Articles by alphabetic order
 Ā Ī Ñ Ś Ū Ö Ō
1 2 3 4 5 6 7 8 9 0

Naga, Yaksini, Buddha: local deities and local Buddhism at Ajanta

From Tibetan Buddhist Encyclopedia
Jump to navigation Jump to search
601415249 txsxn-s.jpg
0707-09 lg.jpg
Itb-tyth-cd 1.jpg
88143 n.jpg
DTS.Nag una.jpg
93 45jg02 n.jpg
Prajna paramita42k.jpg
Chogyal Namkhai Norbu.jpeg
Ated monk1.jpg
Awesa 024.jpg
DSC 7581.jpg
Migyur Namkhe Dorje.jpg
4328 n.jpg
Fifth Dalai Lama.jpg
Va Close Up.jpg
Abstract outer space plh53.jpg
21taras l.jpg
899 n.jpg
AmazinvArt 2.jpg

by by Richard S. Cohen


Several years ago, Dr. William Bodiford sent a message to the electronic discussion group Buddha-l posing the following questions: "Is the worship of local spirit cults an essential part of Buddhist traditions in Asia?--Or, in other words--Are there any Buddhist traditions in which local spirit cults do NOT play a major role?"(1) Bodiford's post generated a particularly stimulating thread of discussion. Indeed, the fact that this thread continued vigorously for the next month and one-half--through Christmas and the New Year, as well as the start of a new semester--testifies to the salient importance of Bodiford's query.

The parameters of discussion were quickly set; every term came under scrutiny. Did Bodiford pose one question or two? Did Bodiford really intend to suggest that local spirit cults are essential to Buddhist traditions? Might such cults be simply normal, but contingent, parts of Buddhist traditions? What did Bodiford mean by "tradition"? What defines a spirit cult? What are the criteria for determining whether a spirit cult does or does not play a "major role"?

When participants ceased problematizing the terms of Bodiford's post and attempted to respond to it, they showed a striking reluctance to answer him in the affirmative. A significant proportion were unwilling even to entertain the notion that local spirit cults may be as integral to Buddhist traditions in Asia as, for instance, the four noble truths. This response surprised me, though it should not have. After all, Bodiford posted his query to Buddha-l precisely because, in his words, "most scholarly accounts of Buddhist life seem to either ignore Buddhist worship of local deities, dismissing it as `folk religion,' or treat it as an example of syncretism."

Presumably, they do so because they do not find such deities or their cults essential. Although nearly all respondents acknowledged that cults for local spirits are normal parts of Buddhist traditions in Asia, there was nevertheless broad agreement that such cults are contingent, cultural accretions to the religion, not necessarily identifiable with "Buddhism" as such. Bodiford's inquiry inspired so vibrant and sustained a discussion because its ramifications spiral beyond the lack of scholarship on spirit cults, folk religion, or syncretism--though these are lacunae that must be filled--to issues of central import for any scholarly study of Buddhism. Where do we locate Buddhism? What is normatively Buddhist?

Almost no participant in the discussion was comfortable with Bodiford's use of the word "essential"; yet, almost every post attempted to pinpoint criteria for delimiting normative Buddhism. Typically, these criteria described a two-tier model, distinguishing a "true" Buddhism, founded in pure philosophy, the Buddha's exact attitude, or the confronting of essentialisms, from a "lesser" Buddhism that involves supernatural powers, the worship of spirits or deities, ordinary folk, and indigenous beliefs. Scholars who divided Buddhism thus tended to represent the upper tier as normative, sociologically as well as ideologically, by associating it with the community of monks. There also were a few discussants who problematized Bodiford's suggestion that local spirit cults may be conceived as essential to Buddhist traditions in Asia, while refraining from positing universalized criteria for normativity. This is the position I take myself.

In my reading, "local," not "essential," is the most important term in Bodiford's post. In seeking a universally generalizable Buddhism, scholars have ignored "place." One aspiring to escape from normative definitions may find, however, that place, locality, provides a powerful category for analytic reconstructions of Buddhism. Place sets the idiosyncratic and indigenous on par with the translocal and universal, the here and there with the everywhere.

This article will explore how a scholar of Indian Buddhism might take place seriously as a ground for interpretation. How does Buddhism come to be of a place? The answer that follows may be understood as a play on Peter Brown's dictum that "the supernatural becomes the depository of the objectified values of the group." That is, the "supernatural" is always emplaced for no better reason than that people live, work, study, meditate, and worship in places; if the term "Buddhism" is to be meaningful, it must be connected to people, to Buddhists, who lived their lives as Buddhists in places.

In this case, the category place will be treated in terms of locale. In particular, my subject is Buddhism as located at the Ajanta, caves, a fifth-century C.E. rock-cut monastic "village" in western India renowned for its paintings, sculptures, and inscriptions. What deities lived at this locale? How were the deities emplaced within Ajanta's monastic architectures? What do these deities' locations tell us about the constitution of Ajanta's Buddhist community and its values? How might cultic activities for the deities located at Ajanta be viewed as contributing to a local Buddhism, Buddhism as a religion of place? My analysis will weave together four discursive threads. The first of these threads is the explication of the theoretical assumptions that lead me to treat Buddhism in its local formation at Ajanta.

The second thread is the actual material evidence from the Ajanta caves themselves, which includes paintings, sculptures, architectural forms, and epigraphs. This Ajanta material will be set in broader contexts through my final two threads. The third is a consideration of models of patronage that had an impact on local community formation; the fourth, a consideration of the political pressures that fostered the development of Ajanta as a site for patronage as well as for the establishment of a local community. The local deities to be considered below will include a naga king, the yaksini Hariti, and the Buddha himself. The inclusion of the Buddha on this list, as a deity of place, may be surprising. However, it is a problematic conceit that the Buddha was a purely translocal religious figure, whose significance may be recovered exclusively through literary and doctrinal sources.

For this site's patrons, visitors, and community, the Buddha at Ajanta was a translocal divinity with a specifically local identity, just as the Great Goddess is Vaisno Devi at her temple on Mount Trikut(3) or Siva is Acalesvara in his temple at Tiuvarur. As I shall show, at Ajanta the Buddha became of this place through his association with and resolution of indigenous and uniquely local problems. Though in nirvana, this Buddha was nevertheless present and active in the world.

As the sangha's head, this localized Buddha, in turn, grounded the Buddhist sangha's participation in local social relations. Social relations necessarily involve interactions among human beings. Within the context of Asian Buddhism, social relations almost invariably involve nagas, yaksinis, and the Buddha as well. At Ajanta this was certainly the case.


As I noted above, scholars who responded negatively to Bodiford's queries often did so through recourse to a two-tier model, distinguishing indigenous religious cultures from a translocal Buddhist norm. Edward Conze, remembered as much for the strength of his opinions as that of his scholarship, elucidated this model with characteristic aplomb.

First, Conze divided Buddhism into two opposing forms: "Much of what has been handed down as `Buddhism' is not due to the exercise of wisdom, but to the social conditions in which the Buddhist community existed.... One must throughout distinguish the exotic curiosities from the essentials of holy life."

Second, Conze associated these two forms of "Buddhism" with sociological and historical realities: "The monks are the Buddhist elite. They are the only Buddhists in the proper sense of the word." Here, proper Buddhists are individuals who reject indigenous acculturation. This transition occurs ritually when, as the stock description reads, these individuals "cut off their beards and hair, put on red robes, and with proper faith follow in renunciation the Blessed One, who himself went forth from the home to the homeless life." Such Buddhists concern themselves little with social relations.

Separated from the pressures of family and work, they devote themselves to gaining spiritual power and spiritual insight leading, ultimately, to absolute social transcendence, nirvana. Conversely, to the degree that an individual does not segregate himself from the world, and becomes physically or mentally caught up in mundane, conditional, social interests, he falls away from this norm. Sherry Ortner explains the logic underlying this model: "It seems fairly safe to say that orthodox, canonical Buddhism was ... a religion of antisocial individualism.... There is, then, an a priori logic to the argument that Buddhism, given its premises, will be antagonistic to social life."(8) If one holds that Buddhism is a priori antisocial, it then follows that social involvements are always somehow extrinsic to Buddhism proper. Buddhism is not a social construction but a discursive construction of and about the truth.

That which is properly Buddhist is universal, never local; it is preserved in texts, not archaeological sites (except insofar as they reproduce textual paradigms); it is realized in mental cultivation, not bodily gesture. Yet, as Conze admits, "wherever the Dharma has been a living, social reality, Buddhist theory has combined lofty metaphysics with a willing acceptance of magical and mythological beliefs." Similarly, Ortner freely acknowledges that for Southeast Asian Buddhism "the a priori argument, predicting conflict between Buddhism and society from the antisocial premises of the doctrine, simply does not hold"; she attributes this failure to follow "logic" to "historical facts" and "social structure." Given Ortner's usual acumen, it is surprising that she did not see the tautologous nature of this argument, which measures a religion--a social and cultural formation--against the yardstick of an a priori argument and then attributes a mismatch to sociocultural factors. It is not Southeast Asian Buddhism that is "illogical" but the model that would valorize a single strand of Buddhist doctrine, normalize it, and seek a determinate logic therein.

Both Ortner and Conze make a "stock anthropological error," in the words of C. J. Fuller, when they "convert an indigenous, ideological distinction into an analytic concept, and then apply it to the empirical evidence to try to divide what is actually united by common underlying themes and principles." This, in short, is the matter at stake in Bodiford's question: How ought scholars to conceive of Buddhism as a social and cultural formation, a religion?

Ivan Strenski offers a solution with his thesis that the "domestication" of the Buddhist sangha was "a fundamental internal factor propelling Theravada along the path to a fully social religious status." Domestication occurs, Strenski writes, "whenever the sangha participates with the laity in institutions" through the mechanism of gift exchange. "Above all, the sangha is a ritual receiver of gifts."

The history of the sangha's dynamic interaction with the laity suggests that these two groups naturally came together to realize social goals; Buddhist cultures, societies, and civilizations emerge out of this union. In short, by unpacking "domestication of the sangha one can account for why Ortner's a priori argument does not hold for Southeast Asian Buddhism and why it makes little sense to delimit the participation of Buddhist monks in society as an illegitimate violation of Buddhist "premises." Here is a premise or logic to counter that promoted by Ortner and Conze. The logic of domestication begins with the empirical observation that one always finds distinct links uniting the sangha with the laity, and that this relationship is almost invariably expressed through gestures of ritual giving.

By subordinating doctrinal apologies for gift giving to the "bare" social fact of its occurrence, one is freed from Ortner's overdetermined premises; the full spectrum of Buddhist history and the fluidity of its local forms become available for consideration within a normalizing discourse. One can take account of place. Indeed, whereas social structures cannot credibly be read through the lens of doctrine, by starting with this empirical social data, we can read Buddhist ideologies, even those valorizing the monk as an antisocial individual, through patterns of exchange. How does domestication work?

Following the lead of Levi-Strauss's analysis of kinship, Strenski proposes that society is synthesized out of exchange relationships. There are two basic patterns by which individuals become bound to one another and to a society through exchange: restricted exchange and generalized exchange. "Restricted exchange" involves what may be called a commercial element: two parties contract an equitable exchange of goods or services on a reciprocal basis. We shall see such a mechanism clearly at work at Ajanta in the cult of a naga with whom the monks share their residence: the humans attend to his shrine/ home, and in turn he does not strike them dead. Each party fulfills its part of the bargain, and social harmony prevails.

For Strenski, relationships of reciprocity between monks and the laity (or nagas) are problematic and play only a secondary role in the sangha's domestication. This is because, although Strenski rejects the premise that Buddhism is fundamentally antisocial, he nevertheless tentatively accepts the ideological construction of the Buddhist monk as "a paradigm of non-reciprocity." In light of this "ideal," Strenski proposes that domestication "both avoids the appearance and substance of reciprocity and expresses more durable and inclusive patterns of relationship." Strenski's analysis went astray here, for, as I shall discuss below, restricted, reciprocal relationships are fundamental to Buddhism as a religion of place. The second modality of exchange, "generalized exchange," is domestication proper; it is not linear but complex in its configuration.

In Strenski's words, generalized exchange "seeks an unbalanced condition between exchange partners, which requires repayment at some unspecified time, typically by another group or person than the original receiver of the first gift: A gives to B who gives to C ... until A finally receives his due.... [Generalized exchange] thus links its members in a theoretically open system of indebtedness, the momentum of which tends to build up social solidarity." Again by reference to Ajanta, so long as the monks attend to the local naga king's needs, he provides steady and plentiful rains, benefiting lay society, which in turn supports the monks. Accordingly, Strenski unpacks the term "domestication" as "the condition of the sangha within a system of generalized exchange. `

Domestication' simply names a process of the sangha's participation in a certain social solidarity."(18) Here I lay a foundation for recovering Ajanta's local Buddhism. If, as Strenski claims, "the problem of how domestication came about is ... the problem of how Buddhist society was formed in the process of ritual giving" then "domestication" provides an analytic concept of exceptional utility. The very nature of the archaeological evidence from Ajanta compels this evaluation.

What we know of Buddhist life at Ajanta derives principally from physical remains that index past acts of giving. Excavatory and decorative work at the site occurred in two phases, the first of which lasted from approximately 100 B.C.E. to 100 C.E., the second from circa 462 and 480. Our knowledge of Ajanta's Buddhism is tied to gifts of monasteries, the wellsprings of Buddhist culture, society, and civilization.

The institutionalization of a Buddhist community at Ajanta through the establishment of a monastic "village" sets this sangha in a network of relationships that circulated obligation as a currency in the local sociopolitical economy. The monks' ritual interaction with local divinities was one means by which they met the obligations attached, like a hen, to their local residence in magnificent caves as well to their more general residence in a social system. As the mechanism of social solidarity, domestication provides a basic model for reconstructing the social processes involved in the foundation of a Buddhist community at Ajanta. Moreover, it is possible to link this generalized-versus-restricted-exchange distinction to another opposition I play with in this article: translocal versus local.

This point can be clarified by way of an example. In the Milindapanha, the first dilemma King Milinda poses to the monk Nagasena bears on the matter of gifts to the Buddha after his parinirvana: If the Buddha accepts gifts he cannot have passed entirely away. He must still be in union with the world, having his being somewhere in it, in the world, a shareholder in the things of the world; and therefore any honour paid to him becomes empty and vain. On the other hand if he be entirely passed away (from life), unattached to the world, escaped from all existence, then honours would not be offered to him. For he who is entirely set free accepts no honour, and any act done to him who accepts it not becomes empty and vain. In other words, if the Buddha is not transcendental and translocal, then he lacks value as an object of worship, so why attempt to engage the Buddha in an exchange relationship? Yet, if the Buddha is not present and local, then he cannot enjoy offerings, so why participate with the Buddha in an exchange relationship? Nagasena's response takes the same normative stance as that embraced by Conze and Ortner.

The Buddha is in nirvana; his is the ultimate in antisocial behavior; "the Blessed One accepts no gift." Nevertheless, Nagasena holds that offerings to the Buddha are efficacious because in this case exchange does not require reciprocity. The Buddha (or his commemorative caitya) provides an occasion, a context, for giving. In this way, the Buddha's translocality does not preclude his location within a network of generalized exchange, for such a network works through the displacement of presence and the postponement of profits. The Buddha is like Johnny Appleseed of US. folklore: both broadcast seeds over fertile public land and then departed for points unknown.

Even though these individuals are no longer present, the community can still enjoy the fruit of their labors. In fact, if Johnny Appleseed or the Buddha did not travel to the beyond (obviously their actions have vastly different cosmological scopes), they would not have become the stuff of legends; their fruits would not be as sweet. Here we see a linkage between generalized processes of exchange, the foundation of generalized community and social structures, and a valorization of translocality.

Such a set of equations may partially explain the apparent disjunction in the field of Buddhist studies between a scholarly interest in Buddhist nomos and one in local spirit cults. As networks of generalized exchange find expression in metaphors of translocality, so restricted, reciprocal relationships find expression in those of place, locality, and presence. Thus, whereas the Milindapanha set the Buddha's translocality as the basis for his absent involvement in acts of ritual exchange, fostering domestication through generalized exchange, this same text sets the monks' localization as the basis for their involvement in exchange relations, fostering domestication through restricted exchange. Nagasena explains: although it may seem contradictory for monks to accept fixed dwellings from the laity, this practice is acceptable because it makes the monks accessible to the laity.

Whether or not monks willfully intend to enter a restricted exchange relationship, their presence functions as the direct reciprocation for the gift of a dwelling. Similarly, the following story from the Mulasarvastivada vinaya (MSV) illustrates the import of place and presence for restricted exchange relationships between monks and donors: "The faithful erected many monasteries, [but] few monks spent the rains retreat in Sravasti. The (monasteries) stood empty. The benefactors did not receive the spiritual merit that comes from [the monks'] use [of their donations]. Instead, scoundrels took over [the monasteries). The Blessed One said, `Count all [the monasteries). Depending upon the calculation, every (monastery) should be used personally by one, two, three, or four (monks). [A monk) should pass the morning in one (monastery), [and] should dwell in a different (monastery) at mid-day, in another during the afternoon, and in [still] another at night.'"

Certain fruits of giving can only be had through monks' living presence. In order to reciprocate the faithful laity for their donations, and to satisfy their desire for spiritual merit, the Buddha disrupt the monks' own fives, enjoining them to play "musical monasteries." Strenski suggests that such a ploy had no role within orthodox Buddhist giving founded in generalized exchange, which "liberates gift-giving from petty calculation ... by appearing to be sacrifice."

 Sacrifice, giving without a discernible guarantee of return, is precisely what the Buddha's order precludes. According to this tale, donors must unequivocally know that they will receive their spiritual due; monks must do what is necessary to satisfy the laity. Reciprocation is a normal and proper part of the monastic life, even at the monks' inconvenience. In addition, despite Strenski's devaluation of spiritual merit as a direct reciprocation for gifts received, one often finds that merit bears a "tangible" fruit and a this-worldly benefit. The donor of a set of Buddha images in Ajanta's Cave 22 expresses the immediate results of giving in a verse included with this painting: "Those who have an image of the Conqueror made, in this [very life) become possessed of beauty, fortune, and good qualities; blazing like the sun in their faculties and sense, they become a delight to the eye."

 Similarly, the Pali commentary on the Dhammapada tells that when King Bimbisara made an offering of robes to the Buddha and community of monks, dedicating the merit from that transaction to a group of ghosts haunting him, those ghosts instantaneously and visibly "were clothed in divine raiments." Again, in the MSV, Sariputra deliberately increases Anathapindada's joy at building the Jetavana monastery by telling Anathapindada that at the same time that he holds a cord to measure the monastery's foundation a golden palace comes into being in Tusita heaven: the larger the monastery, the larger the palace. Restricted exchange, in short, works when the individuals involved are mutually present and when the fruits of the exchange are evident to the parties involved. At the most basic level, presence--for example, the localization of a sangha in a specific place--is itself the desired fruit.

As a social phenomenon, Buddhism cannot be understood apart from its local manifestations, nor solely through its local manifestations. Domestication, the creation of a Buddhist society, works through both modes of exchange relationship; both are normal. Nor is this the case solely for monks. If one is to avoid Fuller's "stock anthropological error," one must entertain the notion that the Buddha, too, might participate in restricted exchange relationships, even though he is sometimes represented in doctrinal texts as utterly beyond samsara. In mythological terms, to recover a local Buddhism at Ajanta one has to problematize the Milindapanha's pronouncement that the Buddha is absolutely translocal and does not enjoy worship himself.

How did Ajanta's patrons, community of monks, and artisans make the Buddha present and use that presence to address mundane and local concerns? How did the Buddha and his community of monks gain institutional momentum in a particular place--Ajanta--through restricted exchange relationships, and how did they diffuse that momentum throughout the broader society (and indeed the cosmos) to become domesticated through generalized exchange?


Before I localize the Buddha at Ajanta or explore the mechanisms of domestication as they pertain to patronage at the site, I will set forth, very cursorily, the geophysical and historicocultural environments that had an impact on Ajanta's establishment. Even today the Ajanta caves are difficult to access. In 1992 a helipad was cleared to expedite the visit of a Japanese prince; the rest of us ride the bus from the nearest rail link, Jalgaon, ninety minutes to the north, or from Aurangabad, three hours to the south. In the late fifth century C.E., when the caves were excavated into a sheer scarp overhanging the Waghora River (fig. 1), they were equally isolated.

Archaeological investigations of the district have uncovered no evidence of villages or towns located in the immediate vicinity of Ajanta's thirty-six monasteries, caitya halls, and shrinelets. The nearest population centers to receive sustained attention from Indian archaeologists would have been a strenuous day's journey from Ajanta: Buldana, approximately thirty miles to the east, and Bhokardan, the same distance to the south. In any event, no evidence links any of Ajanta's rock-cut buildings, or the paintings and sculptures for which they are renowned, to either Buldana or Bhokardan. Rather, epigraphs found on site associate Ajanta's various patrons with an emperor of the Vakataka dynasty named Harisena, who ruled from the city of Vatsagulma (now known as Washim), one hundred linear miles to the east-southeast of Ajanta (fig. 2).

Ajanta's architecture, paintings, sculptures, and epigraphs are representative of a complex and vibrant form of local Buddhism and provide a witness to Buddhism as a constituent of Indian society in the late fifth century. In actuality, Ajanta was created in two phases of patronage, the first stretching from approximately 100 B.C.E. to 100 C.E., the second from circa 462 to 480; my sole interest here ties in donors and donations belonging to the second phase. All evidence points to these fifth-century patrons as being closely affiliated with the Harisena. Walter Spink's chronology and history of Ajanta binds the caves explicitly to the fortunes and social program of the Vakataka polity. According to Spink, a Vakataka ruler named Harisena inherited his father's throne in approximately 460 C.E.

Almost immediately thereafter "a consortium of the emperor's richest and most powerful courtiers" hired architects and artisans to realize this great project. These courtiers included Varahadeva, a minister of Harisena's, responsible for Cave 16; Buddhabhadra, a noble-become-monk, who tells us that he was "bound in friendship with a minister of the mighty King of Asmaka over the course of many lifetimes" (elsewhere I have shown this Asmakan king was likely Harisena), patronized Cave 26; and an unnamed feudatory raja subordinate to Harisena was responsible for Caves 17, 18, and 19. Further, Spink has argued that Ajanta's Cave 1 was the dedication of Harisena himself, and that the neighboring Cave 2 may have been patronized by a close relative, perhaps one of Harisena's wives. Approximately twenty years after the site's efflorescence, circa 480, programmatic work at Ajanta patronized by Vatsagulma's luminaries crashed to a halt with the destruction of the Vakataka family. Due to the loss of sustained, certain patronage, the monks soon left this narrow chasm at the head of the Waghora River for points unknown.

The following study of localization and domestication of Buddhism at Ajanta will proceed through a number of case studies. First, I will treat a "local" deity, by investigating the incorporation of a shrine to a naga king, keeper of this Waghora chasm in Cave 16; here I will set forth a general pattern for localization. The second study will treat a "translocal local" deity. Moving into a more "normative" Buddhist space, both physically and mythologically, I will discuss a pillared chapel dedicated to Hariti--a yaksini whose personal story was known to all the Buddhist world--excavated inside Cave 2, in the rear wall beside the Buddha shrine. I will show how the displacement of this yaksini from her original home in Rajagrha, and her relocation within Cave 2, accomplished both the emplacement of Vakataka society within that cave and the transformation of its resident monks into priests, mediators between the lay and spirit worlds.

The third study will explore the Buddha as a "local translocal" deity. At Ajanta, the Buddha was assimilated in body and role to Harisena; as a king's principal duty is to maintain the social and cosmic orders, so Ajanta's local Buddhas fulfilled this task in the place of Harisena, who lived in distant Vatsagulma. In this way, these Buddhas' local presence played on translocal valences of Buddha as the Unexcelled King of Dharma to incorporate Ajanta and its community into the Vakataka social world. For the reasons explained above, these three studies will not be "completed" by a fourth that treats the Buddha as a purely translocal deity: such studies are plentiful. Although Ajanta is the richest of India's Buddhist archaeological sites for reconstructing a temporally and spatially localized Buddhism, the data found therein do not speak for themselves.

The studies that follow are worked out through a reading of the evidence local to Ajanta through broader patterns of mythology, folklore, and doctrine. My principal literary source for this contextualizing material will be the MSV, a patchwork of monastic regulations, jataka tales, sutras, and retellings of the Buddha's life well known to Ajanta's community. I will supplement the MSV with other literature associated with Ajanta, as well as with the eyewitness accounts of Chinese pilgrims who visited India between the fifth and seventh centuries.


Cave 16 at Ajanta was a monastic residence dominated by a huge, monolithic representation of Buddha cut in a shrine room in the back.

Varahadeva's dedicatory inscription describes this cave as "a splendid dwelling for the ascetic Indra (i.e., the Buddha) excavated on the finest mountain, home to a naga king."(37) Here Varahadeva elucidates two facts of great import: this place became the Buddha's home after formerly having been the abode of a naga king. Varahadeva's claims have two important ramifications for interpreting the site. First, by designating Cave 16 as the Buddha's dwelling, Varahadeva suggests that in some as yet unspecified way the Buddha was "alive" there.

Second, this mountain scarp's original inhabitant, its naga king, was rendered homeless when the Vakatakas began to institute a Buddhist community at the site. The naga needed a new place to live. And so, in addition to creating a home for Buddha, Varahadeva's dedicatory inscription tells that he also excavated a new dwelling for the naga king located immediately inside the cave's entrance, near river level (figs. 3. 4). The entryway in front of this shrine had no provision for a door: Cave 16's naga king sat as an unblinking guardian over the entrance to this monastery and the Waghora River before it. [Figure 3-4 ILLUSTRATION OMITTED] Nagas play an ambiguous role in Buddhist mythology.

Powerful creatures who dwell in a glorious but debased existence underground or in rivers, nagas control patterns of fertility and destruction through their power over rains, which may be sweetly life-giving or torrential and deadly. Nagas' power is distinctly localized, and in Buddhist stories they protect their turf fiercely. Indeed, nagas were among the Buddha's staunchest adversaries: soon after the awakening, Sakyamuni proved his power to a group of ascetics by spending the night in the lair of a fire-breathing naga and taming the beast without causing it harm.(39) Similarly, among the tales the MSV tells of the Buddha's visit to India's northwest, two involve the subduing of the nagas Apalala and Gopala; in both cases, fierce and stormy battles precede the Buddha's inevitable triumph. Once pacified, however, nagas become the staunchest guardians of the Buddha and his doctrine.

Naga king Mucalinda provided a refuge to the Buddha, when a storm threatened his blissful meditation soon after awakening.(41) The nagas in Ramagrama are remembered as preserving and revering one portion of the Buddha's relics.(42) In addition, the MSV tells us that the Buddha established his sutras among the nagas. (43) This latter belief possibly underlies the well-known story of Nagarjuna's retrieval of the Perfection of Wisdom Sutras from the nagas, as well as the lesser-known example of a naga who, having taken the guise of an old monk, preached a "corrupt version" of the Ekottara agama, a version preserved in his watery world. The following tale from the MSV allows a more detailed look at the subduing of nagas. Here we meet Asvaka and Punarvasuka, two nagas who had been Sakyamuni's disciples in their former birth. When human, these two were members of the "gang of six" (sadvargika), disciples who caused trouble in the sangha by violating all rules of propriety.

As one might expect of such students, they blame their benevolent teacher for their own failings and vow revenge. [The Blessed One) went to Nandivardhana. In Nandivardhana, King Bhavadeva, along with his courtiers, the candali's seven sons, and the earth-dwelling yaksas were well established in the Holy Truths. There was a great lake in that place in which Asvaka and Punarvasuka had both been reborn as nagas. Twelve years passed and (Asvaka and Punarvasuka) rose to the [lake's] surface. Angrily, they said: "Because the Blessed One didn't teach us the Dharma, we're now debased, born as nagas. So, let's destroy his religion." The Blessed One then deliberated: "Because these nagas, Asvaka and Punarvasuka, are very mighty and have extraordinary powers, they certainly could grind my teachings to dust after my parinirvana." After considering this [point], the Blessed One went to Asvaka and Punarvasuka. Having approached, he addressed them: "Asvaka and Punarvasuka, there is a teaching known as `The Dharma-discourse in four lines.' Learn it well!" "A reverend (monk) teaches us the Holy Dharma. Who is he?" Wondering this, they sank [back into the lake].

Those two realized, "The Blessed One himself taught us the Dharma, but we didn't recognize him." The Blessed One then fixed his reflection on the surface [of the water). Again and again, Asvaka and Punarvasuka [rise up toward the surface], gaze at the [[[Wikipedia:reflection|reflection]],] believe that the Blessed One remains present, and sink back down. The Buddha subdues Asvaka and Punarvasuka in stages. First, he gives them a Dharma-discourse: a waste of words considering that these two failed to comprehend his teachings even when human. Second. Sakyamuni establishes his definitive presence in these naga's territory by fixing his image at the entrance to their lair. In the end, this is not a story about the syncretic appropriation of a local naga cult into Buddhism. Rather, it tells how Buddhists make use of local deities in order to emplace themselves within a local society.

The MSV does not represent Nandivardhana's king or populace as particularly concerned with Asvaka and Punarvasuka. It is the Buddha who worries about them and Buddhism that needs them. The Buddha enters into a restricted-exchange relationship with Asvaka and Punarvasuka, a relationship based on presence. The Buddha makes two offerings to these powerful creatures of dim intellect: first, his teachings; second, his body. The Buddha gives the nagas these things so that, as the MSV tells us, they will not destroy his religion out of anger that he passed them by. Under the guise of making restitution to these nagas for whatever slight they felt, the Buddha fixed his presence in that very place, corporeally as well as dharmically. Although tantalizing, this text is incomplete: it does not tell us "what happened next" to Asvaka and Punarvasuka. We do not know whether, in Nandivardhana, a shrine to these nagas coexisted with that of the Buddha and, if so, how Asvaka and Punarvasuka would have been integrated into monastic life.

At Ajanta, a shrine for the naga king was placed within the precincts of Cave 16. What did the institutionalized presence of that naga mean for Ajanta's community? The following account from the fifth-century Chinese pilgrim Fa Hien shows one pattern for a sangha's ritual interaction with a local naga. Here Fa Hien describes a naga cult as it was instituted in a Buddhist monastery within the town of Samkasya. This local naga, Fa Men writes, "is the patron of this body of priests. He causes fertilising and seasonable showers of rain to fall within their country, and Preserves it from plagues and calamities, and so causes the priesthood to dwell in tranquility."

In return for the naga's good graces, Samkasya's monks provide Samkasya's naga with a chapel and a seat. Every day three monks bring the naga a religious offering, and once a year they "place in the midst of [the naga's) lair a copper vessel full of cream; and then, from the highest to the lowest, they walk past him in procession as if to pay him greeting all around." This may be what Conze meant by an "exotic curiosity." Just as the Buddha gave his presence to Asvaka and Punarvasuka, similarly, in Fa Men's story about the goings-on in Samkasya, local monks provide the naga a place to live and offer it worship in that place. First, the naga reciprocates in two ways.

The naga supports them. Fa Hien uses a language of direct presence: this snake deity is a patron who enables the monks to dwell in peace. Second, the naga favors the monks by not harming them. The following story from the MSV--in which a local monastic community shares its home with several nagas, mistreats those nagas, and is terrorized by those nagas--helps to clarify this dimension of the restricted, proximate, nature of the relationship between monks and their coresident nagas: "in this case, there is a monastery established in a place inhabited by nagas.

In an inappropriate spot, or dwelling, or path to a dwelling, a certain foolish, stupid, undeveloped, and clumsy old (monk) either shits, coughs phlegm, blows out snot, farts, or scatters befouled bedding about. Enraged, the nagas stand at the dwelling and at the path to the dwelling, as well as on the promenade, in the courtyard, and at the doorway, and strike down the monks." Insofar as this relationship works at the level of restricted exchange, it is primarily a matter of promoting local harmony. The monks and naga are both present; the fruits of their interactions are immediate and apparent. In the above examples, one sees that the presence of a local naga presents the Buddha and his monks with an opportunity to fix their own presence in the same spot.

This is what David Shulman has called the "phenomenon of localization" within his study of Tamil temple myths. The Buddhists enter into a restricted-exchange relationship with a local naga, who is identified with the earth and with all that is indigenous and unique in the site of the shrine or monastery.(49) Through this relationship based on shared presence, the Buddhists establish their local legitimacy. The relationship has discernible local effects: nagas, who might otherwise cause harm, are pacified; Buddhists, who might otherwise be strangers, are neighbors. At the same time, Fa Hien's characterization of the local monastic cult for the naga in Samkasya reveals it to be a paradigmatic example of generalized exchange. The monks propitiate the naga, who in turn causes a good monsoon, enriching the local laity, which in turn enriches the monks. This is a decentralized and dynamic network of exchange; the circulation of benefits and obligations from agent to agent constructs a working society and a domesticated sangha.

This is a neat, idealized example of a generalized exchange relationship. Such a system would, in fact, have been open ended and inclusive of, as Strenski observes, "an indefinite number of members."(50) If, in Strenski's words, "domestication" is "the condition of the (Buddha or] sangha within a system of [generalized exchange]," then "localization" is the condition of the Buddha or sangha, within a system of restricted exchange.(51) And just as generalized patterns of exchange do not occur apart from dyadic restricted interactions, so the domestication of a sangha cannot take place unless that sangha is emplaced within local society.

The interaction of the Buddha or his monks with local spirits is a primary means by which such localization occurs. To reprise my broader programmatic agenda: insofar as actual social interactions on the part of the Buddha or sangha take the form of restricted as well as of generalized exchange, local data is as essential (to recall Bodiford) as translocal to the nuanced reconstruction of Buddhist life. An interpretation of Buddhist life at Ajanta's Cave 16 necessitates consideration of the naga shrine excavated at this monastery's public entrance as certainly as it does consideration of the Buddha deep within the cave's recess.

For the Buddhism local to this spot, neither can be valorized as more essential or normal. We do not actually know what anybody did at Cave 16's naga shrine. We do not know how the local monks may have treated this snake king. And we do not know whether local rains were timely and sufficient or whether floods and epidemics touched the monks' and Vakatakas's lives. Yet, from these various accounts found in the MSV and Fa Hien, examples that could be multiplied, we see a distinct pattern of interaction that is crucial for reconstructing Buddhist life at Ajanta.

Nagas are ambivalent and dangerous beings, each of whom haunts a specific locale. When pacified--through the continuing presence of the Buddha in the form of an image, through the performance of propitiatory rites by the sangha, or through some other means not specified in the stories I cite--these local deities become benefactors for the humans who share their locale. For the Buddha or sangha to have an effective role in the control of local deities, they must themselves become local as well. The same holds true for Varahadeva's Buddha in Cave 16, where a dwelling for the ascetic Indra was incorporated with the home of a naga king.


Nagas were not the only local deities whose pacification by the Buddha and his sangha, enabled localization and domestication. The following story from the MSV shows that this pattern also worked for yaksas, chthonic deities typically associated with terrestrial and human fertility. In this tale, Sakyamuni Buddha visits the northern Indian city of Mathura at a time when a yaksa named Gardabha lives as "an enemy to friends, an adversary to allies, an antagonist to well-wishers, who steals children as soon as they are born."

Mathura's faithful Brahmanas and householders implore the Buddha to assist them. Sakyamuni adjudicates their grievance: if these local devotees will construct a monastery for the local community of monks and dedicate it in Ghardabha's honor, this local yaksa will cause no more harm.(53) The MSV concludes Gardabha's story by telling that the faithful of Mathura tamed a full twenty-five hundred yaksas, through building twenty-five hundred Buddhist monasteries in their honor. The people of Mathura do not want to get rid of their yaksas; they present themselves as the yaksas, friends and well-wishers. The Mathurans aspire to live in harmony with their chthonic deities.

The Buddha fulfills these lay followers' wishes by mastering the yaksas, and harnessing their power for social good. In this story, like those told of nagas above, the Buddha and sangha become localized insofar as they share their dwellings with the deities who are indigenous and unique to the sites of the monasteries. These restricted exchange relationships between Buddha, monks, and yaksas, serve as the bases for the broader project of domestication, an institutionalized social solidarity inclusive of, at least, monks, lay folk, and local deities. Worship, as performed in India with offerings of food, light, and water, typically occurs somewhere. In the above story of Gardabha, the monks live in the monastery, but it was constructed for the yaksa it was dedicated to his worship. One may imagine that Gardabha stood somewhere inside the monastic precinct receiving devotion, like Cave 16's naga king. Indeed, the Milindapanha accepts the fact that offerings to the Buddha will be made somewhere--at a caitya--even if the Buddha is not there to receive them.

The creation of a specific place for the worship of local nagas or yaksas or Buddhas within a Buddhist monastery is the physical means through which localization occurs. The establishment of a shrine (or monastery) is the precondition for a stable and predictable restricted exchange relationship in which offerings are reciprocated through, at the very least, the proximity and presence of the worshiped. Localization contributes to domestication: the Buddha and monks are meaningfully placed within a society when whatever is most particular to that society is placed within their monastic space.

As a sangha becomes geographically localized through the incorporation of a local divinity within its domicile, so a divinity's actual emplacement within the monastic plan might suggest its role or function in the mechanisms of domestication. Varahadeva's shrine for the naga king, for instance, was excavated within the precincts of his monastery, but it was placed outside the living quarters, outside the veranda, and outside the courtyard, down a flight of stairs, near the river, at the furthest reach from the Buddha who fives inside. The placement of this naga king transforms him into a guardian deity for the cave.

He falls within the Buddha's sphere, but his continuing significance has little to do with that specific relationship. This is not to say that this naga does not assist in the localization of Cave 16's Buddha or sangha. Rather, this spatial dynamic suggests that additional factors may have been at work in the establishment of this monastery. I will explore these factors in the next section. A shrine to the yaksini Hariti found inside Ajanta's Cave 2 presents a very different model. Cave 2's Hariti shrine is one of two subsidiary chapels cut into the back wall of the monastery, located on either side of the entrance to the central Buddha shrine (fig. 5).

To the right of the Buddha, the Hariti shrine holds monolithic carved images of Hariti and her consort Pancika (fig. 6); murals depicting devotees performing puja were painted on the side walls (figs. 7, 8). Balancing Hariti's shrine, to the antechamber's left, is a chapel holding monolithic images of Padmanidhi and Sankhanidhi, embodiments of the wealth and power controlled by Pancika These are the only two chapels dedicated to deities other than Buddha that were excavated within an Ajanta monastery. Hariti is an insider. [Figures 5-8 ILLUSTRATION OMITTED] Indeed, one difference between Hariti and Cave 16's naga is attributable to the fact that Hariti functioned within several semantic contexts simultaneously. Hariti fits the model of an ambivalent local spirit.

Like Asvaka Punarvasuka, or Gardabha, her conversion story ties her to a locale, Rajagrha, which she first terrorizes and then protects. Yet, unlike these other figures, Hariti also became a translocal deity for the Buddhists. Shrines to this yaksini were not to be found in Rajagrha alone but wherever Buddhist monks traveled. Hariti was an upasika, a lay devotee, who according to some accounts attained the stage of srotapatti, thereby entering the direct path to liberation; an early Mahayana treatise even calls Hariti a "great bodhisattva."(54) Iconographically, Hariti looks like any other yaksini iconologically, her identity was in the eye of the beholder.

The possibility that Hariti could be taken prima facie for a local yaksini, a translocal Buddhist protector, or even a great bodhisattva, coupled with her widespread inclusion within monastic architectures, suggests that at some level Mad may have served to limit local impact on Buddhist life. The apologetic was always available: she looks like a local yaksini, but "really" is not. Yet, at the same time that Hariti may have insulated monks from local religious pressures, she also aided in localizing a sangha. In a sense, Hariti was a portable local deity, a ready-made, institutionalized, translocal basis for localization. Cave 16's naga was truly of the place; he deserved worship even though he had no direct spatial relationship with the Buddha inside the cave. By contrast, Hariti was neither specifically local nor specifically translocal; her presence in Buddhist monasteries was required by the direct order of the Buddha, and her placement in Cave 2's architecture reflects this relationship. Yet, as I shall show below, Hariti's presence inside Cave 2 also brought the local lay society right into this monastic home as a permanent resident. The story of Hariti's taming has been recorded throughout the Buddhist world.(55) Here I reproduce the version told by the seventh-century Chinese pilgrim I Tsing. Although I Tsing never visited Ajanta, depictions of Hariti's conversion sculpted in Cave 2 correspond to the story as he tells it

 At the former birth of this mother, she from some cause or other, made a vow to devour all babes at Rajagrha. In consequence of this wicked vow, she forfeited her life, and was reborn as a Yakshi; and gave birth to five hundred children. Every day she ate some babes at Rajagrha, and the people informed Buddha of this fact. He took and concealed one of her own children, which she called Her Beloved Child. She sought it from place to place, and at last happened to find it near the Buddha. "Art thou so sorry," said the World-honoured One to her, "for thy lost child, thy beloved? Thou lamentest for only one lost out of five hundred; how much more grieved are those who have lost their only one or two children on account of thy cruel vow?" Soon converted by the Buddha, she received the five precepts and became an Upasika.

"How shall my five hundred children subsist hereafter?" the new convert asked the Buddha. "In every monastery," replied the Buddha, "where Bhikshus dwell, thy family shall partake of sufficient food, offered by them every day." For this reason, the image of Hariti, is found either in the porch or in a corner of the dining-hall of all Indian monasteries depicting her as holding a babe in her arms, and round her knees three or five children. Every day an abundant offering of food is made before this image. Hariti is one of the subjects of the four heavenly kings.

She has a power of giving wealth. If those who are childless on account of their bodily weakness (pray to her for children), making offerings of food, their wish is always fulfilled.(56) According to the MSV, after Hariti converted she gave her demon children to Rajagrha's sangha for safekeeping because the other yaksas mocked them for their new religion.(57) I Tsing traveled to India in order to observe and record the ritual life of Indian monks so that he might reform the Chinese sangha. This pilgrim's discussion of Hariti comes in the midst of a description of the posadha, a biweekly ceremony in which a patron worships the Buddha (sometimes with dancing girls), listens to the Dharma, and feasts the sangha. Though scholars often have viewed the posadha as a preeminently monastic occasion, a time for the sangha's ritual declaration of moral purity, I Tsing presents the posadha as the principal occasion at which exchange relationships were realized between monks and the laity, coalescing Buddhist communities into a social whole.

Indeed, in I Tsing's telling, the very constitution of sacred space for the posadha observance constructed a complete cosmos, everyone in his or her place. At the room's front was placed an offering for the arhats, and perhaps the Buddha and bodhisattvas; following that, the monks were seated in a row in order of seniority; and finally, in the hall's recess, at the lowest end of the row, was placed an offering for the yaksini Hariti.(58) One may see the posadha as a performative event that created a hierarchy of the sacred, analogically expressed by the spatial dynamics instituted for the rite.

During the posadha ceremony, Hariti was mapped as the sacred opposite of the Buddha, bodhisattvas, or arhats, who were placed at the posadha, hall's front. Together with these epitomes of Dharma, she functioned to bracket the ritual space. Placed between these antipodes, the monks were defined spatially as mediators between transcendental and chthonic powers, between the Buddhist ideal and the all too socially real. Like the ideal figures--arhat, bodhisattva, Buddha--who received offerings at a posadha, hall's other end, Hariti's presence served as the condition of possibility for the sangha's domestication: one chaos that made acts of ordering, such as the posadha, meal, not only possible, but also necessary. The incorporation of a yaksini in the posadha, ceremony served to localize the sangha. being feted; the incorporation of a yaksini specifically in the guise of Hariti ensured that this chthonic presence fit within a translocal Buddhist cosmos as well. In this way, Hariti fulfilled the structural role played by local deities in the localization of Buddhist sanghas without herself necessarily being of the place at which the posadha, was held.

The geography of ritual in Ajanta's Cave 2 is not isomorphic with that of the posadha hall. In the latter, sacrality was constructed through the binary opposition of potential harm and potential perfection, a space mediated by the sangha. In Cave 2, a binary opposition does not obtain, but a complementarity between the figures in the side chapels and the central Buddha does. Treating this cave's space as a programmatic whole, one finds that no matter what the order or direction of approach a devotee may have taken to the various icons, the spatially central Buddha never lost its ritual centrality. Here the yaksas and yaksini, in the side chapels do not bracket the space so as to delimit a dualist cosmology. Instead, this configuration suggests a continuous, perhaps functional, hierarchy of sacrality. Granted, Hariti's occasional demonic nature was recollected for the worshiper in the little carvings behind her monolithic image (figs. 9, 10). Nevertheless, as a presence she is regal, maternal, and eminently approachable.

She is in the Buddha's sphere and portrayed as if the monks living at this site have maintained the diurnal duties to her and her sons set on them by the Buddha: her icon is the ever-present and unchanging sign of the Buddha's power and the sangha's performative success. Of course, this incongruity between the certain knowledge that Hariti is potentially deadly and the portrayal of her as benign is the very basis of her ritual feeding. Even though Hariti's iconography represents her as a sweetly maternal figure, the surmise that Cave 2's inhabitants would nevertheless have fed her bespeaks a concern for her ever-present demonic appetites, which would return in the absence of the Buddhist control mechanisms. In point of fact, although Cave 2's space is configured differently than that of the posadha hall, this shrine emplaces the local monastic community into the role of mediator between Buddha and deity, as well as between deity and laity, identical to the posadha ritual described by I Tsing. Here, too, the laity is brought into the picture, quite literally, through murals painted on the shrine's walls. On the right (fig. 7), one finds a scene in which laywomen and their children bear offerings to Hariti, place them in a pile before her, and then proceed to pay homage at her feet.

Although there is a feeling of narrative continuity between the sculptured figures and those painted next to them on the wall, there is an artistic discrepancy. The sculpted figures are hard, not in substance alone, but also in spirit: they maintain a quasi-iconographic stoniness absent from the painted figures, which are mannered, yet naturally individual. This dynamic interplay of formal and informal figures within a continuing narrative sequence is particularly evoked by the nonchalance with which a languid painted fly-whisk bearer seems about to step into the sculpture and take the place of the stolid stone attendant at Pancika's side.

The left wall of this chapel (fig. 8) differs from the right in that there is no sense of a direct interplay between it and the sculpted figures. Rather, breaking the flow of action toward the chapel's rear is one disrespectful woman facing away from the sculpted image of Hariti looking out from the shrine just as Hariti does. Unlike the other figures in these murals, this woman's hands appear to be held in positions characteristic of a divinity rather than a devotee. Her right hand suggests the varadamudra, the gesture of gift giving; the fact that it is also holding a child suggests the giving of children. Her left hand is in the sri-mudra, signifying good fortune, wealth, and royalty, formed by the joining of the ring finger and thumb.

The frieze's composition also sets this woman apart. By placing her to the extreme right of the group of four votaries--all of whom appear to be in attendance on her, though in informal poses--the artist highlights her position as the focus of the scene. The sense that she is the center of attention is augmented by the placement of a stylized mountain scarp directly to her right: she becomes impassable, the action of this scene having no other direction than toward herself. Finally, she has two children with her, also heading away from the sculptural group. I would suggest that figure is Hariti herself, slimmed and beautified in accord with the mannered naturalism of these friezes, come to speak with her constant devotees. If this figure is Hariti, then there is not a spatial but a temporal continuity between the scenes on the right and left walls of this chapel.

The right wall depicts the performance of a puja; thus, the painted figures proceed toward those in stone. The left wall portrays the desired outcome of that puja, in which the great yaksini Hariti grants her devotees darsan and satisfies their hopes and desires in accord with their petitions. The fact that these images of lay folk performing a ritual to Hariti were painted in a place where monks would have been expected to feed her encapsulates in living color the ideological tensions involved in monastic localization and domestication.

The sangha and the yaksini have a relationship based on restricted exchange. The monks live with Hariti and care for her physical needs; she not only agrees to peaceably coexist with them, but even gives her children to the monks for safekeeping according to the MSV. By appeasing this demoness and accepting her children, the monks create a world in which security, health, babies, and ample crops are the norm for every member of a local society, regardless of any individual's personal relationship with the sangha The sangha become domesticated in a network of generalized exchange. Cave 2's Hariti, shrine is a unique text in that its murals confirm and problematize these exchange processes. According to the paintings, it is not the monks who are in a restricted exchange relationship with Hariti but the laity. Such is precisely the ideal outcome of the monastic ritual.

The restricted exchange relationship between the monks and Hariti, should bring lay society "into the picture"; the generalized exchange ramifications of this monastic cult should work as if the local laity performed a successful puja for Hariti, Indeed, in a sense the emplacement of this sangha within local society could not have been complete were the laity not explicitly implicated in the worship. Hariti, is a translocal Buddhist yaksini. By placing lay society inside Hariti's chapel, through the paintings of local laity performing her puja, she was transformed into a particular local deity. The Buddhist monks become meaningfully placed within the local society as the society becomes visibly emplaced as a presence within the monastic space. Merely by living in the presence of Hariti, who lived in the presence of Buddha, Cave 2's monks were performatively transformed into priests, intermediaries between the human and spirit worlds.

 BUDDHA: AJANTA IN THE SOCIAL LANDSCAPE OF VAKATAKA INDIA Localization is one moment in the domestication of a sangha. A Buddha or community of monks becomes localized by entering into a restricted exchange relationship with a divinity intimately associated with a place; this relationship can be read, in part, through its expression within a monastery's architectural plan. The relationship works in both directions: an indigenous deity, such as a naga, is pacified by the Buddha's continuing presence; for the local community, the Buddha himself is bound with the identity and characteristics of the naga as well. Buddhists do bring certain translocal, generalizable values into a local economy of belief, but those values become coin only insofar as they are converted into a currency whose exchange is accepted in that locale. This process of conversion, in turn, symbolically emplaces society at large within the monastic precincts: this is domestication.

As we have seen, Cave 2's Hariti chapel is a remarkable text through which to illustrate the processes of localization and domestication at Ajanta. In the present section, I will treat the Buddha at Ajanta. How did this figure enter into a restricted exchange relationship with Ajanta's patrons? What roles did Buddha play within the generalized structures of Vakataka society? What was it about Ajanta's Buddhas' locations and presence that induced members of the Vakataka court to create magnificent homes for them, monasteries "almost measureless and inconceivable to the mean-spirited," in the words of Cave 17's donor?(59) The model presented above suggests that the localization of Cave 2's Buddha within the ideological and ritual structures of Vakataka India would have been intimately associated with Hariti's specific place within those same structures.

Accordingly, I will return to Hariti. As a translocal Buddhist divinity, Hariti performed many of the roles and took on many of the symbolic values associated with indigenous deities. As a typical "tooth mother," Hariti was given local legitimacy at Ajanta through her spatial relationship with Cave 2's Buddha and her ritual relationship with the sangha; this in turn visibly brought the laity into the recesses of that Buddha's home. In Vakataka India, however, Hariti had associations beyond that of Buddhist divinity or chthonic divinity.

Hariti had special meaning for the luminaries in Vatsagulma responsible for major patronage at Ajanta. Considered an ancestor of the Vakataka emperor Harisena. Hariti was significant in the Vatsagulma Vakatakas' conception of themselves as a royal family. Harisena was the last known monarch from the Vakatakas, a lineage of kings who ruled large swaths of central India following the dissolution of the Satavahana empire in the mid-third century. Epigraphic and puranic sources alike name Vindhyasakti as the family progenitor; his son, Pravarasena I, was deemed the greatest of the Vakatakas. Following Pravarasena I, the Vakataka family split into two collateral lines: one based in Nandivardhana;(61) the other in Vatsagulma. Six generations and about two hundred years after Pravarasena I, Harisena took on the Vatsagulma Vakataka mantle. As a token of Pravarasena I's glory, his descendants in Vatsagulma bestowed the epithet Haritiputra on him; their cousins in Nandivardhana did not.

The association of Pravarasena I with Hariti was an important symbolic means by which the Vatsagulma Vakatakas distinguished themselves from the Nandivardhana Vakatakas. Haritiputra translates literally as "the son of Hariti." There is no certainty whether Haritiputra was intended as a matronymic, a literal genealogical claim, or even an evocative epithet for military prowess, since Hariti's five hundred sons form the bulwark of the yaksa army. No medieval Indian king ever called himself a Haritiputra. Rather, this epithet was typically given to the founder of a royal dynasty or used to characterize the lineage as a whole. Nevertheless, beyond being a Buddhist divinity and a local yaksini. Hariti seems to have functioned for the Vatsagulma Vakatakas in particular as a clan deity. Writing on village goddesses, Fuller observes that a significant proportion of such figures "are tutelary deities of specific social units ... whose boundaries define the spatial extent of their powers."

Hariti's specific social unit would have been the Vatsagulma Vakataka ruling family; her domain, coextensive with Harisena's own. Monastic rituals for Hariti at Ajanta would have introduced the sangha into a generalized, structural system of exchange. Within the localized context of Vakataka concerns, however, this ritual also would have served as a basis for political legitimation, reinforcing whatever claims Harisena's family made based on its association--genealogical or metaphoric--with this yaksini. Such legitimation is a matter of both restricted and generalized exchange.

Through the ritual appeasement of Hariti, Ajanta's sangha, located near the Vakataka kingdom's geographical periphery, ritually straddled the one-hundred-mile distance, in direct service to King Harisena's political center; the establishment of a monastic "village" at Ajanta served to legitimate and strengthen Harisena's reign throughout his kingdom. Given this expanded understanding of Hariti's significance at Ajanta--site intimately tied to the fortunes and social plan of the Vatsagulma Vakatakas--one can better grasp the Buddha's local significance. Certainly as the guru of this Vakataka ancestor, Buddha earned the utmost reverence.

But Ajanta's Buddhas were not merely figures from the past. In a remarkable study, Gregory Schopen has demonstrated that stone Buddhas, such as those found in the recesses of Ajanta's Caves 2 and 16, "were actually thought to live in these establishments." Indeed, "the Buddha was considered to be the legal head of the group, and ... both the Buddha and the monastic community were thought to reside in the same monastery." As the leader of a monastic community whose rituals to Hariti legitimated Harisena's reign, the Buddha localized at Ajanta can be likened to a raja whose "troops-were mustered in the service of Harisena. This may be more than mere metaphor. Eighty percent of the monks who left literary records of their presence at Ajanta refer to themselves by the epithet Sakyabhiksu.

I have shown elsewhere that this term, in its most fundamental meaning, identifies an individual as a member of Sakyamuni's own family: Sakyabhiksus are bhiksus who are Sakyas. And one finds within the MSV that Sakyamuni's relationship with his clan members was conceptualized through royal ideologies and martial metaphors. Had Siddhartha become a universal emperor, it was argued, the Sakyas would have been his army and followed him into war. The Sakyas' obligations to and relationship with Siddhartha did not alter when he instead became the Unsurpassed King of Dharma. The Sakyas were still bound to follow Sakyamuni, but now as monks. Indeed, the MSV's ordination formula highlights the Buddha's kingship over the Sakyas: "I follow in renunciation the Blessed One, Tathagata, Arhat, Complete and Perfect Buddha Sakyamuni, the Lion of the Sakyas, the Overlord of the Sakyas."

In turn, by reason of blood alone, Sakyas were entitled to be ordained as monks immediately, without the customary four-month probationary period. Eighty percent of the Ajanta monks whose identity we know represented themselves as Sakyabhiksus: members of Sakyamuni's family and by extension the "army" of this Dharmaking. Buddha became localized in Cave 2 through his relationships with Hariti. As the head of a family association, a clan, which maintained a righteous order through its interactions with the Vatsagulma Vakatakas' own clan deity, this Buddha performed as vital a service to Harisena as any military ally.

In Cave 16, as well, the Buddhas local significance was intimately bound up with an attempt to legitimate Vakataka control over Ajanta through a local play on Buddhahood's translocal royal valences. Here the Buddha was not emplaced in the role of Harisena's family member or ally. Rather, Cave 16's Buddha was a double of Harisena himself. To grasp the role Cave 16's Buddha played as lord over Ajanta, one must consider the historicopolitical context surrounding patronage at the site. Epigraphic records show that Harisena's family colonized an empire that centered on Vatsagulma, but also stretched to the south, to Nanded and later Bidar. Harisena seems to have set his ambitions even wider: he created an empire stretching from the Arabian Sea to the Bay of Bengal, according to one reading of Varahadeva's Cave 16 inscription. This hyperbole is certainly suspect-Nevertheless, though the precise extent of Harisena's empire cannot be certified, we do know that Harisena began to colonize new lands early in his reign. A copperplate inscription found in Thalner, on the north bank of the Tapti River, records that in the third year of Harisena's reign he donated five villages to a number of Brahmanas with the permission of Gomikaraja.

There are a number of interesting and puzzling details to the Thalner grant, none more so than the question of Gomikaraja's identity and role in the Thalner region. But whoever Gomikaraja may have been, it is clear that Harisena was concerned to consolidate this outlying area into his, Harisena's, not Gomikaraja's, political center at Vatsagulma. The granting of land and villages to Brahmana settlers is a well-attested means for the cultural, ritualistic, and bureaucratic incorporation of newly acquired territory into a polity. Vakataka grants to Brahmanas typically fostered the colonization of farmlands necessary for an agricultural surplus to support the Vakataka court.

The names of two of the villages Harisena donated, Kamsakarakagrama (village of bronze workers) and Suvarnakaragrama (village of goldsmiths) suggest that artisanship and trade were also crucial to Vatsagulma's prosperity. Located in an area with ready access to the sea, this western colony's settlers could receive and process raw materials, sending their profits back to Vatsagulma in the form of taxes. Struck in the third year of Harisena's reign, the Thalner grant dates to approximately the same year that, according to Spink's chronology, members of the Vakataka court initiated patronage of the caves at Ajanta, Ghatotkaca, and Banoti.

In Vakataka times, Ajanta lay on a route that led from Vatsagulma to Harisena's Tapti River communities and out to the sea at the ancient port cities of Surat and Bharuch. Ajanta was a crucial stage on this route. In Vakataka times, as now, Ajanta was the pass in the Sahyadri range favored by caravans, merchants, armies, and monks for crossing down from the Deccan plateau to the Tapti River valley (or vice versa). In short, although Ajanta was (and still is) geographically remote from major population centers and seats of power, its spatial location did not translate into a similarly isolated social location. Located at a pass along the road from Vatsagulma to the east, Ajanta was a site of crucial strategic value for Harisena's expanded realm.

I have characterized Ajanta as a monastic "village" several times because the Vakataka court's patronage of Buddhists at this site served many of the same social functions for the Vatsagulma polity as Harisena's granting of five Brahmana villages near Thalner. The renaissance of work at Ajanta in the early years of Harisena's reign fits within a broader strategic program to secure cultural, economic, and administrative control over territories distant from Vatsagulma. Hermann Kulke has identified three means--in addition to a strong military-by which a "victorious conqueror succeeded in unifying the newly conquered areas permanently with his own homeland": royal patronage of important places of pilgrim-age; the granting of lands to Brahmanas; and the construction of new "imperial temples" within the core of region of the kingdoms.

 Ajanta can be loosely viewed as satisfying all three means. Intended as a one-quarter-mile series of cave monasteries, Ajanta would have been a village of sorts, albeit nonagrarian; the many shrines that were excavated into the scarp outside of the monasteries and accessible to the public reveal a marked concern with satisfying transient pilgrims; and though not in the Vatsagulma Vakatakas' core region, Varahadeva unreservedly attempted to make Cave 16 an "imperial temple.

" Vakataka support for Ajanta can be understood to fall within the first of Kulke's three ritual means for a polity to establish hegemony over peripheral territories: the establishing of pilgrimage sites (tirtha). Indeed, Varahadeva--"intent upon his duty" as a minister to Harisena--likens Ajanta to Mount Mandara, by whose foot runs the Ganges, the most revered tirtha of all. Varahadeva thereby transmutes the Waghora River into the Ganga and the Vakataka kingdom into Aryavarta, the "middle kingdom" in Brahmanic sacred geography. This explicit equation of Ajanta with Mount Mandara fits within a larger set of rhetorical strategies for reinforcing the legitimation of Harisena's royal power. Varahadeva's personal endeavor to unmistakably associate Ajanta with its Vatsagulma Vakataka overlord also is seen in his dedicatory inscription. Although Varahadeva transferred the spiritual merit accruing from his cave to his parents, Varahadeva's inscription details and celebrates Harisena's imperial lineage; it all but ignores Varahadeva's own family background. However, the piece de resistance of Varahadeva's legitimatory enterprise is seen in the iconology of the Cave 16 Buddha. In Cave 16, the Buddha was sculpted in an iconographic form known as bhadrasana, that is, seated on a royal throne, his legs pendant "European style" (also called pralambapadasana). Cave 16's massive central image was the very first fully plastic, sculptural rendering of this iconographic form in western India.

In fact, according to Spink's analysis, Varahadeva originally intended to place a Buddha seated in the cross-legged lotus position, not bhadrasana, in the central shrine but changed his plan midway through the site's brief history, in the wake of a regional war that may have placed Harisena's control over Ajanta in jeopardy. Stronger ritual medicine was needed! Introduced to India through the tradition of Kusana royal portraiture, the bhadrasana has been interpreted as iconologically "imbuing sacred images with a majesty and presence lacking in the rather compressed outline of the regular ascetic seated pose." This majestic presentation of the body is bolstered by accompanying details, which also bespeak royalty: a lion throne with a wheel of law at its base, attendants bearing fly whisks, and so on. Moreover, this Buddha's unique placement within the cave complemented the iconography. All other central Buddha figures within Ajanta's viharas were carved in shrines set apart from their monasteries' principal space. In some caves, the Buddha was placed within a separate chamber attached to the main pavilion; more typically, the Buddha was twice removed, his chapel set behind a shrine antechamber. By contrast, Cave 16s' Buddha was separated from this vihara's main space by only a pair of pillars.

Thus, Spink describes Cave 16's as "a revolutionary new Buddha, authoritatively posed, and looming directly above the devotee, rather than set back within a conventional shrine." In Cave 16, one experiences the Buddha's social presence as an awesome fact, in a way not matched anywhere else at Ajanta. "Compared with the other shrine images at Ajanta," Sheila Weiner observes, "there is a prepossessing and overbearing majesty to this figure that sets it apart conceptually."(85) Cave 16's massive central Buddha image is uniquely large and uniquely regal at Ajanta; seated with legs pendant, a position associated with royalty and worldly action, this is a Buddha who is not only operative in the world but acts therein as a king. Given Varahadeva's project of glorifying Harisena through this cave, this image may have functioned as something of a "portrait sculpture"; it may have allegorized Harisena as Buddha, whom Varahadeva likens epigraphically to an ideal king who "extinguishes the flames of wickedness" and enables the world to "enter that peaceful and noble state free from sorrow and disease, (attained) by eradicating the many faults." If so, the Cave 16 Buddha would fit into a pattern observed by Kulke.

As he notes, "one of the characteristic features of the cults at these centers of pilgrimage [which functioned to legitimize a ruler's hegemony] was an increasing process of a ritual `royalization' of these deities. Making such a homology complete, Varahadeva's inscription likens Harisena to Indra, Lord of the Gods, and also calls the Buddha the "ascetic Indra,"(88) Indeed, Varahadeva carries the identification to its logical conclusion: Cave 16 is characterized as the "home" of "ascetic Indra" and is given the name "Vaijayanta," eponymous with Indra's divine palace in the Heaven of Thirty-Three Gods.(89) Above I argued that Cave 16's shrine for the naga king assisted the Buddha's localization at Ajanta, and that the spatial disconnection between the naga and the Buddha indexed the nature of their relationship.

Just as Ajanta is at a far remove from the center of Vakataka power, so the naga, the chthonic embodiment of this place, is at a remove from the double for Harisena inside Cave 16. Buddha/Indra/ Harisena is present in this cave and this presence is itself sufficient basis for a restricted exchange relationship between him and the local naga. In return for its own pacified presence, the naga has the privilege of having its shrine located within this sacred domain. Sharing a place, Buddha/Indra/Harisena and the naga become accessible to the humans of that place.

Those humans' ritual actions, in turn, redound throughout all Vakataka society. Above we saw that the murals in Cave 2's Hariti shrine gave Vakataka society a symbolic presence at Ajanta; this emplacement was crucial, for the mutual presence of exchange partners is a prerequisite for the restricted-exchange relationships whose emergent outcome is a generalized social structure. Were Cave 2's monks truly antisocial individuals, they would not have entered into institutionalized relationships with the members of Vakataka society; they would not have become domesticated. Similarly, the local significance of Cave 16's Buddha is tied to his social function. For Harisena, this Buddha played the role of a local deity. Varahadeva symbolically placed Harisena in a restricted exchange relationship with this Buddha, a relationship of shared presence, thereby legitimating the Vakataka emperor's local control over the Ajanta pass. Just as the establishment of Brahmana villages, pilgrimage sites, and imperial temples are all ritual means for multiplying royal presence, the Vakatakas' establishment of Ajanta was a means of consolidating their rightful presence in this territory. Varahadeva, a minister "intent upon his duty," ensured that Cave 16's Buddha played an explicit role in this local process.

 CONCLUSION The monk Buddhabhadra, donor of Cave 26, proclaims that the wise "perform devotion intensively to the Tathagatas," since "even a single flower" offered to them causes the attainment "heaven and final emancipation."(90) Applying this logic to his own involvement at Ajanta, Buddhabhadra holds that "powerful and affluent bodhisattvas who are desirous of mundane pleasures as well as liberation," like himself, have an obligation to commission cave monasteries. Worship and giving are acts of the wise that result in both worldly and transcendental boons. Buddhabhadra does not set the mundane and supermundane in a hierarchy; he does not portray one as a precursor to the other; he does not characterize one as essentially Buddhist, the other as an accommodation to laity or indigenous concerns.

Indeed, Buddhabhadra publicly acknowledges that one of the motivations for undertaking this cave was to enable his recently deceased friend, Bhavviraja, a minister of the mighty king of Asmaka, to enjoy himself in heaven. Nor does one have to read too closely between the lines to learn that Buddhabhadra also used this donation as a means of currying favor with Bhavviraja's son, Devaraja, who took his father's position in the Asmaka court. The monk Buddhabhadra intended his patronage of Cave 26 to meet personal and political concerns; he was public and unabashed about this; he made no attempt to hide it. Yet at the same time, Buddhabhadra also intended his donation to meet a universalized Mahayinist concern: that all living beings might attain Buddhahood.

He dedicates the merit from this act accordingly. In using the donation of a cave monastery to address local concerns alongside the translocal, the monk Buddhabhadra does not differ from Varahadeva or Cave 2's anonymous donor. If Buddhabhadra makes no apology for this, why should we? To gain a nuanced and dynamic grasp of Buddhism as a living religion in ancient India, one must begin by considering the fives and actions of Buddhists. I have attempted to do so here by exploring patronage at Ajanta through a dual filter, laying Strenski's treatment of exchange relationships over Peter Brown's dictum that "the supernatural becomes depository of the objectified values of the group."

That is, I looked at how local deities' participation in exchange relationships served to form, legitimate, and give meaning to a local Buddhist community at Ajanta. This study led to a series of conclusions. First, in existential terms, Buddhism in India was as much a means for meeting personal local concerns as it was a means for meeting generalized, translocal goals. Present happiness was valued at least as much as ultimate perfection, with no indication that the latter was deemed normative for Buddhists. Second, restricted exchange works to build local systems of social solidarity through a valorization of direct and present participation by the exchange partners.

The reciprocity of restricted exchanges was a normal dimension of Buddhism as a religion of place. One takes place seriously as an analytic category only to the extent that one accepts reciprocity as a factor in Buddhist life. Third, nagas and yaksinis--superhuman figures whose principal values are associated with reciprocal relationships, presence, and the immediate satisfaction of desires--do not come from outside Buddhism. The reasons for the incorporation of shrines to the naga king and Hariti in Ajanta's architecture are not clarified through the rhetorics of syncretism, overlay, or appropriation any more than such language would clarify shrines to the Buddha, bodhisattvas, or arhats. Varahadeva's building of a home for the naga king or Samkasya's monks' worship of a naga must be viewed as properly, fully, and fundamentally Buddhist practices. Fourth, worship, like politics, is a local matter, however wide-ranging the consequences.

As the Unexcelled King of Dharma, the Buddha's reign was cosmologically limitless. One cannot know whether this mattered much to Harisena or Varahadeva, except insofar as the Buddha's translocal glory served Vakataka interests in governing the Ajanta pass. The Buddha was at Ajanta: his presence was indexed by his restricted exchange relationships with a naga, a yaksini, and a king. Generalized and displaced, the dharmic harmony resulting from this localized presence legitimated both the sangha's and Harisena's presence at Ajanta. As a "local deity," the Buddha brought the sangha and Vakatakas together in an institutional relationship, domesticating Ajanta's sangha within Vakataka society.

(1) Posted by William Bodiford, associate professor. University of California, Los Angeles, to on December 16, 1994. Messages to electronic distribution lists, like Buddha-l, are an informal medium of exchange between scholars. There is no expectation on the part of participants in these discussions that their observations and opinions are, or should be, finely crafted and suitable for publication. For this reason, the following account of the discussion inspired by Bodiford's post will be generalized and impressionistic. I will neither name names nor cite direct quotations. Peter Brown, "Society and the Supernatural: A Medieval Change," in his Society and the Holy in Late Antiquity (Berkeley and Los Angeles: University of California Press, 1992), p. 318. Kathleen M. Erndl, Victory to the Mother. The Hindu Goddess of Northwest India in Myth, Ritual, and Symbols (New York: Oxford University Press, 1993). (4) David Shulman, Tamil Temple Myths: Sacrifice and Divine Marriage in the South Indian Saiva Tradition (Princeton, NJ.: Princeton University Press, 1980), p. Edward Conze, Buddhism: Its Essence and Development (New York: Harper Torchbooks, 1975), p. 12. (6) Ibid., p. 53. Raniero Gnoli, ed., The Gilgit Manuscript of the Sayanasanavastu and the Adhikaranayavastu, Being the Fifteenth and Sixteenth Sections of the Vinaya of the Mulasaravastivadin (Rome: Instituto Italiano per il Medio ed Estremo Oriente, 1978), p. 15. I have read the manuscript's anagarad as agarad Sherry B. Ortner, Sherpas through Their Rituals (New York: Cambridge University Press, 1978), p. 157. Conze, p. 72. (10) Ortner, p. 159. (11) Ibid., pp. 159-60. C. J. Fuller, The Camphor Flame: Popular Hinduism and Society in India (Princeton, N.J.: Princeton University Press. 1992), p. 27. Ivan Strenski, "On Generalized Exchange and the Domestication of the Sangha," Man 18 (September 1983): 464. Ibid. pp. 464-65. Ibid., p. 472. Ibid., p. 473. (17) Ibid., p. 471. Ibid. Ibid., p. 470. For these dates, see Walter M. Spink, "The Archaeology of Ajanta," Ars Orientalis 21 (1991): 67-94. I will discuss Ajanta's history at greater length below. T. W. Rhys-Davids, trans., The Questions of King Milinda (New York: Dover, 1963). 1:144-45. Ibid., p. 146. One finds a similar question raised in Vasubandhu's Abhidharmakosa: How can a gift to a caitya be meritorious if there is nobody there to enjoy it? The Kosa explains that a gift may be meritorious either because it is directly enjoyed by the recipient or because it is abandoned with no one to receive it and no expectation of recompense (Dwarika Das Shastri, ed., Abhidharmakosa and Bhasya of Acarya Vasubandhu with Sphutartha Commentary of Acarya Yasomitra (Varanasi: Bauddha Bharati, 1987], p. 747, verse 4.121). Adrian Mayer points out a similar distinction within Vaisnava devotionalism. Puja is an exchange "made in connection with benefits for the worshiper," whereas seva is performed "Without thought of benefit or return" (cited in Fuller [n. 12 above], p. 71). (24) Strenski (n. 13 above), p. 473. Gnoli, ed., The Gilgit Manuscript of the Sayanasanavastu (n. 7 above), Strenski, p. 475. I reedited and retranslated all Ajanta's inscriptions in my "Setting the Three Jewels: The Complex Culture of Buddhism at the Ajanta Caves" (doctoral diss., University of Michigan, 1995). At times my readings coincide with those previously published; at times I diverge widely from previous scholars. All translations here are my own and are based on reconstructions of the Sanskrit that can be found in appendix A to my "Setting the Three Jewels." John S. Strong, The Experience of Buddhism: Sources and Interpretations (Belmont, Calif.: Wadsworth, 1995), p. 80. (29) Gnoli, ed., The Gilgit Manuscript of the Sayanasanavastu, P. 24. (30) For a survey of the few archaeological remains from the Buldana district dating to Vakataka times, see Ramesh Chandra Agrawal, Archaeological Remains in Western India (Delhi: Agam Kala Prakashan, 1989). For a survey of the few archaeological remains from Bhokardan district dating to Vakataka times, see Shantaram Bhalchandra Deo and Ramesh Shankar Gupte, Excavations at Bhokardan (Bhogavardhana) 1973 (Nagpur and Aurangabad: Nagpur University and Marathwada University, 1974). Spink, "Archaeology of Ajanta," p. 69. For Varahadeva's inscription see My "Setting the Three Jewels," pp. 357-62. For Buddhabhadra's inscription see my "Setting the Three Jewels," pp. 378-81. On the identification of Harisena as the mighty king of Asmaka referred to in Buddhabhadra's inscription, see my "Setting the Three Jewels," pp. 57-62. For the Cave 17 inscription see my "Setting the Three Jewels," pp. 367-71. (33) Walter M. Spink, "Ajanta's Chronology: Politics and Patronage," in Kaladarsana, ed. Joanna G. William (New Delhi: Oxford and IBH in collaboration with the American Institute of Indian Studies, 1981), pp. 119-21. (34) Walter M. Spink, personal communication with author, 1989.

(35) Although I have accepted the general outline of Spink's chronology for Ajanta's Vakataka phase, my own reconstruction of Vakataka history differs radically from the one Spink proposes. See my "Problems in the Writing of Ajanta's History: The Epigraphic Evidence," Indo-Iranian Journal 40, no. 2 (April 1997): 125-48, for a discussion of where Spink has gone wrong in his historical reconstruction, and my "Setting the Three Jewels," pp. 63-77, for an alternative, tentative reconstruction. (36) On the pros and cons of using the MSV for discussions of Ajanta, see my "Setting the Three Jewels," pp. 122-27. (37) Ibid., p. 361. (38) Ibid., p. 362. (39) Raniero Gnoli, ed., The Gilgit Manuscript of the Sanghabhedavastu, Being the Seventeenth and Last Section of the Vinaya of the Mulasarvastivadin (Rome: Instituto Italiano per il Medio ed Estremo Oriente, 1977), 1:217-18. See John S. Strong, The Legend and Cult of Upagupta: Sanskrit Buddhism in North India and Southeast Asia (Princeton, NJ.: Princeton University Press, 1992), pp. 26. 28. (41) Gnoli, ed., The Gilgit Manuscript of the Sanghabhedavastu, p. 126. (42) John S. Strong, The Legend of King Asoka (Princeton, NJ.: Princeton University Press, 1983), p. 219. (43) Gnoli, ed., The Gilgit Manuscript of the Sayanasanavastu (n. 7 above), p. 24. (44) S. Bagchi, ed., Mulasarvastivadavinayavastu (Darbhanga, India-Mithila Institute of Post-Graduate Studies and Research in Sanskrit Learning, 1970). 2:86. The same tale is told in Edward Byles Cowell and R. A. Neil, eds., The Divyavadana, A Collection of Early Buddhist Legends (Delhi: Indological Book House, 1987), p. 329. (45) Derge Kha, 120B4-121A3. I am working with the text as reproduced by the Asian Classics Input Project (ACIP), catalog number KD0001B. (46) Samuel Beal, Si-Yu Ki: Buddhist Records of the Western World. Translated from the Chinese of Hiuen Tsiang (A.D. 629) (Delhi: Motilal Banarsidass, 1981), p. xli. (47) Ibid., p. xlii. (48) Bagchi, ed., p. 134. (49) Here I paraphrase Shulman's definition of this phenomenon ([n. 4 above], p. 51). (50) Strenski (n. 13 above), p. 471. (51) Ibid. (52) Bagchi, ed., p. 18. (53) Ibid. (54) Noel Peri, "Hariti la Mere-de-Demons," Bulletin de l'Ecole Francaise d'Extreme Orient 17 (1917): 20, 23, 32. For the most detailed account of Hariti, see ibid., pp. 1-102. For Further references, see Strong, The Legend and Cult of Upagupta (n. 40 above), p. 303, n. 66. (56) I Tsing, A Record of the Buddhist Religion as Practiced in India and the Malay Archipelago, A.D. 671-695, trans. J. Takakusu (Delhi: Munshiram Manoharlal, 1982), p. 37. (57) Peri, pp. 13-14. The MSV story continues, telling that when the women of Rajagrha saw Hariti give her sons to the Buddhist monks they did the same. Later, these women paid compensation for the children's upkeep and still later "redeemed" their children and brought them home. This is significant for interpreting the shrine in Ajanta's Cave 2, since this story warrants the incorporation of youngsters into the monastic community, without requiring that these children become formally ordained. In other words, this is a myth that charters Buddhist monasteries to act as schools.

One sees precisely this on the base of Cave 2's sculpted image of Hariti, (seen at the bottom of fig. 6): a school room in which the good students sit in the front of the class, paying attention to their teacher, and the poor students hang to the rear of the class, playing with toy rams. Indeed, I Tsing notes that in India, the monks took on many students, whom they instructed in secular literature, these students had to live at their own or their parents' expense (p. 106). In the same vein. I Tsing presents the great Buddhist monastic universities at Nalanda and Valabhi as institutions that prepare young men to enter courtly life and advance to a high rank through the skills they learn therein (pp. 177-78). (58) Ibid., pp. 35-37. (59) Cohen, "Setting the Three Jewels" (n. 27 above), p. 371. For epigraphic sources, see ibid., p. 361; P. V. Parabrahma Sastry, "Hyderabad Plates of Vakataka Devasena, Year 5," Journal of the Epigraphical Society of India 13 (1986): 71-75; Krishna Mohan Shrimali, Agrarian Structure in Central India and the Northern Deccan (C. A.D. 300-500): A Study in Vakataka Inscriptions (Delhi: Munshiram Manoharlal, 1987), pp. 82-83.

For puranic sources, see Frederick Eden Pargiter, The Purana Text of the Dynasties of the Kali Age (Varanasi: Chowkhamba Sanskrit Series Office, 1962), pp. 48, 50, 72-73. (61) This is not the Nandivardhana in which Asvaka and Punarvasuka lived. (62) Vasudev Vishnu Mirashi, The Inscriptions of the Vakatakas, Corpus Inscriptionum Indicarum, vol. 5 (Oocatamund: Government Epigraphist for India, 1963), p. 96; Shrimali, p. 82; Sastry, p. 74. Fuller (n. 12 above), p. 43. (64) Gregory Schopen, "The Buddha as an Owner of Property and Permanent Resident in Medieval Indian Monasteries," Journal of Indian Philosophy 18 (1990): 203. (65) Ibid., p. 191. See my "Setting the Three Jewels," pp. 191-269, for a discussion of the evidence concerning the use of Sakyabhiksu at Ajanta and a fuller interpretation of this epithets significance. As is the case with the Vatsagulma Vakatakas' use of Haritiputra for Pravarasena I, one cannot determine whether the use of Sakyabhiksu is meant literally or metaphorically. (68) Gnoli, ed., The Gilgit Manuscript of the Sanghabhedavastu (n. 39 above), p. 200. Anukul Chandra Banerjee, Buddhist Vinaya Texts in Sanskrit Pratimoksa Sutra and Bhiksukarmavakya (Calcutta: World Press, 1977), p. 60. (70) Derge Ka I, 150B6-50B7 (ACIP catalog number KD001A; n. 45 above). See also I. B. Horner, trans., The Book of Discipline (Vinaya-Pitaka), vol. 4, Mahavagga (London: Pali Text Society, 1982), pp. 85-89. (71) Mirashi, The Inscriptions of the Vakatakas, pp. 94-100. Sastry (n. 60 above). Varahadeva tells that Harisena had some relationship with the lands between the Bay of Bengal and the Arabian Sea: Kuntala, Avanti, Kalinga, Kosala, Trikuta, Lata, and Andhra. Unfortunately, Varahadeva's inscription is damaged here, and the verb that delimits this relationship has been lost.

 Despite this uncertainty, however, it is probable that Varahadeva's verse is a digvijaya prasasti, a celebration of territorial conquest. See my "Problems in the Writing of Ajanta's History" (n. 35 above) for a more detailed discussion of this problem. (74) Vasudev Vishnu Mirashi, Indological Research Papers (Nagpur Vidarbha Samshodhan Mandal, 1982), 1:78-87; Ajay Mitra Shastri, "Thalner Plates of Vakataka Harishena: A Re-Appraisal," Journal of the Epigraphical Society of India 11 (1984): 15-20. (75) Shrimali (n. 60 above), p. 26. (76) None of the villages named in Harisena's Thalner grant were given any privileges, such as freedom from harassment by internal security police, freedom from corvee labor, or freedom from taxation. This is notable, for the majority of Vakataka land grants release their grantees from these and other obligations to the central polity as an inducement for individuals to settle there. (77) One millennium later the Mughals built forts in precisely the same locations that the Vakatakas excavated their Buddhist caves: the Fardapur and Ajanta forts guard the front and back of the Ajanta pass; the Abhasgash and Vetalwadi Forts protect the pass in which one finds Varahadeva's Ghatotkaca cave; the same relationship holds between the Sutanda fort and the Vakataka cave at Banoti. Indeed, these Sahyadri forts were instrumental in preventing the British from taking the Deccan plateau.

This region, along with the Ajanta caves, remained under the suzerainty of the Nizam of Hyderabad until Man independence; Fardapur, at the head of the Ajanta pass, marked the Nizam's northern border. (78) Hermann Kulke, "Royal Temple Policy and the Structure of Medieval Hindu Kingdoms," in his Kings and Cults: State Formation and Legitimation in India and Southeast Asia (Delhi: Manohar, 1993), pp. 9-10, quote on p. 9. (79) My "Setting the Three Jewels" (n. 27 above), p. 362. (80) For a similar strategy of geographical displacement, as used by the Rashtrakutas through their construction of the Kailash temple at Ellora, see Ronald Inden, Imagining India (Cambridge, Mass.: Blackwell, 1990), pp. 256-62. (81) Varahadeva's glorification of Vakataka accomplishments in his Cave 16 dedication is especially notable in light of a second inscription from Varahadeva, which dedicates a cave temple at Ghatotkaca. By contrast with the Cave 16 commemoration of Harisena and the Vakatakas this Ghatotkaca inscription, attached to a single cave in a lonely pass, details Varahadeva's own genealogical line of descent. At Ghatotkaca. Harisena receives only a brief mention.

This contrast between Varahadeva's Ajanta and Ghatotkaca inscriptions highlights that Varahadeva's role as the good minister was central to his undertaking at Ajanta. (82) Walter M. Spink, "Ajanta's Chronology: The Crucial Cave," Ars Orientalis 10 (1975): 143-69, pls. 1-16. (83) John M. Rosenfield, The Dynastic Arts of the Kushans (Delhi: Munshiram Manoharlal, 1993), p. 186. (84) Walter M. Spink, "A Scholar's Guide to the Ajanta Caves" (University of Michigan. Ann Arbor, typescript). (85) Sheila Weiner, Ajanta: Its Place in Buddhist Art (Berkeley: University of California Press, 1977), p. 98. (86) My "Setting the Three Jewels," pp. 361-62. (87) Kulke, p. 11. (88) My "Setting the Three Jewel," p. 361. (89) Ibid., p. 362. (90) Ibid., p. 380.