nectar of saliva,/Four drops fell,/Blood and saliva got mixed, and from that/The great yakṣa Kha ’bar appeared./[He] went to Śākyamuni/And offered his cut-off tongue./<…>/From the two drops that fell to the ground,/The wrathful goddess Great Glorious Blaze (Dpal ’bar chen mo) appeared./She went to the Bhagavān/And offered [her] cut-off tongue./<…>/From a single drop falling to the ground,/The preta Kha ’bar ma was born./He went to Śākyamuni,/And offered [his] tongue as a gift./Nectar dripped and dripped from the wound (?),/It was born as a small tree (?),/With roots that twirled and twirled with vigor (?),/Its flowers were blue and shaped like tongues./It is a very cooling medicine,/Not to be processed [anyhow]:/By simply eating it, illnesses are cured./A single drop fell to the realm of the gods,/And a female goose was born…54

The parallels with the Indic traditions of the Jālandhara Pīṭha are striking. At the level of narrative, the story echoes the origin myth of Vajreśvarī and the rulers of Kangra, Kha ’bar ma here being born from a drop of saliva falling to the ground and mixing with blood (possibly that of Rudra). In terms of ritual practices, the offering of tongues closely corresponds with the cult of the goddess of Kangra and, to a lesser extent, that of Jvālāmukhī. From an intra-Tibetan perspective, we may note that the quoted fragment shows that, apart from a preta, an almost identical [?] name referring to the image of the blazing mouth can be applied to another type of demon—in this case, yakṣas. This is consistent with certain other “old tantras” that use the name Kha ’bar ma to designate demons or minor Tantric deities of various types: a host of herukas (blood-drinkers) serving as punishers of those who violate the samaya (Tantric pledges);55 one of the so-called “vajra messengers” (rdo rje pho nya), which is synonymous with ḍākinī, female partners of yogīs in their secret practices;56 a rākṣasī cannibal whose husband was the god of wind (rlung lha) and whose child was a discord-causing demon turned protector deity;57 one of the consorts (or sisters?) of Yama;58 one of ma mo (mātṛkā) demonesses;59 even one of the two assistants of the god of fire.60 The multiplicity of Kha ’bar ma as both fierce and protective thus underscores a broader trend in goddess worship, observed in both Indic and Tibetan contexts, where goddesses transition between wrathful and protective roles depending on the ritual and regional context.61

One more text, Āryavidyottamamahātantra, an old tantra that is part of the core Kangyur,62 mentions Kha ’bar ma as one of the six “messengers” (pho nya mo drug) and provides her ritual, which has no parallels with what we know from the dhāraṇīsūtra discussed above, as this pho nya mo (ḍūtī) seems to have no link to the queen of pretas.63

Another important tantra included in the Kangyur, Śrīḍākārṇavamahāyoginītantrarājanāma, mentions the ḍākinī Vajrajvālāmukhī in its Sanskrit version (Sugiki 2019, p. 20), but the Tibetan translation refers to her as Rdo rje ’bar zhal (with zhal being an honorific counterpart for kha, a word of common language).64

Similarly, the Sanskrit text of Vajraḍākamahātantrarāja mentions Jvālāmukhī as the devī (goddess whose role is played by a local girl)65 of the sacred district Kollagiri, although another devī is also assigned to this location, and the Tibetan translation mentions only the latter. Notably, this goddess, Mahālakṣmī (dpal chen mo), has a “face of fire” (agnimukhī = me yi kha), and her male partner is directly named Agnimukha (me yi gdong can).66

The list of twenty-four districts, each hosted by a devī and her male partner, a “field-protector” (kṣetrapāla), likely follows a similar list in the Hindu Tantric text Kubjikāmatantra. However, in that text, Jvālāmukhī is assigned to a different sacred district, Jayantī (Sugiki 2003, pp. 65–67, 83). As a matter of fact, this is an important detail that will be addressed in Section 3.3.

The second part of the Tibetan Buddhist canon, the Tengyur, has three texts where the expression Kha ’bar ma is attested. The oldest of them, Triśata-kārikā-vyākhyāna by Vinītadeva (fl. 7th or 8th century), applies this name to a demon wearing clothes made of rugs from dust heaps.67 This identification not connected with the realm of pretas seems to be a singular instance, not supported by other texts.68

Another text, *Śrīsarvadurgatipariśodhanamaṇḍalavidhikṛpāvalī, attributed to Ānandagarbha (9th c.), mentions “the Preta King Kha ’bar ma, of red-black color, very emaciated, holding a blazing iron club in his hand”69 as one of the seven demonic beings comprising one of the circles visualized by a yogī. The text is said to have been translated by *Kiraṇākaravarman and Khyung grags (11th c.). The latter is a controversial figure, as some Tibetan authors, such as Bu ston rin chen grub (1290–1364), accused him of forging several texts that he claimed to have translated. However, the text mentioned above is not included in these accusations (See Bajetta 2019, pp. 17–23).

Finally, the text *Śmaśānasaṃskārasādhanavidhi, whose author and translators are not identified in the colophon, mentions Kha ’bar ma as one of the six munis (Tôh. 1647, f. 312a5).70

3.1.2. The Subsequent Tibetan Buddhist Tradition

Tibetan practices associated with the queen of pretas, Kha ’bar ma, are primarily focused on the ritual of the great torma offering. The manuscript XT-72 (preserved at the Institute of Oriental Manuscripts of the Russian Academy of Sciences), from Khara-Khoto, which served as a fort of the Tangut State until it was conquered by the Mongols in 1227, is probably the earliest extant physical document testifying to the spread of the ritual in Tibet.71 Although the manuscript may have been produced later, during the period of Mongol rule, lasting until the early 1370s.72 In any case, XT-72 records information about two traditions of the ritual, both referring to Indian Buddhist masters.

One of them is Atiśa (982–1054), and we have found only three other instances in the Tibetan literature confirming that this famous propagator of Buddhism in Tibet promoted this ritual, two from the pre-Geluk period (Ibid., pp. 46–47) and one dated to the first half of the 19th century. The early Geluk authors, who regarded Atiśa as their predecessor, do not seem to have utilized this ritual in their practices. XT-72 briefly mentions the story of the Buddha’s disciple’s encounter with the queen of pretas, as recorded in the canonical dhāraṇīsūtra, only in connection with Atiśa.

A much more widely accepted account is the story of transmission of the ritual by Bari Lotsāwa (1040–1112) from the Indian master Vajrāsana the latter (11th century). The earliest rendition of this account is found in the text by Bsod nams rtse mo (1142–1182), one of the first five hierarchs of the Sakya school, to which Bari Lotsāwa also belonged:

The story [is] as follows. Bari Lotsāwa went to Nepal and invited a pandit to come with him. Having no gold, they arrived in Purang. Upon meeting Tsepo Tsede,73 [they] conferred empowerment on the king and taught [him] the Dharma. The king offered one hundred ounces of gold—[so they] say. The pandit put the gold away, and the lama asked [him] for a little bit. [The pandit] gave nothing, so they had a quarrel. Deeply upset, [Bari] took the pandit back to Nepal—[so they] say. Then, on the way to Nepal, he met a tīrthika (non-Buddhist) pandit named Bhavyarāja, who said, “You must become my translator and come with me to Tibet”—[so they] say. The lama said to him, “[I] will not”. That [tīrthika] said, “I will not teach non-Buddhist teachings. I will teach the Buddhist dharma.” The lama still refused, and that [tīrthika] said, “I will use my magic on you.” The lama said, “I have done nothing wrong to you. There is no need for your magic”, and went to India—[so they] say. Then, when he met guru Vajrāsana and was receiving teachings, [Vajrāsana] said, “You will be harmed by the tīrthikas”. “May it not happen?” [he] asked. “It will happen”, said [Vajrāsana]. “Well, I met a tīrthika named Bhavyarāja in Nepal, and this and that happened”. The guru gave [him] the “Bha-ya-na-sa” mantra74 and [the instruction] on this torma. [Bari] recited the “Bha-ya-na” mantra and threw the great torma three times. In the end, [the tīrthika] died—[so they] say. Therefore, this is said to be a very powerful blessing.75

This story, in an abridged form76, is repeated in many other texts, including XT-72. In this version, Bhavyarāja’s anger is explained by his defeat in a dispute and some rude remarks that Bari Lotsāwa made following his victory.77

The name Bhavyarāja must refer to the well-known Kashmiri Brahmanical paṇḍita who assisted Tibetan Lotsāwas in translating Buddhist treatises on logic. Notably, one of his close Tibetan associates, Ngok Lotsāwa (Rngog lo tsā ba) Blo ldan shes rab (1059–1109) (See Jackson 2012, pp. 92–95), from the Kadam school of Buddhism, authored a text on the Black Kha ’bar ma.78 Unfortunately, this text remains unavailable to us, and we only have some excerpts translated into English by Irmgard Mengele (Mengele 2010, pp. 120–21). There are no details concerning the source from which Ngok Lotsāwa obtained the ritual he described. Consequently, we can only guess whether it could be Atiśa, the founder of Kadam, some other Indian or Nepalese Buddhist master, or a local Tibetan tradition.

One more Indian name must be mentioned here, as it is referenced in several texts. This name is Padmasambhava, a great yogī of the 8th century venerated by the Nyingma school as the Second Buddha. One text, which belongs to the Nyingma tradition, describes a ritual for averting obstacles and “deceiving death” and simply mentions Urgyen Padma (i.e., Padma from Oḍḍiyāna) in its colophon;79 another Nyingma text claims to present a ritual of the torma offering to White Kha ’bar ma in Urgyen Padma’s tradition.80 The Sakya author Zhu chen Tshul khrims rin chen (1697–1774) mentions Padmasambhava’s hidden instructions discovered by Guru Chos kyi dbang phyug (1212–1270).81

A very significant text was compiled by Tāranātha Kun dga’ snying po (1575–1634), the eminent master of the Jonang school. It consists of several parts, beginning with a close rendition of Bsod nams rtse mo’s story about Bari Lotsāwa. One of these parts is titled “The ransom ritual of Kha ’bar ma” (Kha ’bar ma’i glud rabs). It opens with the following short introduction: “This is said to have been composed by the master Padma, but the Nyingma school modeled it on a similar Bön tradition. It was adopted by the ”new schools” as well, and it is faultless because it delights the “eight classes of gods and demons of Tibet”. This is followed by a prayer to Kha ’bar ma to serve as a protector. The initial stanza condenses a myth about the Buddha’s taming of the demoness, which, for the sake of coherence, must have preceded the story of the dhāraṇīsūtra:

The loving protector, the compassionate Śākyamuni

Hid [your] eldest son in the small vessel.

Because [you] were tormented, [he] took [him] out, and [you] offered [him your] essence of life [in response].

Great Khabar[ma], stern in command, to the deed—BHYOḤ!82

This myth is presented in more detail in a later text, composed by a Geluk author, Blo bzang bkra shis rnam rgyal (late 17th or early 18th century), a disciple of the Mongolian lama Dzaya Pandita Blo bzang ’phrin las (1642–1708/1715). It depicts her as a major, almost universal demoness, the one that can be only tamed by the Buddha himself.83

Blo bzang bkra shis rnam rgyal’s (later?) contemporary, the Third Gzims ’og, Bstan ’dzin phrin las (b. 18th century) of the Sakya school, opened his text on the ritual of the great torma offering to Kha ’bar ma with a brief account of Bari Lotsāwa’s quarrel with Bhavyarāja and Vajrāsana’s intervention, whose instructions ultimately enabled Bari to bring about Bhavyarāja’s death. This account is supplied with a notable remark:

From that point on, it [the great torma offering] became renowned as a great source of blessings and spread widely throughout the upper and lower regions [of Tibet]. However, over time, because of the weak mental powers of people, it came to be seen as difficult to practice and gradually fell into obscurity.84

It is tempting to follow this passage and believe that the Kha ’bar ma ritual became forgotten or almost forgotten in Tibet, as the texts we have analyzed (primarily those available in the BDRC online library) do not reference authors who lived between the fourteenth and late sixteenth centuries. Perhaps it was thanks to Tāranātha that the ritual returned to wide circulation. He included two short texts on it in his famous collection of sādhanas, Yi dam rgya mtsho’i sgrub thabs rin chen ’byung gnas,85 and also compiled the separate long text quoted above. There are texts on various types of this practice composed by authors from all the major schools of Tibetan Buddhism starting from the second half of the 17th century. Perhaps the earliest of these is by Mkhas grub chen po ka+rma chags med (1613–1678)86 from the Kagyü school, which has not been mentioned so far. And we can be confident that the torma rituals connected with Kha ’bar ma have not been forgotten by Tibetan Buddhists since then.87

We will not delve into the various types and technical details of this ritual, as it is a lengthy topic. Instead, we will limit ourselves to basic observations. In the earliest texts, five aims for the torma offering to Kha ’bar ma are delineated, such as in XT-72:

(1) “Deceiving death by offering the torma as a ransom effigy” (gtor ma mi glud du gtang ba’i sgo nas ’chi ba bslu ba). (2) “Reversing external curses” (pha rol gyi byad bzlog pa). (3) “Killing hostile demons using a karttṛka chopper” (gri gug la brten te dgra bgegs gsad pa). (4) “Destroying the enemies’ citadel by throwing cursed water [towards it]” (chu zor ’phang ba la brten te gzhan gyi mkhar gzhig pa). (5) “The constant gathering of accumulations” (rgyun du tshogs bsag pa).88 The first three aims directly correspond to the two legendary stories: that of Nanda/Ānanda, who escaped death with the Buddha’s help, and that of Bari Lotsāwa, who avoided death, reversed Bhavyarāja’s magic, and caused his death by following Vajrāsana’s instruction. Deceiving death is the most prominent element in Ngog Lotsāwa’s text, illustrated by the story of his own son’s recovery after the “father respectfully beseeched the goddess Black Flaming Mouth for help”.89 In later texts, this application of the ritual becomes predominant.

The above-quoted fragment specifies the color of Kha ’bar ma’s body as black, which may be the earliest indication of this kind. Later texts describe three forms of the demoness: white, black, and multicolored they usually capitalize the colors of forms of Tantric deities, but it’s not something essential for me. These forms are sometimes assigned different functions, with the white form responsible for deceiving death, the black for averting harm from enemies and obstructors, such as sorcery, spells, and curses, and the multicolored for averting slander and disputes.90 However, as we can see, Ngog Lotsāwa’s text does not follow this scheme.

Texts prescribing the visualization proceedings typically do not focus much on Kha ’bar ma herself but emphasize the visualization of the maṇḍala, with Buddha Śākyamuni in the center91 and his fourteen disciples around him,92 with the yogīs themselves visualized as their yidams. The potentially malevolent demons, led by Kha ’bar ma, are briefly invoked to keep their pledge, take the torma as ransom, and depart. However, in some instances, Kha ’bar ma is described in more detail, and invocations to her can be relatively extensive, as seen in the versified text composed by the Second Dre’u lhas grub dbang, Kun dga’ mi ’gyur rdo rje (1721–1769) from the Drukpa subschool of Kagyü:

[The yogī] visualizes himself clearly as the [yidam] goddess./From [one’s] heart, light rays radiate/Like shooting stars, summoning [demons]/From the cities of the nine subterranean realms,/The northeastern abodes of mamos, and other places in the ten directions./[They are led by] the great Goddess of Death,/Kha ’bar ma, the mamo of action,/Who has a dark, naked, emaciated body,/With a gaping mouth, bared fangs, and outstretched hands,/With withered hair and pendulous breasts./[She is] surrounded by a retinue of life-stealing yamas/And mamos of a thousand diseases,/Leading the eight classes of spirits./[They] are summoned helplessly and dissolve/Into the form [visualized by the yogī] in front of oneself,/Becoming inseparable from it./The [chief] mamo [and her retinue] appear clearly in actuality. <…>

Listen, [oh] Goddess of Death, Chief of the mamos!/Abandon your evil and wrathful thoughts!/This person is under the protection of me, the goddess,/So from now on,/Do not harm them!/This ransom torma, complete with all the sense pleasures,/Is made from the finest substances./[I] dedicate it as a ransom for this person./This ransom, more pleasing than a human, is for you./Depart to the great city of the pretas!/If you still disobey [my] command,/By the power of my compassion,/With the sharp and fierce pikes of fire garlands,/I will reduce your body to dust!/So do not transgress your oath,/But quickly return!93

This text portrays Kha ’bar ma as a protective deity of the lowest class, i.e., as a demon bound by an oath unwillingly. However, some texts seem to show her more respect. For instance, an interlinear note in XT-72 suggests that Kha ’bar ma was Vajrāsana’s personal Dharmapāla, and the same could probably be said about Ngok Lotsāwa. An impressive example of this can be found in Tāranātha’s previously mentioned invocation to Kha ’bar ma, which is written in a notably solemn style. For a student of Tibetan Buddhism, it is no surprise that the same original image can lead to multiple diverse manifestations.

3.2. Evidence in Sources Relating to Bön Tradition

Despite significant progress in the field of Bön studies, the ritual literature of this tradition, in all its diversity, remains understudied and largely inaccessible for keyword searches in online Tibetan text libraries such as BDRC and rKTs. As a result, we were only able to cover a limited range of sources. Nevertheless, in three academic publications, several instances of interest for this paper have been identified.

One is found in Samten Karmey’s publication of a large painting94 that reflects Dzogchen theories about “the vision of the world and what is beyond the world of the Dzogchen practitioner, who is represented by the central seated figure as he understands the four wheels”.95 The painting is abundantly supplied with inscribed annotations. At the bottom of the painting, we find the representation of six realms of sentient beings that Bön borrowed from Buddhism. The realm of pretas consists of three sections, with the upper one occupied by “Kha ’bar ma, the queen of the pretas who have outer (physical) obstacles; [thus, they are] unable to eat”.96 The picture portrays Kha ’bar ma as an ugly demoness of pale grey color, with two tongues of fire emanating to the left and right from her mouth, thin arms, and pendulous breasts. She is surrounded by pretas of the aforementioned category. The lower part of her body is not depicted; instead, the image shows a preta in profile and a priest clad in a yellow and red robe, wearing a hat with red borders and a yellow peak. He stretches his right (?) hand towards the preta’s mouth. Next to them, bowls with offerings are depicted. We can only speculate as to whether this picture refers to the Buddhist legend or if there is a Bön version of it. However, we can be confident that the concept of offering to Kha ’bar ma as a ransom remains consistent in this context. It is evident that her image was significant enough to be prominently reflected in the depiction of the world by Bön practitioners.

A very similar instance is found in a 14th-century Bön text on cosmography, analyzed by Katsumi Mimaki (2000). In this text, Kha ’bar ma and another figure named Kha ’dra ma are listed as representatives of the “gods of pretas” (Mimaki 2000, p. 111). Both instances belong to the tradition “of the so-called ’gyur bon (transformed Bön), which seems to have borrowed and adopted Buddhistic ideas beginning from the eleventh century in order to form its own doctrine” (Ibid., p. 89). Even though this may be the case, the acceptance of Kha ’bar ma as the primary representative of the realm of pretas by this tradition may suggest its widespread recognition in Tibet, rooted in deeper layers of local culture.

Just as we observed that Indian Buddhist material shows diversity in the bearers of the name Jvālāmukha/Jvālāmukhī, Bön texts studied by René de Nebesky-Wojkowitz (1923–1959) also demonstrate that the name Kha ’bar ma (Kha la me ’bar) can refer to various minor demonic beings such as bdud mo or gshin rje.97 A similar example concerning a minor deity from the retinue of the major Buddhist goddess Dpal ldan lha mo is attested in texts from the Nyingma and Kagyü schools (Ibid., p. 37). The close contacts between Bön and Nyingma suggest possible mutual influence, and the imagery of a demon(ess) with a flaming mouth might have developed locally, independently of the cult “invited” from abroad.98

3.3. Tibetan Textual Sources, Sacred Geography, and Pilgrimage Routes

The earliest of the canonical texts mentioned above is Suhṛllekha, attributed to Nāgārjuna. It exists in three Chinese translations: one by Guṇavarman (dated 431), another by Saṅghavarman (434), and a third by Yijing (673). These translations exhibit significant discrepancies not only among themselves but also with the later Tibetan translation made by Sarvajñadeva and Dpal brtsegs we should capitalize only the first part of this name. Recently, the Sanskrit version of the text was fortuitously discovered in Tibet and published by Dngos grub tshe ring (2020). The passage of interest, which describes various types of pretas, contains the following line that can be translated from Sanskrit as follows: Some have faces that [flame like] firebrands at night and eat moths falling into their blazing mouths.99 The first and third Chinese translations follow this imagery,100 so we may be certain that no later than the early 5th century, the Buddhist literature included an image of ghost spirits whose faces blazed, attracting night moths to burn and provide them with some nourishment.

It is tempting to speculate that the origin of this image must have a natural explanation, and there are not many phenomena that could be proposed as such. “Eternal” fire rising from the earth, caused by the burning of natural gas, is certainly among them, as it could produce the described effect. At least three locations in the Himalayan region have incorporated such phenomena into religious cults. In addition to the Jvālāmukhī temple of Kangra, these include the fire temples in Dullu and Muktināth, both in Nepalese territory.101 The latter, located in Mustang with a predominantly Tibetan-speaking population, sees equal worship of the local fire temple by both Hindu and Buddhist devotees.

Tibetans call this place Chu mig brgya rtsa, meaning “a hundred-odd springs”. For them, this site combines fire with water, as emphasized in one of the texts (in David Snellgrove’s translation): “[T]he precious fire that burns from water has the way of burning changeless and unceasing from the hidden parts of Samvara Male and Female. This precious fire that burns from stone serves as a symbol of the light-giving butter-lamp… These streams of water, one hundred and eight are a place which the Eighty-(four) Great Magicians protect, each one of them” (Snellgrove 1979, pp. 108–9). Buddhists, apparently, preferred to remark on the combination of fire, water and earth (from where fire emerges) rather than one element solely. Thus, they interpreted the fire as produced by one of the major Tantric gods, Cakrasaṃvara, and his consort, while the water was associated with the renowned group of great Indian Tantric masters (mahāsiddhas).

Moreover, as Snellgrove observes, the springs “are watched over by a Serpent-Divinity known as Gawo Jogpa (dGa’-bo ’Jogs-pa), who is identified by Hindus as Vishnu or Narayan” (Ibid., pp. 81–82). While Hindus regard Muktināth as one of the 51 Śakti Pīṭhas, Buddhists include it in their own list of Tantric pīṭhas.102 This is a clear example of the sacred “contest” over territory between major Hindu and Buddhist deities. In this particular case, it seems to have resulted in a “truce”.103

The fact that the Chu mig brgya rtsa area in Muktināth has no connection to Kha ’bar ma may suggest a non-Buddhist origin for her image. Intuitively, if Kha ’bar ma had been conceived within a Buddhist framework, her association with fire worship would likely have linked both sacred sites dedicated to this element. This is further supported by her identification as the queen of pretas, as found in one of the two variants of the canonical dhāraṇī. The other variant does not classify the character as female. It seems likely that the general concept of face-blazing pretas underwent certain developments before being personified as their queen, who then adopted this name as her own.

The circumstances of her threatening appearance to Ānanda (or Nanda) are described as occurring in a secluded place, but not in a mountainous area, since the Buddha, according to both variants of the text, was staying in Kapilavastu.104 This implies that, if the present hypothesis is correct, the deity’s image must have already been detached from its possible origin. Such a transition is entirely plausible, given that the first Chinese translation of the dhāraṇī is dated between 700 and 704, which is much later than the composition of the Suhṛllekha the Sanskrit titles should be italicized here and below. The image had ample time to undergo the necessary transformation.

As mentioned above, the Hindu Tantric text Kubjikāmatantra, which presents a list of 24 sacred districts, assigns Jvālāmukhī to the place called Jayantī. At that point, we deliberately kept the intrigue, omitting that this location is in Kangra105 and that the Devī’s male partner is identified as none other than… Mahāpreta! Surprisingly, the corresponding Buddhist list in the Śrīḍākārṇavamahāyoginītantrarājanāma—which, as Sugiki convincingly demonstrated, was based on the Kubjikāmatantra—alters her image. It mixes her up with the preceding figure, losing her association with Jayantī and assigning her a different male partner. Whether this was an unintentional mistake, as Sugiki suggested (Sugiki 2003, pp. 65–67), or a deliberate attempt by Buddhists to dissociate her from Kangra and the preta realm, remains open to speculation. The alternative pair of the devī and kṣetrapāla, Mahālakṣmī and Agnimukha, as noted above, are directly associated with fire.

Significantly, not only Muktināth but also the temple of Jvālāmukhī and the town of Jalandhara were once part of an important pilgrimage circuit for Tibetans. In Giuseppe Tucci’s book (Tucci 1940) and Lobsang Shastri’s dedicated paper (L. Shastri 2009), sufficient evidence proving this fact is gathered. The most notable pieces of evidence pertain to the 13th century. The area was associated with flourishing of Buddhism in the 1st Millennium CE, and was part of the Kushan Empire, whose rulers were great patrons of this religion.106 Later on, Jālandhara was cast as one of twenty-four pilgrimage sites recommended by the Cakrasaṃvaratantra. The list of the eighty-four mahāsiddhas (great Buddhist tantriks) includes Jalandharapa.107 According to his hagiography, “[a]fter undertaking intense meditation at Jalandhara,108 he was blessed with the vision of Vajrayogini and attained enlightenment. He was named after the holy site where he meditated” (L. Shastri 2009, p. 8). Tibetan texts also identify this figure with Jālandharanāth of the Nāth tradition (ibid., pp. 8–9). This intersection alone might be seen as an important sign of the extremely intricate nature of Hindu–Buddhist (Tantric) relations in this region.

Tibetan geographic descriptions and travel records of this part of India do not indicate any association between Kha ’bar ma and Jvālāmukhī, whose name is usually rendered in phonetic transcription in these texts. The most interesting information is probably provided by Bu ston rin chen grub (1290–1364) in his text on the history of the cult of Cakrasaṃvara (Bde mchog chos ’byung), present in North India at the time and closely related to the worship of yoginīs in Hindu Tantras (Gray 2007, pp. 3–5). Bu ston mentions “Kṣetrapāla goddess Jvālāmukhī (zhing skyong lha mo Dzwa la mu khi), whose [image] miraculously formed of stone and was lying face down”.109 Bu ston also refers to a temple (gtsug lag khang) built on this spot but does not specify whether it was Buddhist or Hindu. Not far, there was a place with “one hundred caves for the siddhas, one hundred springs, and one hundred tree-root [dwelling yogīs]” (the phrase ”one hundred springs” inevitably recalls the Tibetan name of Muktināth), and there was “a waterfall where tīrthikas (Hindus) perform [ritual] baths” (Bu ston rin chen grub, f. 28a = p. 55). It is not clear where Bu ston obtained the information about the stone figure, but his identification of the goddess as a Kṣetrapāla (protector of the ground) suggests that Jvālāmukhī was indeed assigned this function locally—perhaps by both Buddhist and Hindu Tantrics.

The idea of joint worship of the site is supported by the description found in the biography of Rgod tshang pa mgon po rdo rje (1189–1258), an eminent master of the Drukpa Kagyü tradition.110 This passage may be quoted in full (in Tucci’s translation):

[T]here is a great temple called Dsa va111 la mu gi in which both believers and unbelievers112 offer their worship. Thirty villages113 are in charge of this temple. The very day the pilgrim114 arrived and went to Dsva la mu khe, in the night there were in the temple sixty or seventy girls, all virgins, beautiful and charming, adorned with five kinds of symbols like divine girls, dressed and adorned with various ornaments such as the jeweled crown. Some of them carried in their hands flowers and other things for the pūjā such as incense, etc. The girls having covered their head with a cotton veil, entered the temple. The pilgrim followed them, but a man of low class holding the door-bolt did not allow him to go farther; but he, without hesitation, pushed the door and went in. The other stood up but was unable to hit him, (the pilgrim) went inside. One of the principal ladies said, ”Sit down here, all these are ḍākinīs”. Then that lady began to sing some songs. The other girls sang as if they were either the sixteen mystic wisdoms (vidyā) or the twenty goddesses, made the offerings with the various ingredients of the pūjā such as flowers, incense, etc. They sang songs and danced accompanying the dance with gestures of the hands.

Another version of the same master’s biography adds that a ḍākinī named ”The Bee-Faced” was the chief deity (gtso mo) of the place, and Rgod tshang pa obtained some teaching from her (Rgyal thang pa, p. 386). L. Shastri (2009, p. 15) identifies this ḍākinī with Jvālāmukhī, although their names are quite different and the title gtso mo is somewhat distinct from that of a Kṣetrapāla (zhing skyong). One might speculate that the bee, with its sting, and fire, with its sharp tongues, could be symbolically linked. It is also notable that the Tibetan words sbrang ma (‘bee’) and ’bar ma (see the next paragraph) share similarities, making it plausible that a linguistic confusion could have led from one to the other. However, it is safer to leave this issue unresolved, hoping that it may be clarified over time.

Later texts, such as Panchen Lama Dpal ldan ye shes’s (1738–1780) “Guidebook to Shambhala” (Sham bha la’i lam yig) and Btsan po no mon han’s (1789–1839) “Geography of the World” (’Dzam gling rgyas bshad), do not add anything substantial. However, the former provides the Tibetan name for the Jvālāmukhī site—rdo le me ’bar (“Fire Blaze on Stones”)—while the latter refers to the goddess residing there as ’Bar ma (“Blazing”) (L. Shastri 2009, p. 15). This represents an interesting shift from the phonetic transcription of the name to its translation into Tibetan, though without any association with Kha ’bar ma. These later texts suggests that by the 18th century, joint worship at the site had likely ceased.

The Muktināth material reminds us of another sacred contestant Buddhism had to face in the Himalayan region, namely Bön. Charles Ramble’s study of a Tibetan harvest festival in Muktināth detected traces of old enmity between Buddhism and Bön in this locality, although members of both communities now show mutual respect for each other’s cults. The Bön community is now significantly smaller, and Buddhists control all the Muktināth shrines apart from the Hindu one (Ramble 1987, pp. 238–42).

It is highly likely that this was not always the case, and that Bön cults could have actually preceded the spread of Buddhism there. It is hard to believe that this religion could remain unimpressed by the natural fires emerging from stone. Although this issue requires further study, perhaps through fieldwork, we think it is entirely possible that the image of a deity with a face radiating a blaze or fire could have been integral to the beliefs of local people since ancient times. This might explain the application of the name Kha ’bar ma (or its variants) to various types of demonic beings, as reflected in both Buddhist literature (including “old tantras” that may have been produced in Tibet rather than India) and Bön literature.

Charles Ramble kindly brought to our attention a short Bön text, translated by him, which contains a myth about the killing of a vampire with a firebrand. The text incorporates all the natural elements necessary to potentially give rise to a fire cult similar to the one found in Muktināth: “At the edge of the lake, where the mountain rose up, was a white rock. On the white rock there grew a red plant; the white rock and the red plant came together, and the white fire of the mountain above came into being. The Lady of the Rising Mountain blew on the fire, and the land of the Rising Mountain was brightened. This is how the killing of the vampires by the bright fire came into being” (Ramble 2019). We may expect that, with more Bön texts being introduced to academia, we may find more examples relevant to the ancient fire cults in Tibet.

The religious “career” of Kha ’bar ma in Tibetan Buddhism from the late 11th century onward, though depicted in various forms and ritual details, is quite typical of beings categorized as mundane protectors—usually demons bound by vows to serve Buddhists. Her image, however, possesses a peculiar feature—some texts portray her as a kind of primordial demoness, whose taming could only be accomplished by the Buddha himself. In this regard, she resembles the famous rākṣasī from the apocryphal “Pillar Testament”, depicted as a monstrous being covering the entire land of Tibet. To subdue her, Buddhist temples had to be constructed at key locations piercing the vital points of her body. Jacob Dalton, in his analysis of this story, likened it to Purāṇic dismemberment myths (Dalton 2011, pp. 113–18). This bears a clear parallel to the Hindu myth of Jālandhara’s subjugation (see Section 2.1, above). Although the rākṣasī myth may be imported from India, its thematic resonance with regional dismemberment motifs is also congruent with the wider pattern of Himalayan cultural exchanges. Strictly speaking, we cannot be certain that the influence always flowed from south to north and never in the opposite direction.

4. Conclusions

Our study was instigated by a simple question: Is there any link between the Hindu goddess Jvālāmukhī, worshipped in the Kangra Valley, and the Buddhist demoness Kha ’bar ma, as depicted in Tibetan texts? While canonical dhāraṇī-sūtras suggested Indic provenance for the figure of Kha ‘bar ma, the complexity and depth of her Tibetan representations—emerging from at least the 11th century—hint at a potential local heritage, perhaps shaped by Bön or regional folk traditions. This inquiry was grounded in a broader recognition of the cultural proximity between Himalayan Indians and Tibetans, which provided the theoretical background for our comparative investigation. We approached the question open to the possibility that no substantial link would emerge—recognizing that a negative result would still hold scholarly value. Nonetheless, our findings revealed several points of convergence that, considered collectively, suggest the correspondence of the Hindu and Tibetan names may not be a mere coincidence:

• Sectarian overlaps and spatial organization.
  The goddesses worshipped in the Jālandhara Pīṭha—primarily Vajreśvarī, Tārā, Cāmuṇḍā, and Chinnamastā/Cintāpụrṇī—bear clear Buddhist associations. Spatially, Vajreśvarī and her surrounding goddesses evoke the layout of a classical maṇḍala, recalling the five tathāgatas with Vairocana at the center, whose divine consort is Vajradhātvīśvarī.
• Tantric pairings and symbolic functions.
  The pairing of Jvālāmukhī with Mahāpreta in the Kubjikāmatantra as field protectors (kṣetrapālas) for the Kangra region is striking—especially in light of Jvālāmukhī being identified as a pretī in Buddhist literature.
• Parallel creation myths and ritual motifs.
  The narrative of Kha ’bar ma’s birth in an “old tantra” of the Nyingma school—emerging from Rudra’s blood—parallels Hindu myths in which fierce goddesses arise from drops of divine essence during moments of cosmic upheaval. This echoes the mythic origins of the Katoch rulers, born from the goddess’s perspiration during her battle with Jālandhara. Moreover, the motif of tongue offerings, present in the Kha ‘bar ma ritual, resonates with reported practices at both the Vajreśvarī and Jvālāmukhī temples.
• Subordinate status and dual nature.
  Both Kha ’bar ma and Jvālāmukhī occupied liminal positions within their respective traditions as minor, subordinate figures who nonetheless embody fierce, transformative energy (pretīs, yoginīs). The former retained this position to the present, the latter became important in Kangra over time. Their dual role as wrathful and protective deities further aligns within the broader category of liminal goddesses within their respective religious systems.
• Sacred geography and shared custodianship.
  While Muktināth remains an important center of Tibetan Buddhism, Jālandhara also attracts Buddhist pilgrims as one of the sacred sites of the Cakrasaṃvara cult. Several Tibetan texts from the 13th century associate the temple of Jvālāmukhī with the cult, and a number of these suggest joint worship of the site by Buddhists and Śaivites.
• Elemental cults and natural phenomena.
  Muktināth and Jvālāmukhī both center around natural flames—a fact that may reflect ancient fire cults grounded in elemental worship. While Hindu traditions emphasize fire as a solitary element, Buddhist sources often present it in conjunction with water and earth. The presence of natural flame over water in the Nāth shrine at Jvālāmukhī suggests possible cross-adaptations from earlier Buddhist or folk traditions that are consistent with the Nāth appropriation of Buddhist sites and relics in South India (Mallinson 2019).
• Bön as a residual presence.
  Bön remains a “dark horse” in this investigation. Nevertheless, scattered references to Kha ‘bar ma in Bön iconography and cosmology imply a longstanding familiarity with the demoness of the flaming mouth. Certain non-canonical Buddhist narratives portray her as a primordial force that had to be subdued by the Buddha himself—reminiscent of Purāṇic dismemberment myths and possibly reflecting deeply rooted Himalayan religious imaginaries.

Certainly, none of these observations seems conclusive on its own. However, taken together—and considered within the context of centuries of religious exchange, shared history, and the geographic proximity of Tibetan and West Himalayan societies—they suggest a compelling, if still hypothetical, link between Jvālāmukhī and Kha ‘bar ma. Thus, while this study does not claim to offer definitive proof of this connection, its findings are sufficiently robust to justify its continued exploration through both textual analysis and field research.