Alternate Narratives for the Tamil Yoginis: Reconsidering the ‘Kanchi Yoginis’ Past, Present, and Future

Abstract

This essay revisits a group of stone goddesses that once shared a temple in southern India, together with the god Shiva and perhaps other deities. Considering the different paths sacred sculptures in India take after becoming separated from their original temple contexts suggests that there were multiple possible histories for these works. The authors reveal a newly discovered goddess from the group and reconsider the significance of the works, including the original temple and the deities it enshrined. Finally, they propose the possibility of bringing these sculptures back together in the context of an exhibition.

1. Introduction

Two stone sculptures of goddesses in the collections of the Detroit Institute of Arts (DIA) and the Smithsonian National Museum of Asian Art (NMAA) (Figure 1 and Figure 2) once shared a temple in northern Tamil Nadu, together with other similar goddesses, the god Shiva, and perhaps other deities.1 Slightly larger than life-sized, these fierce, tantric goddesses combine features identified with an idealized female beauty together with elements that signal danger. They have voluptuous bodies and intricately carved jewelry, but they also wear adornments made of writhing snakes and they hold threatening objects, such as skull cups and weapons. When they were part of an active temple, they were probably believed to bestow great powers upon devotees but to be dangerous to people not initiated into their religious traditions (Dehejia 1986; Kaimal 2012; Hatley 2013, 2014, 2019). It is this embodiment of both alluring and threatening qualities, as well as the large number of goddesses from the group—at least thirteen are known—that identifies them as yoginis.

Stylistic comparisons with stone and bronze sculptures from Tamil Nadu, and relative dating of yogini temples elsewhere in India, suggest that these yoginis were carved sometime between the late ninth and mid-tenth century.2 Their story then continues in the twentieth century, when they were found at an unspecified site in 1925–26 by the French archaeologist Gabriel Jouveau-Dubreuil and his Indian associate, N. Tangavelou Pillai.3 Jouveau-Dubreuil subsequently sent most of the known yoginis to the Paris-based art dealer C. T. Loo, whose gallery in the ensuing decades sold or donated them to various museums and private collectors.4 One yogini remaining in India went to the Government Museum in Madras (Chennai) in 1937. Today, fourteen sculptures from the group have become dispersed across museum collections spanning North America, Western Europe, and India, where each one sits as an art object in relative isolation.5 No trace of their original temple has been located.

In this essay, we revisit these Tamil yoginis and propose alternate narratives for the sculptures that took place during the thousand-plus years between the ninth and twenty-first centuries. In calling this group of sculptures the “Tamil yoginis,” rather than the “Kanchi yoginis,” as they have previously been known, we argue that the sculptures did not originate in the city of Kanchi (Kanchipuram), and we use “Tamil” as a geographical designation—the group constitutes the only known set of yoginis from within the present-day state boundaries of Tamil Nadu, and no other yogini temples are known to exist in the region.6 Situating the yoginis in the changing historical circumstances of northern Tamil Nadu, we reevaluate their original temple, including the deities it enshrined, the materials used to build it, where it stood, and its possible patronage and ritual community. We then address its dismantling and the dispersal of its deities, both before and during the twentieth century. The different paths sacred sculptures in India take after becoming separated from their original contexts suggests that there were multiple possible histories for the yoginis and their companions and that not all the sculptures from the group shared the same fate. A newly discovered goddess from the group—not in a museum but a local Tamil shrine—illustrates one alternate narrative and demonstrates another mode of collecting and reuse for the yoginis. Finally, we explore the possibility of bringing the sculptures now stewarded by museums back together in the context of an exhibition.

2. What Are Yoginis

As two members of a larger group, the yoginis that now reside at the DIA and the NMAA exemplify many of the qualities that characterize yoginis in general and the Tamil set in particular. Both goddesses appear both attractive and threatening—a combination of traits frequently encountered in yoginis’ widely varying iconography—but the DIA yogini is the fiercer of the two. She fixes her face in a menacing expression, staring out with bulging eyes under an intensely furrowed brow and opening her mouth to bare her teeth and fangs. Her unbound hair radiates around her head with a wild energy, signaling her transgression from cultural norms that prefer neater, more controlled coiffures.7 She wears a diadem, necklaces, and bracelets that could be shared with more peaceful goddesses, but where those gentle goddesses would wear jeweled armlets and a sacred thread resting diagonally across the torso, live snakes wind their way around her body, rearing their hooded heads. Other attributes have clear associations with death: the cup she holds in her lower left hand is understood to be made from a human skull, and she sits above a headless corpse, which is carved in low relief onto her pedestal. In her two upper hands, she holds a shield and a club—objects that can both harm and protect (Kaimal 2012, p. 91). Drawing attention to the twofold nature of these weapons, Kaimal also makes a larger point about the character of yoginis themselves: they embody both auspicious and inauspicious qualities (Kaimal 2012, pp. 90–96; Kaimal 2013).

Threatening elements may be comparatively toned down in the NMAA yogini, but they are unmistakably present. In her lower left hand, she holds a skull cup, as do all the yoginis from this group whose lower left hands remain intact. In her two upper hands, she holds a broom made of sacred darbha grass and the type of basket-weave winnower (an agricultural tool used for separating the chaff from grain) that is regularly used as a dustpan by priests and temple attendants while cleaning their sacred spaces.8 Much of the jewelry she wears could be mistaken for that of a gentle goddess or a human queen, but in her ears, she wears a slithering snake and a writhing makara (mythical sea creature). Her hair extends freely to each side, curling at the ends, unable to be contained by the tall, stately crown upon her head. On her crossed ankles are strands of bells that recall the sounds of the yoginis’ descent invoked by the poet Somadevasuri in his Yaśastilaka, composed around the same time as the sculptures’ creation:

The companions of the gods were alarmed by the vibrating resonance of their anklet bells as the Yoginis danced the distance between the sky and the earth. Their dark loose tresses splaying out across the daytime sky, darkened it, and their terrible yet glowing skull crowns stood out among the black tresses like stars at night.9

Yoginis, as their iconography suggests, are extremely powerful.10 As goddesses associated with tantric traditions, their power is in part connected to their overturning of orthodox norms and their embracing of what more mainstream schools of thought considered impure—objects such as the corpse the DIA yogini sits upon, or the skull cups that she and many of her companions hold.11 Yoginis guard tantric teachings against the uninitiated, restricting the powerful knowledge of these traditions to the initiated few. To individuals who hold this esoteric knowledge, and who correctly perform the arduous practices it demands, yoginis may grant victory over one’s enemies or great magical powers, such as the ability to fly or to become very large or very small. However, to individuals who do not follow their practices correctly, yoginis can be lethal.

A yogini does not act alone. These goddesses always occur in groups—most commonly, texts refer to groups of sixty-four yoginis, though they may occur in other numbers too. Extant yogini temples enshrine dozens of goddesses along the inner walls of open-air structures—because yoginis can fly, their temples must be roofless (Figure 3a,b; see below for a further discussion of extant yogini temples). Both within their temples and among known yogini sculptures from different groups, no two are exactly alike. Some are more ferocious, appearing furious and holding objects such as weapons, skulls, or severed heads, while others are more benign. Some yoginis have idealized bodies suggestive of a woman in her sexual prime, while others appear more aged; still others are withered and emaciated. Some yoginis appear human in form, while others have the heads of various animals. Certain religious texts even describe yoginis as shapeshifting beings. Their forms are inherently multiple.

Even with all their iconographical variety, yogini sculptures from a particular group share certain qualities that indicate they belonged together. Because sculptors carved them as sets to be enshrined together within the same temple, yoginis from any given group are made of the same type of stone and are similar in size (though their exact dimensions may vary).12 Having been inserted into niches within their temples’ walls, yogini sculptures are typically steles, and those from the same set share a common format—each of the Tamil yogini figures is attached to a solid, curved backdrop. Bodily positions are also usually consistent among yoginis of the same group. Each member of the group of Tamil yoginis sits with one ankle crossed over the other, their bent legs projecting into the surrounding space.

As in the Tamil group, each yogini within a set shares certain conventions with her companions but has her own individualized features and attributes. Rarely does this individualized iconography repeat exactly, within a group or between different groups of yoginis. In at least some cases, yoginis were assigned individual names, too. For instance, in the yogini temple at Bheraghat, Madhya Pradesh, in central India, names are inscribed into the pedestals of each sculpture. However, as Vidya Dehejia demonstrates in her foundational book on yogini temples, these names are not consistently repeated at other temples, nor do they correspond exactly to the various lists of yogini names that appear in religious texts (Dehejia 1986, pp. 5–7). Although no names are inscribed on the Tamil yogini sculptures, it is possible that visitors to their temple knew each goddess by an individual name or epithet. Yoginis’ names are individualized and multiple, as varied as their visual iconography.

3. The Yogini Temple in Tamil Nadu and Its Deities

Although no trace remains of the Tamil yoginis’ temple, extant yogini temples in India, further north, begin to suggest what it looked like. Dehejia located seven surviving yogini temples, all in the present-day states of Odisha and Madhya Pradesh, though additional sets of yogini sculptures from northern and central India, as well as textual references, suggest that yogini worship was more widespread (Dehejia 1986, pp. 77–184; see pp. 175–84 for the sculptures we discuss in this article, which constitute the only known evidence of a yogini temple in present-day Tamil Nadu). All of the known yogini temples in India were constructed as roofless structures open to the sky, and all feature multiple shrines for yoginis lining the periphery of their interior walls. Most of the yogini sculptures remain in the temples at Hirapur (ca. mid-ninth–early tenth century), Ranipur-Jharial (ca. early tenth century), and Bheraghat (ca. late tenth–early eleventh century) (Figure 4) (see Dehejia 1986, pp. 91–114, 125–40). Today, these temples are commonly referred to as “Chausath Yogini”—or “Sixty-four yoginis”—regardless of the actual number of yoginis or yogini shrines they contain. While sixty-four is the most common number for a group of yoginis, some temples only accommodate forty-two, and the temple at Bheraghat has eighty-one peripheral shrines (Dehejia 1986).

Because the number of yoginis in temples and texts varies so widely, we cannot ascertain how many yoginis were originally part of the temple in Tamil Nadu. Kaimal (2012) has published twelve from the set, eleven of which are in museums in North America, Western Europe, and India, and one whose location is now unknown.13 If there were originally sixty-four, then a great many are unaccounted for. Schmid (2013, p. 149) has proposed that the group originally contained only twelve, but no other known yogini temple enshrined so few goddesses. As we will discuss below, a recently discovered yogini from the Tamil set establishes that there were at least thirteen—and there were likely even more.

In addition to the variation in numbers of yoginis, yogini temples sometimes also included a related set of goddesses, called matrikas (“mothers”).14 At Bheraghat, sculptures of the matrikas predate the yoginis, but they too are enshrined along the inner periphery of the temple. Whether these earlier matrikas were installed within the Bheraghat yogini temple at the time of its consecration or were brought there later remains an open question.15 Similarly, three matrika sculptures that were found by Jouveau-Dubreuil and Tangavelou together with the Tamil yoginis may or may not have been enshrined alongside the yoginis in their original temple (Figure 5; see also Kaimal 2012, figs. 14, 17 and 18).

At 70–75 cm in height, these matrika sculptures are much smaller than the Tamil yoginis, and they are backed by an open stele framework, with much of their bodies carved in the round, in contrast to the yoginis’ solid stone backdrop. Kaimal argues that these differences in format do not preclude the matrikas’ inclusion within the Tamil yogini temple and suggests that their cells may have been smaller than those of the yoginis (Kaimal 2012, pp. 116–17). While such a scenario is certainly possible, we think it is more likely that the matrikas were brought together with some of the yoginis only after their respective temples had fallen out of use.16 These smaller goddesses probably once belonged to a set of seven matrikas enshrined in the southwestern part of a temple dedicated to Shiva. Separate shrines for matrikas and other auxiliary deities were (and still are) a standard part of Shiva temples in southern India.

Shiva finds a place in many yogini temples, too, often within a shrine standing at the center of the open-air structure.17 Quite possibly, the Tamil yogini temple also had a central shrine for Shiva.18 Indeed, a Shiva-Vinadhara now at the Museum of Fine Arts, Boston, was found together with another one of the yogini sculptures in 1926 (Figure 6). Carved from the same stone as the yoginis, on the same scale, and with a solid, curved stele backdrop, his original presence with the yoginis is all but certain. Whether or not he occupied a central shrine remains a matter of speculation. At 122 cm in height, he is shorter than the tallest yogini.19 One might expect a central deity’s image to be larger than the surrounding figures, as is the case in the yogini temple at Ranipur-Jharial, Odisha, where the original sculpture of Shiva remains (see Dehejia 1986, pp. 103–114).

R. Nagaswamy (2006, pp. 91–94) and Charlotte Schmid (2013, pp. 140–49) have both proposed a different sculpture for the Tamil yogini temple’s primary god. Once located in a village called Tirumalaicheri (also called Melcheri), near the town of Kaveripakkam, this sculpture of Shiva has three faces, the central one in fierce aspect, and like the Tamil yoginis his body projects from a curved stele. Stylistic elements, such as facial features and jewelry, are also consistent with the yoginis’ visual vocabulary (Schmid 2013, p. 142). The sculpture’s size—some 140 cm in height—is just slightly greater than that of the largest known yogini from the group, which further suggests that this sculpture could have been the primary god enshrined at the center of the temple (Schmid 2013, p. 142). Intriguing as these arguments are, we cannot be certain of this Shiva’s association with the yoginis because firsthand study of the sculpture is not presently possible. According to Nagaswamy (2006, p. 92) and Schmid (2013, p. 142), the three-faced Shiva was previously kept at the Government Museum in Chennai. However, during our visits in 2019 and 2020, it was no longer there—we were told it had been moved to an unspecified location.

However many Tamil yoginis there were, whether or not they sat alongside the smaller matrikas, and whether or not the three-faced Shiva from Tirumalaicheri had a presence in their temple, we can be fairly certain that the goddesses were enshrined around the inner periphery of a roofless structure, facing the building’s interior. The temple may have been circular in form, as most extant yogini temples are, or it could have been rectangular, like the yogini temple at Khajuraho (see Dehejia 1986, pp. 115–17). Its exterior walls were probably relatively plain. Although some yogini temples feature carved imagery and moldings that share a vocabulary with the architecture of Brahmanical, Jain, and Buddhist temples of their respective regions, a starker, fortress-like appearance is more common.20 The yogini temple at Hirapur features austere architecture but enshrines additional deities around its exterior and doorway (see Figure 3), as well as on the exterior of its central shrine.

Yogini temples across the Indian subcontinent supply possible morphologies for the Tamil yogini temple’s architectural structure. However, local practices within northern Tamil Nadu are more relevant when considering how it was built and with what materials. Looking at texts and comparable temples, we propose that the Tamil yogini temple was constructed with brick walls above stone foundations.

We know from inscriptional references and surviving examples that brick was a common building material for temples in Tamil Nadu throughout the first millennium (Stein 2021, pp. 47–51). Temples were frequently constructed of brick walls on top of stone basement moldings. Many—but not all—of these temples were rebuilt in stone, either in part or in full, starting in the late tenth century (some were upgraded to stone earlier). Although many brick temples from the period of our yoginis have either crumbled or have been converted into stone, some still survive in northern Tamil Nadu. Two examples are the Cuntaravarāta Perumāḷ and Kailāsanātha temples at Uttiramerur, about 25 km south of the city of Kanchi (Figure 7). Both feature brick walls above granite foundations and both date to the ninth century. The stucco that covers these temples’ walls makes the brick difficult to discern in photographs, but our site visits confirm that brick lies beneath.

Brick is far less durable than stone, and without the kind of upkeep the temples at Uttiramerur have enjoyed, a brick structure is more likely to fall into disrepair than a stone one, especially over a span of centuries. If the yogini temple had walls of brick, it is likely they would not have survived. Like the temples at Uttiramerur, the yogini temple probably also had stone foundations to support such weighty sculptures. However, it is not altogether surprising that they are unaccounted for today. After the yogini temple fell out of frequent use, fragments of the stone foundation may have become buried deep in the ground, reused in other structures, or otherwise lost over time. It is quite possible that they were dismantled and repurposed for constructing or enlarging the massive tank bund in Kaveripakkam, the once-prominent village that is the most likely site for the yogini temple, as we will discuss below. Kaveripakkam’s tank bund was fortified in the mid-twentieth century, in part using sculptures and architectural fragments made of stone (Sivaramamurti 1955, p. 7). Today, some of these fragments are still visible, and some may have supported the yoginis.

4. Siting the Yogini Temple

There has never been doubt that the yogini temple was situated in northern Tamil Nadu, in or more likely near the city of Kanchi. However, the precise location has remained unconfirmed. Correspondence from Jouveau-Dubreuil and Tangavelou to C. T. Loo and Musée Guimet staff (which we will discuss in detail below) establishes that by 1925, when Jouveau-Dubreuil and Tangavelou located them, the yoginis no longer inhabited their original temple. These circumstances make tracing the location of this temple even more difficult.

The closest stylistic comparisons for the Tamil yoginis can be found in ninth-century and some tenth-century temples and fragments from villages near Kanchi, the city most frequently associated with the Tamil yoginis in previous publications. Kanchi is located approximately 75 km southwest of Tamil Nadu’s capital, Chennai (formerly Madras). It once served as a royal capital of two major dynasties in Tamil Nadu, the Pallavas (ca. third–ninth century) and then the Cholas (ca. ninth–thirteenth century), and it remains a living city replete with densely packed residential enclaves, ponds, and temples, many of which were built during the reigns of these two dynasties (Stein 2021).

Kanchi is famous for its temples, but none of its sculptures or architectural relief carvings entirely correspond in style to the Tamil yoginis. Instead, closer comparisons stand in the villages of Kaveripakkam (25 km west of Kanchi) (Figure 8) and Tiruttani (45 km north of Kanchi), and as far afield as the region of the Kaveri River delta (more than 250 km south). At Kilaiyur-Melappaluvur, for example, the door guardians of the late ninth-century Agastyēśvara temple have facial features that are remarkably similar to the Tamil yoginis (Figure 9a,b). Considering the mobility of artists in premodern South Asia, it is even tempting to speculate that the Tamil yoginis and the Kilaiyur guardians may have come from a common workshop.21 Although the yogini sculptures have previously been dated to the early to mid-tenth century, the stylistic parallels between the Tamil yoginis and the materials from Kaveripakkam, Tiruttani, and Kilaiyur suggest that we should not rule out the possibility that they date more specifically within the late ninth century.22 We prefer a date range of late ninth to mid-tenth century.

Among all the sites that offer stylistic parallels to the Tamil yoginis, the most likely place of origin for the set is Kaveripakkam, or somewhere in its close vicinity. We are not the first to suggest this—Vidya Dehejia (1986, pp. 181–82), R. Nagaswamy (2006, p. 94), and Charlotte Schmid (2013, pp. 138–50) have all attributed the yoginis to Kaveripakkam or its immediate environs. Numerous sculptures and architectural fragments from Kaveripakkam are now housed in the Government Museum, Chennai, and more remain in the village itself. Several inscriptions found in Kaveripakkam date to the second half of the ninth century, which suggests the presence of multiple temples there at that time, and a smaller number date to the tenth century (Mahalingam 1985, pp. 3–5). The ninth- and tenth-century temples must be the source of at least some of the sculptures and fragments presently found in the museum and on-site.

Today, Kaverippakam is a small village with a few locally active temples and underproductive rice fields abutting a massive reservoir. A bund 5 km long separates the water from the rice fields, which indicates that this reservoir was once much larger and more abundant (Figure 10).23 The roads that lead to Kaveripakkam are no longer major, and few people find reason to visit. However, a look at the site as a whole helps us begin to reimagine Kaveripakkam as it existed in the ninth and tenth centuries.

Around the turn of the first millennium, Kaveripakkam was part of a constellation of villages and provincial towns that had Kanchi as their main urban hub (Figure 11). There were at least three major temples within the settlement—surviving temples may stand on the same footprint as the earlier ones. Plentiful resources of water fertilized the agricultural lands. The then-flowing Palar River lay just to the south, and the large human-made reservoir (ēri in Tamil) stood to the north of the settlement. The long bund suggests the remarkable labor force that the reservoir’s construction would have required. Lakes like these often also served as stone quarries that provided construction materials for local sacred architecture. Throughout premodern Tamil Nadu, temples were built using the type of stone most immediately available. Temple sites were likely even selected in part because of their proximity to stone (Stein 2021, pp. 100–1).

Most of the early sculptures and architectural elements that remain in Kaveripakkam have become disengaged from their original contexts and stand as loose fragments on the premises of later shrines (Figure 12). However, it is highly likely that the later complexes are reconstructions of shrines made of less durable materials, such as timber and brick. Notable sculptures that are roughly contemporary with the yoginis include two door guardians and a Vishnu in the Gōṅkaṇēsvara temple, several architectural elements and a Ganesha at a Kaṇṇi temple on the banks of the reservoir near a sluice, and significant portions of the Kōṭṭai Perumāḷ temple (Figure 13).

Whether found on-site or in the museum in Chennai, all the remains from Kaveripakkam have one thing in common: they are all made of a uniquely dark, fine-grained form of stone that matches the appearance of the Tamil yoginis. Whereas sandstone and lighter-colored granite are common types of stone for sacred architecture in Tamil Nadu, this particular material is exceptionally rare (Stein 2021, pp. 92–101). According to conservation scientists at the Smithsonian and geologists in Detroit, the yoginis are made of metagabbro, a basalt variety.24 What this indicates is that Kaveripakkam—or somewhere quite close to it—must have been an excellent source of this special stone, which provided materials for the local temples, the yogini temple among them.

5. Building and Using the Temple

With its proximity to Kanchi and its status as a sizable town that held two of southern India’s most desirable natural materials—water for irrigation and stone for building temples—in the ninth and tenth centuries, Kaveripakkam would have attracted elite patronage. Some yogini temples were supported by royal patronage (Dehejia 1986, pp. 88, 138–39; Hatley 2014, pp. 204, 206, 216, and n. 23, p. 218), but with no inscriptions definitively associated with the Tamil yogini temple, the identity of its patron or patrons is uncertain. Schmid (2013, pp. 143–46) and Nagaswamy (2006, p. 94) argue—on the basis of their proposed association between the yogini sculptures and the three-faced Shiva discussed above—that the patron was the Rashtrakuta king Krishna III (r. ca. 939–67), who gained power in northern Tamil Nadu after a victory over the Cholas in 949 C.E., and who seems to have remained in the region until ca. 960.25 If the sculptures were carved earlier in the late ninth- to the mid-tenth-century timeframe we have proposed, the source of patronage may have belonged to another ruling family or to a different social group altogether—not all temple patrons in premodern India were royal.26 Whoever they were, they must have had access to significant resources in order to secure the land on which the temple stood, purchase the materials and labor necessary for construction, pay artists for the numerous life-sized, expertly carved sculptures the temple enshrined, and sustain the temple’s ritual life.

Yogini temples accommodated multiple communities and modes of worship. As Shaman Hatley has shown, yogini goddesses emerged from tantric traditions, but their temples were not strictly tantric in nature (Hatley 2014, 2019). While individuals who had been initiated into the tantric traditions of the yoginis probably had a prominent role in the temples’ ritual life, the spaces were likely open to a wider range of devotees. Looking to a variety of ritual and literary texts as well as the temples themselves, Hatley convincingly argues that yogini “temples mark the goddesses’ entry into a more public religious sphere,” where practices included “image worship, fire ritual, night vigils, and animal sacrifice, with a variety of aims reflecting the goddesses’ diverse identities” (Hatley 2019, p. 4).27 In addition to practices such as these, which were likely open to non-initiated participants, it is possible that the Tamil yogini temple also served as a site for more secretive, powerful rituals that were restricted to tantric initiates, as Hatley (2019, p. 4) suggests was the case for yogini temples more generally.

In both esoteric and exoteric contexts, the yoginis’ imagery, which combines elements of the beautiful and the terrifying, would have encouraged practitioners to think beyond the assumptions of mainstream convention, and perhaps to confront their own fears. The Yaśastilaka, a Sanskrit literary text composed in 959 C.E. by Somadevasuri, illustrates the terrifying aspects of yoginis. Written in the Rashtrakuta kingdom while Krishna III was in residence at his Tamil capital, Melpadi, the plot involves a power-hungry king planning to sacrifice animals and even human beings to the goddess Chandamari, whose temple is inhabited by yoginis.28 Pointing to the geographical proximity of Melpadi to the likely location of the Tamil yogini temple, Dehejia suggests that Somadevasuri’s vivid descriptions of the yoginis may have been inspired by a visit to this very temple (Dehejia 2021, p. 277). As the poet imagined it, these loud, unruly goddesses struck fear in every being who encountered them, wreaking havoc across the entire cosmos:

As abruptly as darkness descends at nightfall, even so, without warning did the Mahayoginis appear out of the sky, the earth, the depths of the nether regions, and the four corners of space … In their hands they held staffs topped with skulls and decorated with myriads of little bells which jingled furiously with the speed of their flight … Sparks issuing from the third eye on their foreheads were fanned into flames by the gasping of the helpless serpents ruthlessly enmeshed in the tangled masses of their hair; and these flames leapt forth so high as to singe the banners of the Sun’s aerial chariot. The ornamented designs on their cheeks were painted with blood which was being lapped up by the many snakes adorning their ears. Hovering over gruesome human skulls decorating their heads were vast numbers of giant vultures who obstructed the rays of the Sun. Tripping over one another in their haste, the Yoginis glowered repulsively and unleashed a host of tremendous and terrifying howls. Startled by the uproar, the Moon’s deer bolted off, trailing behind it scrambling constellations of stars entrusted with its care …29

If Somadevasuri did visit the Tamil yogini temple, it was not as a devotee—for he was a Jain, and his representation of the yoginis and the bloodthirsty goddess Chandamari was decidedly unfavorable (Hatley 2014, pp. 211–12). Whether he experienced the temple firsthand or learned about yoginis and their imagery from another source, to him, they were dangerous and malevolent goddesses, best to be avoided.

Somadevasuri was not the only one to view yoginis with trepidation. Fear has been a common response to yoginis, especially among people who do not understand them. Religious texts on the goddesses abound with warnings to anyone who might anger the yoginis by sharing ritual knowledge inappropriately, or by conducting rituals inaccurately—such individuals will incur the curse of the yoginis or even become food for the yoginis (see Dehejia 1986, pp. 34–35; Hatley 2013, p. 25 and nn. 22 and 27, pp. 28–29). Dehejia relates that during fieldwork in 1979–80, local people in many villages where yogini temples are found were reluctant to approach the structures, or even to speak about them (Dehejia 1986, p. ix). In the early 2000s, István Keul found that in Hirapur, where the ritual life of the yogini temple now focuses primarily on a single goddess, many residents viewed the yoginis as capable of both protection and destruction, but that elsewhere in Odisha people had largely unfavorable and fearful impressions of yoginis (Keul 2013, pp. 4–7).

Every one of the Tamil yoginis has endured damage—the figures have lopped-off noses, arms, legs, and even heads—and most of the breaks are now well weathered, which indicates that this destruction took place a very long time ago. At least some of it may have happened during the thirteenth or fourteenth century, which was a time of religious and political transformation for northern Tamil Nadu. More may have taken place later. Kaimal notes that the yoginis appear to have been deliberately broken, suggesting that their destruction was motivated by fears that people may have harbored towards the goddesses and their temple (Kaimal 2012, p. 2). Vandalizing the sculptures would have neutralized their power.

6. Scattering the Temple

There is little evidence of the Tamil yoginis’ history between the tenth and twentieth centuries. At some point in time, their temple fell out of use, and at some point in time, the sculptures were damaged. We cannot confirm precisely when these events took place or in what sequence. However, primary source materials in Paris pick up the story in 1925. Preserved in the archives of the Musée Guimet, these valuable documents have only recently been made available for research, and they remain unpublished.30 We are grateful to have been granted access in May 2022, though we cannot treat them fully in the space of this article (see note 3 above). Most of the materials are letters written by Jouveau-Dubreuil to “Monsieur” (Sir)—identifiable as C. T. Loo—and to an unspecified “Ami” (Friend), who was apparently a staff member of the Musée Guimet. Based in Pondichéry (Pondicherry), Jouveau-Dubreuil had been sourcing sculptures in southern India since 1924 for Loo to sell in his Paris and New York galleries. An undated document titled “Rapport sur les Antiquités hindoues” (Report on Hindu Antiquities), which may date to early 1926, informs us that Jouveau-Dubreuil worked with a number of Indian agents who negotiated with locals to purchase sculptures and that Tangavelou was his chief agent for Hindu antiquities.31

The most detailed account of the yoginis—identified as “Mêres” (or mères: "mothers” or matrikas) by Jouveau-Dubreuil and Tangavelou—is found in another undated document, titled “Résumé des Opérations faites par M. Tangavelou pour essayer d’obtenir des antiquités pour M. Loo” (Summary of Operations made by Mr. Tangavelou in trying to obtain antiquities for Mr. Loo), filed with correspondence from 1927 (Tangavelou, n.d., pp. 9–10, 12, 16–22). Written and signed by N. Tangavelou Pillai on his own letterhead, it is this document that informs us of Tangavelou’s full name.32 In it, he explains that he and Jouveau-Dubreuil first found the “Sept mères” (Seven mothers) together, in a small shrine, during a trip to Kanchi in August 1925. He describes the neighborhood as a village-like area consisting of four streets, but he does not say precisely where in Kanchi it was located. There are many such residential enclaves in Kanchi, even today. Over the next few months, Tangavelou returned on his own multiple times to negotiate the purchase of the sculptures. After collecting sixty signatures consenting to the sale, he determined that a majority of households had agreed, despite the ongoing objections of some residents, and he had the sculptures removed from their shrine and sent to Pondicherry.33

If the yoginis and their companions came from Kaveripakkam, then we might reasonably ask why they were found in Kanchi. How did they get there? When? And why? What type of journey did they have between their site of fabrication and a neighborhood in Kanchi? Although none of these questions can be easily answered, the scenario itself—deities re-enshrined away from their original home—should not be altogether surprising. In fact, it is quite typical throughout India to find that stone sculptures have moved, or more precisely that they have been relocated. We can turn to a range of scenarios for examples.

When sculptures become separated from their original context but remain in India, the most radical departure is that the sculpture enters a museum collection, where it is removed from worship entirely. This may entail traveling a long distance, for example to a national or state museum, or remaining close to the original location in an archaeological site museum. Alternately, a sculpture may be left untended on the premises of the temple with which it was originally associated. It happens with relative frequency that a sculpture or a fragment of a sculpture can be found within the boundaries of temple grounds but not inside a shrine. In other instances, a sculpture is moved to a different temple for safekeeping, but it is either not worshiped or not worshipped formally. The mark of worship on the forehead of the Durga in the garden of Uttiramerur’s Murugan temple reveals that at least one local devotee is aware of her presence (Figure 14). This sculpture would have originally been installed in a niche on the northern exterior wall of a temple, as is traditional for this goddess in Tamil Nadu. However, the Murugan temple does not have niches for deity sculptures on its exterior, which means that this sculpture must have belonged to a different temple—most likely one of the many in the same village, but it could have come from a more distant location.

A very common scenario is that a sculpture is moved to another temple location and is put into worship, sometimes as a different deity or type of icon than originally intended. This often happens with sculptures of the Buddha, which are commonly reinstalled in Hindu temples and worshipped in the same manner as Hindu deities (see Stein 2021, pp. 188–92). A more elaborate version of this is that a sculpture gets completely re-enshrined, sometimes in a temple that is built for it specifically. For example, in the village of Kumaravadi (near Uttiramerur), it became customary for people to install fragmentary sculptures found while plowing land or laying roads under a particular tree near the communal pond. In 2020, a local resident personally sponsored the construction of a shrine for the most complete figures of Shiva and Nandi, and he planted trees to accompany this new temple.

The Kumaravadi example returns us to the travels of the yoginis. It is possible that at least some of these goddesses—including those found by Tangavelou and Jouveau-Dubreuil—were relocated from Kaveripakkam to the more prominent urban center of Kanchi sometime after their temple fell out of use. This transfer may have happened around the sixteenth century, when the relocation of stone sculptures and architectural elements became more widespread as part of ambitious construction campaigns by new political players in southern India. Since the mid-fourteenth century, most territories in peninsular India—including Kanchi and the surrounding region—had been claimed by successive dynasties who built their capital at Vijayanagara, in what is today northern Karnataka. Beginning in the sixteenth century, new architectural practices arose throughout the Vijayanagara empire and in the territories of its Nayaka successor states, in which centuries-old stone temple sculptures and architectural elements were transported across great distances to be reused at new sites of power (Eaton and Wagoner 2014; Kasdorf 2016). Richard M. Eaton and Phillip B. Wagoner have convincingly argued that these new modes of architectural reuse—which prevailed in both Hindu and Islamic contexts in the Deccan—linked architectural patrons to an idealized historical past (Eaton and Wagoner 2014, pp. 77ff). We cannot say whether such a motive was at play for the patron(s) who moved the yoginis and Shiva-Vinadhara from Kaveripakkam to Kanchi, but the act of transporting stone sculptures across a long distance to be incorporated into a new context aligns with practices of architectural reuse that arose in the sixteenth century.

It was also in the sixteenth century that most of Kanchi’s large temple complexes were expanded to their present-day footprint. The Vijayanagara- and Nayaka-period (ca. mid-fourteenth–early eighteenth century) construction projects are best known for their towering gopurams (gateways), but shrines were also built inside the new enclosure walls. Abundant reuse of earlier construction materials is evident in the new structures, including in Vijayanagara-era temples in Kaveripakkam. Amid this shifting political and religious landscape, the goddesses and Shiva-Vinadhara could have been moved. Once they arrived in Kanchi, some may have become venerated and installed in existing or new shrines. In these new contexts, their original meanings would have been transformed.

It is also possible that the movement of the yogini sculptures to Kanchi took place earlier than the sixteenth century. In the center of the city, the Kamakshi Amman temple was consolidated into a single complex, dedicated to the goddess, in the middle of the fourteenth century. Previously, it had been a loose cluster of shrines dedicated to various deities, including some Buddhist and some Jain (Venkataraman 1973; Stein 2021, p. 192). It is possible that the transport of the yoginis from Kaveripakkam coincided with a rising focus on goddesses in Kanchi.

7. The Tamil Yoginis in the Twentieth Century

In total, Tangavelou and Jouveau-Dubreuil discovered at least ten yogini sculptures, in various states of completeness. By the end of 1925, they had located five, along with two smaller matrikas—the sculptures Tangavelou discusses in his “Résumé des Opérations” (Tangavelou, n.d.). These seven are most likely the same sculptures that are illustrated together in a photograph believed to have been taken at Jouveau-Dubreuil’s Pondicherry villa (Figure 15). In this photograph, we see the five yoginis—one of which may have been comprised of a body from one sculpture and a head from another—and two matrikas arranged in a neat row against a plain backdrop.34 The documents in the Musée Guimet’s archives corroborate Kaimal’s suggestion, based on the C. T. Loo archives, that the sculptures in this photograph were found inside the modern shrine. Kaimal has further suggested that their arrangement in the photograph replicates their arrangement within the shrine where they were found, but of this we cannot be sure (Kaimal 2012, pp. 13–14).

Letters from Jouveau-Dubreuil to Loo, dated 5 May 1926 and 27 July 1926, respectively, account for four more yoginis and a sculpture of Shiva described as being the same type as the “Mêres” found in Kanchi the previous year.35 In the first letter, Jouveau-Dubreuil explains that Tangavelou had succeeded in purchasing another “Mêre” and a Shiva that he identified as Bhairava. Photographs enclosed in the letter confirm that these sculptures were the yogini now in Toronto (Royal Ontario Museum acc. no. 956.181) and the Shiva-Vinadhara now in Boston (Museum of Fine Arts acc. no. 33.18). In the letter dated 27 July 1926, Jouveau-Dubreuil provides the cumulative list of sculptures he had sent to Loo thus far: Seven female figures (“Mêres”) were shipped to Paris in January 1926, three in April 1926, and one—together with the Shiva-Vinadhara—in May 1926. We have not yet located correspondence that specifically discusses the three “Mêres” shipped in April 1926, but they may be the yoginis that are now at the British Museum (acc. no. 1955,1018.2) and the Minneapolis Institute of Art (acc. no. 60.21), and the one at the Musée Guimet that is not pictured in Figure 15 (acc. no. MG18507). Archival photographs from Jouveau-Dubreuil that show each of these yoginis individually appear to have been taken at the same time, which further suggests their identification with the three sculptures found in April 1926 (see Kaimal 2012, figs. 22–24, pp. 45–47). The fragmentary torso of a yogini now at the Museum Rietberg (Zurich; acc. no. RVI-210) may be mentioned in a letter from 29 September 1926 that is published by Harle without its date line (Harle 2000, p. 294).36

In the decades that followed, C. T. Loo sold or donated the sculptures that he had received from Jouveau-Dubreuil to various museums and private collectors. He brought the Shiva-Vinadhara, the matrikas, and a selection of yoginis to New York for an exhibition in 1931; some were included in a second New York exhibition in 1942 (Kaimal 2012, pp. 142–45). In 1933, he donated three of the yoginis to the Musée Guimet in Paris, and in the same year, the Museum of Fine Arts in Boston acquired the Shiva-Vinadhara (see Kaimal 2012, pp. 8, 155–57, 165). After that, it was nearly thirty years before Loo’s gallery, by then under the direction of Frank Caro, sold the last of the yogini sculptures to the Minneapolis Institute of Art in 1960. The Detroit Institute of Arts purchased theirs in 1957. In the intervening years, however, various museums in the United States displayed yogini sculptures in their galleries, on loan from Loo. Documents in the DIA’s archives show that in 1929, the DIA borrowed one that had previously been in Philadelphia.37 In 1933, Loo’s office wrote to Benjamin March, the DIA’s curator of Asian art at the time, to request that this sculpture be shipped to the William Rockhill Nelson Gallery in Kansas City—now the Nelson-Atkins Museum of Art.38 Recent archival research by Michele Valentine at the Nelson-Atkins Museum has confirmed that this is the very same yogini accessioned into the Nelson-Atkins collection eleven years later, in 1944 (acc. no. 44-27).39

C. T. Loo was not the only one to lend yogini sculptures to museums. The yogini that is now at NMAA passed through the hands of two private collectors before landing in Washington, D.C., both of whom also lent the sculpture to public institutions.40 Between about 1959 and 1966, this yogini could be found at the Cincinnati Art Museum. During this time, she was part of the collection of Christian Humann, a famously private collector who remained mostly anonymous during his lifetime but whose collection became well-known as the Pan Asian Collection. By 1968, however, this yogini had been purchased by Arthur M. Sackler, who then lent the sculpture to the Brooklyn Museum of Art. It remained there until 1986 and then went to Washington in 1987, when the Arthur M. Sackler Gallery was established (now within the National Museum of Asian Art) as part of the Smithsonian Institution.

However, Jouveau-Dubreuil and Tangavelou did not find all of the known Tamil yoginis, and not all of the sculptures from the original group left India. One yogini, now at the Government Museum in Chennai (Figure 16) may have been separated from the others well before Jouveau-Dubreuil and Tangavelou came along in 1925–26.41 The first known record of this sculpture dates more than ten years later when the district collector stationed at Kanchi brought it to the museum in 1937.42 By this time, Loo had already begun exhibiting and selling the sculptures that Jouveau-Dubreuil had sent to him. Jouveau-Dubreuil’s correspondence with Loo makes no clear mention of this yogini, and she does not appear in any of the photographs taken of the sculptures before they were shipped to Paris.43 Kaimal, following Chance, suggests that Jouveau-Dubreuil gave this sculpture to the museum in exchange for permission to export the rest (Kaimal 2012, pp. 18, 139–40; Chance 2000, p. 13). This scenario does not explain the eleven-year lapse between Jouveau-Dubreuil’s shipment of the other sculptures to Loo and the accession of the Chennai yogini into the Government Museum’s collection. More likely, the Chennai yogini was found later than her one-time companions and was brought by the Kanchipuram district collector to the region’s primary museum.

The yogini now in Chennai is not the only sculpture from the set that Tangavelou and Jouveau-Dubreuil missed. We know that at least one more Tamil yogini followed a different path than the others. In addition, unlike her one-time companions, today this sculpture is enshrined and worshiped as a goddess.

8. A Kanchi Yogini

In February 2020, we traveled to Kanchi together and checked in on a goddess sculpture that we believed to be important for our study of the Tamil yoginis. This goddess was enshrined in a local temple, where she was worshiped as a fierce goddess, but not a yogini. The temple was primarily tended by women, and we are grateful to them for welcoming us into the shrine and allowing us plenty of time for close-hand observation. In recent years, this temple had been renovated, the lighting improved, and the sculpture deep-cleaned—inches of wax, ghee, sandal paste, pigment, and a lifetime worth of worship had been removed from the goddess’s face. Now the sculpture’s similarity to the other Tamil yoginis could finally be confirmed (Figure 17).

Now we could see the contours of the goddess’s open eyes and arched brows, the finely articulated ornamentation on her headdress, the long earrings and curls of wild hair framing her face, and her multiple necklaces barely revealed by a fold in her pomegranate-colored sari. Although the sculpture has sustained visible damage, it is still deemed worthy of worship. An elderly woman tending the shrine kindly raised the goddess’s garments to let us see that the legs are crossed at the ankles, just like the others from the group. The size of the sculpture corresponds well with the Tamil yoginis, and she has specific features that reminded us of individuals from the group (Figure 18). She wears a makara-shaped earring in her left ear, like the NMAA yogini. The spiral-shaped ends of her loose hair are similar to the curls of many yoginis from the group, including the one at NMAA. Her cheeks are incised with circles like the yogini in Minneapolis (Figure 19). The correspondences spoke for themselves—this had to be one of the yoginis.

The discovery of this enshrined yogini raises many new and intriguing questions and possible new narratives. Did she leave Kaveripakkam with the other yoginis that were found in Kanchi? If she was known to Tangavelou, then why was she able to remain? Are there other yoginis from the group in other Kanchi shrines today? In Tangavelou’s report, he explains that the local residents wanted to obtain permission from their goddess before agreeing to sell the sculptures (Tangavelou, n.d., pp. 17–18). Could this yogini have been the goddess whose blessing they sought? If in fact she were considered primary among a group of similar deities, it would follow a more widespread pattern that continues throughout India, even today. Often one image becomes more highly worshiped over time without an immediately apparent reason. The single yogini that is a focus of contemporary worship at Hirapur is one example, but the same holds true for a relief carving of a goddess on a pillar at the Minakshi Sundaresvara temple in Madurai. The subject need not always be female.

The temple that houses the yogini in Kanchi is quite recent, but there may have been a shrine to shelter her for hundreds of years. Her story may be similar to the sculptures in the village of Kumaravadi that recently had a shrine built around them on the banks of the local pond. Eventually, one way or another, this particular yogini came back into worship and became a singular goddess.

9. Reuniting the Temple?

Displayed today in museum galleries, the Tamil yoginis are immediately positioned as works of art, but in their presence, some viewers still perceive their power. (We certainly do.) Different strategies of display and choices around the content of labels can encourage a range of approaches and responses—visual, spatial, intellectual, and emotional. In Detroit, the DIA yogini is placed alongside sculptures representing Shiva and Vishnu—she stands for Shakti, the supreme goddess (Figure 20). If visitors read her label, they will learn that she once sat in a temple alongside numerous other yoginis, each one combining threatening and alluring qualities, and that they embodied the power of Shakti together. In the museum, however, she does this work by herself.

Similarly, in most other museums, each Tamil yogini stands alone—accompanied, perhaps, by other South Asian sculptures, but not by her original companions. The one exception to this is at the Musée Guimet in Paris, where three yoginis sit together. Additionally, in 2013–14, Debra Diamond brought three others together for her exhibition Yoga: The Art of Transformation (Diamond 2013) (Figure 21). Presenting multiple yoginis together, as these installations have, helps viewers better appreciate the spatial relationship the sculptures likely had to one another in the temple for which they were made. However, three sculptures can only begin to suggest the layout of a yogini temple, where goddesses surround their visitors on all sides.

As we write, we are in the early stages of planning an exhibition to reunite the Tamil yoginis that are now scattered across museum collections, as well as their Shiva-Vinadhara and the matrikas that accompanied some of them by the early twentieth century, if not before. At the heart of the exhibition, we envision the yoginis and Shiva in a semicircular formation, an installation that will evoke their probable arrangement within the temple they once shared. Although we cannot recreate this temple, we will invite visitors to imagine the experience of that space, where yoginis would have filled their field of vision. By placing emphasis on the entire group of sculptures and by reintroducing the spatial logic of a yogini temple into their presentation, this installation will depart from traditional modes of museum display, in which objects are typically exhibited as individual works of art or as cultural exemplars, with little acknowledgment of their previous architectural contexts. Kaimal identifies three strategies of traditional museum display that encourage readings of imperial triumph over colonized people and places: (1) placing objects from colonized regions in marginal areas of the museum; (2) presenting objects in a way that “erase[s] traces of the objects’ previous homes”; and (3) “subjugat[ing] these objects to new systems of ordering” (Kaimal 2012, p. 176). Our proposed installation of the yoginis and Shiva resists such imperialist modes of display.44 While many museums that steward the sculptures today address yogini temples in their labels, contextualizing the sculptures as effectively as possible within the limits of their collections, gallery text can go only so far.

In the proposed exhibition, we want also to prompt visitors to think critically about the multilayered histories and identities of the Tamil yoginis, including the multiple modes of practice they engaged in their shared temple and their transfer to new places and contexts of worship after the abandonment of this temple. The exhibition will also address and critique the complex structures of power that led to the sculptures’ removal from India during the colonial period and their dispersal to the museums that house them today. Beyond addressing the history of their colonial past, we are exploring new approaches for the future, considering how we might engage and support present-day communities with connections to the Tamil yoginis, and exploring ways to make the exhibition accessible to audiences in India.

Writing about the DIA’s 2017–18 reinstallation of its Asian art galleries—a project that involved extensive community participation—Swarupa Anila (then-director of interpretive engagement at the DIA) remarks, “In this work is the attempt … to decolonize our process” (Anila 2017a, p. 115).45 Building upon this experience as we plan the yogini exhibition, we are pursuing inclusive, collaborative approaches, inviting participation from individuals with multiple kinds of expertise and experience. So far, exhibition planning has involved workshops and brainstorming sessions with scholars and museum professionals based in North America, Europe, and India, as well as discussions with focus groups and other forms of community engagement.46 Participants in these activities—who bring a wide range of perspectives to the conversation—have already contributed valuable insights that we are carrying forward as planning proceeds; as of this writing, further symposia, workshops, and community engagement initiatives are under preparation.

In planning this exhibition, we are seeking ways to bring in the multiple narratives and many discourses that have contributed to our knowledge of and perspectives on the Tamil yoginis and their companions—from their original installation in a temple in Kaveripakkam, to their movements through northern Tamil Nadu, and, for some, their journeys out of India, to what we can know about them today. For it is the responsibility of an exhibition that reunites the Tamil yoginis to ask why and how they have become dispersed, and to consider new possibilities for the future.