An Archaeological History of Indian Buddhism

Acknowledgments ix

A Note on Transliteration xi

  1. Introduction: The Archaeology and History of Indian Buddhism 1

  2. The Material of Religion 34

  3. From the Buddha to Ashoka: c. 600–200 bce 70

  4. The Sangha and the Laity: c. 200 bce–200 ce 104

  5. The Beginnings of Mahayana Buddhism, Buddha Images, and Monastic Isolation: c. 100–600 ce 146

  6. Lay Buddhism and Religious Syncretism in the First Millennium ce 180

  7. The Consolidation and Collapse of Monastic Buddhism:

    c. 600–1400 ce 202

  8. Conclusion 225

Bibliography 235

Index 247


Throughout this book I phonetically transliterate Sanskrit terms rather than use the more modern system of diacritical marks that is commonly employed in Buddhist studies. This was a difficult decision. Diacritical marks undoubtedly allow for the most accurate pronunciation of Sanskrit. I chose to use phonetic transliteration for two reasons. First, archaeologi- cal sites are by convention named by their excavators. In almost all cases the names of archaeological sites are phonetically transliterated. For reasons of consistency, other Sanskrit words should be similarly trans- literated. Second, I expect that many of the people reading this book will be scholars who work outside South Asia. Buddhist scholars and South Asian readers already know how to pronounce Sanskrit terms, as shown by the common practice of omitting diacriticals marks but otherwise leav- ing the spelling unchanged (e.g., “Asoka” rather than “Aśoka”). For the non-specialist, however, neither “Asoka” nor “Aśoka” would suggest the correct pronunciation—“Ashoka”—of the great Mauryan ruler’s name. While phonetic transliteration is less accurate and seems old-fashioned, it is the quickest and easiest way to approximate proper pronunciation for the non-specialist.

Using phonetic transliteration, Sanskrit and other terms are pro- nounced pretty much as an English speaker would expect. The one differ- ence is the use of the letter “h” to mark certain consonants as aspirated (e.g., “th,” “bh,” “chh,” and “dh”). With the exception of “th,” this does not cause any significant confusion. As for “th,” it is never pronounced as a fricative as in “them,” but rather as an aspirated “t” as in “stop”; “ch” and “sh” are pronounced as they are in English.

There are a few instances where I do not phonetically transliterate. The first are the names of the few archaeological sites that are not phonetically transliterated. Rather than alter the name of these sites, I use the name under which they are commonly published. I also have chosen not to alter the transliterations of any scholars that I quote in the text. All quotes are as originally written, with diacriticals omitted. Finally, when referring to

( xi )

Buddhist texts, I use the title of the cited translation, however transliter- ated by the translator. In all of these cases I have sought to preserve the intentions of the scholars I rely upon. While I may choose to use an older transliteration system to aid novice readers, I do not wish to force this decision on other scholars by altering their words.




The Archaeology and History of Indian Buddhism

The study of Buddhism, like the study of most major world religions, has long focused on written accounts of transcendent beliefs concern-

ing the spiritual world at the expense of material expressions of faith in the mundane, earthly world. Temples, however beautiful, are understood as a reflection of faith, not faith itself. At some level, this perspective makes sense. Buddhists, like the adherents of most other modern world religions, explicitly champion the transcendent over the illusion of the earthly and mundane. Being part of the mundane world, Buddhist tem- ples are merely way stations on the path toward enlightenment. To study Buddhism from a material perspective, then, necessarily imposes an alien understanding that is antithetical to what Buddhists themselves believe. Yet, this is exactly what I aim to do.

At its heart, archaeology is the study of the material remains of past cultures—the empty buildings, the discarded tools, and the garbage that people leave behind. While ruins inspire a peculiar sense of melancholy in those who visit them by forcefully demonstrating the impermanence of even the greatest of human accomplishments, archaeological remains also provide a window into the worldly actions of the long dead. Though different from the transcendental ideas preserved in ancient texts, these worldly actions are no less important for understanding the history of any people or any religion. This is not say that the study of ruins should sup- plant the study of texts. Rather, history needs to examine the interac- tion between the spiritual and the material, between the transcendent

( 1 )

and the mundane, and between faith and the practicalities of daily life. Accordingly, an archaeological history of Indian Buddhism must combine both textual scholarship and archaeological scholarship to produce a more complete understanding of the origin, development, and eventual collapse of Buddhism in the place of its birth.

The standard, textually derived history of Indian Buddhism begins in fifth or sixth century bce with Siddhartha Gautama, a prince who renounced his life of privilege for the life of an ascetic (e.g., Basham 1967; Davids 1910; Lamotte 1988; Lopez 2001). After many trials, Siddhartha achieved enlightenment, seeing the path to ending the cycle of rebirth and the end of suffering. For the rest of his life, he taught this path to an increasing number of disciples. After his death, his disciples contin- ued to promote the Buddha’s teachings and established a community of monks and nuns, known collectively as the sangha. Initially members of the sangha were wandering ascetics, living on the margins of society, beg- ging for their food, and practicing meditation and other ascetic rites. In contrast to the ascetic practices of the sangha, the Buddhist laity began practicing pilgrimage to key sites of the Buddha’s life, and to burial tumuli—stupas—that held portions of his cremated remains.

Over the centuries, and out of their desire to assist the laity on their path to enlightenment, the sangha gradually settled into monasteries and nunneries that drew the favor and financial support of the elite and non-elite Buddhist laity. According to the standard history, by the end of the first millennium bce, the sangha was becoming domesticated within monasteries, with ever escalating obligations to their lay followers. With the contact that these obligations demanded, the sangha began adopting the practices of lay Buddhism—worshiping at stupas and, by the early to mid-first millennium ce, Buddha images. Thus began a gradual degrada- tion of the pure, ascetic tradition of forest monks in favor of scholastic Buddhism centered in monasteries. Scholastic Buddhism focused on the study of texts and the formal education of novitiates in ever-larger monasteries in the north of India.

In the mid- to late first millennium ce, increasing numbers of lay followers abandoned Buddhism in favor of Hinduism, Jainism, and, eventually, Islam. Without lay support, Buddhist monasteries and pil- grimage centers were particularly vulnerable to Turks from central Asia, who invaded North India beginning in the second millennium ce. With their monasteries in ruins and their pilgrimage centers abandoned, the sangha either abandoned India for the Himalayas, China, Sri Lanka, and Southeast Asia, or abandoned Buddhism for Hinduism, Jainism, and Islam. By the fifteenth century ce in India, Buddhism survived in a

few small isolated pockets and as traces in the ritual practices of other religious sects.

Adding archaeology to the standard history of Indian Buddhism does not wholly change the narrative. Particularly in the latter half of the standard history, from the mid-first millennium onward, textual and archaeological evidence are in broad agreement. This shouldn’t be surpris- ing. Beginning in the mid-first millennium ce, the quantity, quality, and diversity of historical source materials increase dramatically. With better textual sources on which to base the standard history, it is not surpris- ing that archaeological evidence is largely in accordance with it. The same cannot be said of the first half of the standard history. When compared to what is known of the archaeological record of Indian Buddhism, much of the standard history appears to be wrong. Most important, archaeological evidence of early Buddhism indicates that from at least the third century bce—the earliest period in which substantial archaeological evidence is available—the sangha was fully domesticated. Where the standard history posits an ascetic sangha that slowly coalesced into monasteries, archaeological evidence suggests something like the opposite. From the earliest period in which archaeological evidence is available, the sangha was monastically based. Rather than developing early, it appears that the ascetic ideal only emerged in India in the mid- to late first millennium ce, and even then only a small minority of the sangha abandoned monasteries for the ascetic life of the forest.

While archaeological and textual evidence about early Buddhism seem- ingly contradict each other, it is not my aim to simply debunk the stan- dard, textual history of Buddhism and replace it with an archaeological history of Buddhism. The generations of textual scholars who developed the standard history were often brilliant, and the historical narrative they developed was based on real evidence. That said, the Buddhist textual sources that historians relied upon were written by a small, literate minor- ity for specific reasons in the early to mid-first millennium ce, some five to ten centuries after the events they describe. I argue that the Buddhist textual sources used to make the standard history of early Buddhism are colored by the concerns of the sangha during the periods in which they were composed.

In the mid-first millennium ce, the sangha was divided between

those who desired a more contemplative life as ascetics and those who desired a scholastic life as monastics. Much of the standard history of early Buddhism is based on readings of early to mid-first millennium ce Buddhist texts written by the pro-asceticism faction of the sangha. The authors of these texts sought to delegitimize established, monastic

Buddhism by portraying the Buddha as the prototypical ascetic. By rely- ing on these accounts, those scholars who created the standard history of early Buddhism confused later polemic for actual history. This does not mean that Buddhist texts that describe early Buddhism have no historical value. Obviously, they have immense value in illuminating the sangha’s concerns and interests at the time of their composition in the first mil- lennium ce. But the texts also have value for understanding the early history of Buddhism. In some instances, descriptions in Buddhist texts strongly concord with what has been found archaeologically. By combin- ing archaeological and textual sources, it is sometimes possible to separate those textual accounts that have a high degree of historical accuracy from those that are later interpolations. More so, Buddhist textual sources also illuminate a central, long-standing tension in Buddhism—the tension between the individual and the group. Relying on textual sources, many scholars have argued that this persistent tension was central to Buddhism from the very start (e.g., Carrithers 1979, 1983; Tambiah 1976, 1982). Following this insight, I argue that the strategies that different groups of Buddhists used to address and ameliorate this tension were central to the history of Buddhism in India.

I do not expect anyone to be convinced of my argument based upon this short introduction. I also admit that my version of the standard his- tory just presented ignores many significant debates among older gen- erations of Buddhist scholars and the study of more quotidian Buddhist texts by a new generation of Buddhist scholars (e.g., Lopez 1995; Schopen 1997, 2004, 2005; Strong 1983, 2004; Trainor 1997; White 2000). That is what the rest of this book is for. Here I am laying out only the most basic contours of an archaeological history of Indian Buddhism—a history that spans almost two millennia and includes the sangha, the laity, and the ruling dynasties that both were subject to. It must also be remembered that Buddhism was always one of several religions in simul- taneous practice in India. The history of Buddhism in India, then, is not simply the history of Buddhist thought, but rather the history of the Buddhist thought, Buddhists themselves, and the non-Buddhists they interacted with.

While this book is primarily intended as a study of the history of Indian Buddhism, it is also intended as an exemplar for the archaeological study of religion. It might surprise a non-archaeologist to know that, until recently, archaeologists have avoided studies of religion. For the most part, religion was the explanation of last resort, with “religious” often meaning little more than “I have absolutely no idea what this thing is or what it was used for.” Its not that archaeologists have not found ancient temples and

religious artifacts—they are almost unavoidable. Rather, archaeologists have not known what exactly to do with them. Following the tradition of seeing religion as belief, archaeologists often did little more than identify sacred sites as sacred and religious artifacts as religious. Where possible, archaeologists have also taken the bold step of correlating known archae- ological sites with sites named in historical or ethnohistorical sources. In American archaeology this is known as being the “handmaiden to his- tory” (Hume 1964), and as most would likely guess, archaeologists don’t want to be anybody’s handmaiden.

Beginning with the pioneering work of Colin Renfrew (1985) at Phylakopi sanctuary on the Island of Milos in the Aegean and John Fritz (1986; Fritz and Michell 1989) at Vijayanagara in India, archae- ologist have begun to take up the challenge of studying religion (e.g., Bradley 2005; Fogelin 2007a, 2008a; Insoll 2001, 2004; Kyriakidis 2007; Shaw 2013a; Walker 1998; Whitley and Hays-Gilpin 2008). For the most part, this challenge has been addressed by borrowing recent anthropological theories that see religion as something that people do, not simply something that people think about. Religion is created and revealed in the performance of a ritual, the construction of a temple, and the attendance at a ceremony. If people practice religion, then they are likely to leave at least some material traces of their actions. Those material traces are what archaeologists can use to identify and recon- struct ancient religions. For the most part, however, most archaeo- logical studies of religion have been synchronic—identifying religious practices in specific archaeological contexts or identifying the function of religion at a particular time. With a few exceptions (e.g., Bradley 2005; Shaw 2013a), archaeologists have not studied long-term religious change. They have not studied the ways that the material aspects of religious experience can shape and alter future iterations of that reli- gion. In Chapter 2, I will develop the methods necessary to animate the archaeology of religion and incorporate its insights into the study of long-term religious change.


The ways that an author defines and writes about a subject are critical to understanding the arguments an author is making. The choice of words, and the way an author uses those words, significantly frame the bound- aries of debate and discussion. These choices are important, as different words, even those that nominally refer to the same thing, have different

meanings. When choosing the best terms, it is necessary to recognize that no term perfectly expresses the desired meaning, and that all terms have their own history and baggage. In writing this book, I have been forced to make several critical choices concerning what will, and will not, be covered. These choices are reflected in the language I employ and the arguments I make. Among the most important of these decisions is the geographic region I intend to address.


Throughout this work I use the term “India” to refer to mainland South Asia south of the Himalayas. This use of the term “India” has no relation- ship to the borders of the modern nation of India. Rather, India, as I am using it, roughly corresponds to the territories of the modern nations of Pakistan, India, Bangladesh, and lowland Nepal. Though I wish it were otherwise, the region I am calling India matches the borders of British colonial India. By using “India” in the way I am, I am not trying to refer- ence colonial India, but I cannot fault anyone who is troubled by the cor- respondence. I am using the term “India” in the way I am simply because it is less cumbersome than consistently referring to the region as “mainland South Asia south of the Himalayas” or “the regions now called Pakistan, India, Bangladesh, and portions of southern Nepal.” Without a better term, “India” is the best I’ve got. To avoid confusion, throughout the text I use “modern India” to refer to the modern Indian state and “colonial India” to refer to British India. I use “India” to refer to the region in which Buddhism originated and initially spread in the centuries immediately after the Buddha’s death.

The Buddha was born and raised a small kingdom in what is now south- ern Nepal. The other key moments of the Buddha’s biography (enlight- enment, first sermon, and death) occurred in North India. Within a few centuries after the Buddha’s death, the practice of Buddhism spread to peninsular India and present-day Pakistan and Bangladesh. Later still, Buddhism was adopted in Sri Lanka, Central Asia, East Asia, Southeast Asia, and the Himalayas. The spread of Buddhism beyond India is among the most important and complex topics in the history of Buddhism. Wherever Buddhism was adopted, it was partially transformed to suit the needs of its new practitioners. As such, the history of Buddhism in China is different from the history of Buddhism in Sri Lanka, which is differ- ent from the history of Buddhism in Cambodia. A complete account of the spread and fragmentation of Buddhism in all the nations in which

it is, or was, practiced would require multiple volumes. Given this com- plexity, I have chosen to focus only on India, the place of Buddhism’s origin, development, and initial expansion. That said, I recognize that it is impossible to completely divorce Buddhism in India from Buddhism elsewhere. Throughout this book I make frequent use of information con- cerning India preserved in Chinese and Sri Lankan texts. I also note when doctrinal and other developments beyond India affected the practice of Buddhism in India.

Early Buddhism

Another term that should be noted is my use of “early Buddhism.” As used here it is intended to be a substitute for the term “Hinayana Buddhism.” Hinayana translates as “lesser vehicle” and refers to the earliest forms of Buddhism that existed in India. This was not the term used by early Buddhists themselves; it was applied by later Mahayana (greater vehicle) Buddhists as a somewhat derisive label for those who preceded them. In some of the earlier academic literature of Buddhist studies, “Hinayana” also became a term used to describe the later forms of Buddhism found in Sri Lanka and parts of Southeast Asia. The use of “Hinayana” for these later forms of Buddhism was predicated on the idea that these forms of Buddhism, particularly those found in Sri Lanka, had undergone less change (or corruption) than the Mahayana Buddhism of Central and East Asia. Rather than Hinayana, I refer to Sri Lankan and Southeast Asian Buddhism as Theravada Buddhism, the term preferred by Theravadin Buddhists themselves.

I am also skeptical that Theravada Buddhism is a better model for early Buddhism than any other form of Buddhism. There is ample reason to suspect that it has gone through as many profound changes as any other Buddhist tradition. Almost all Buddhist sects claim that their teach- ings represent the original words of the Buddha. From my perspective, I see no reason to privilege one a priori. For this reason—in addition to the negative connotations of Hinayana—I prefer to use the term “early Buddhism” until such time as Mahayana and Theravada Buddhism began to emerge and diverge. Early Buddhism, as I use it, dates from the time of the Buddha through the second or third centuries ce. It should be noted, however, that the terminal date is extremely fluid. The process in which Mahayana Buddhism developed is exceedingly complex and existing academic understandings of it contested (see Schopen 2000). It is likely that some Buddhist sects adopted Mahayana practices more quickly than

others. Even within individual sects some practices changed quickly, while other practices continued to follow earlier forms. Thus, the bound- ary between early Buddhism and Mahayana Buddhism is permeable and likely indefinable.

Buddhist Sangha/Buddhist Laity

Throughout this book I frequently contrast the actions of “the Buddhist laity” and “the Buddhist sangha.” I use these terms for the same reason I use the term “India”: because all of the other options are worse. The primary problem with these terms is that they imply far more coherence among “the laity” and “the sangha” than likely ever existed. Neither the Buddhist laity nor the sangha were ever homogenous. The sangha, as already noted, was divided into several different, often rival, sects, and Buddhist tex- tual sources note several failed attempts at preventing sectarian schisms. Even within individual Buddhist sects there were marked status differ- ences between members of the sangha, with abbots and those who had been ordained the longest having the highest status. These differences were even more pronounced in regard to gender. In some accounts, the Buddha was said to have only reluctantly accepted nuns into the sangha, and only after stipulating that even the most senior nun was subordinate to the most recently ordained monk.

If anything, the Buddhist laity were even more diverse than the sangha—with the same divisions in terms of status, wealth, and gender, but with additional divisions as subjects of rival kingdoms. The other problem with contrasting the actions of the Buddhist laity and the sangha is that neither were wholly separate from the other. Members of the laity, particularly older wealthier members of the laity, were often ordained late in life. Similarly, young people often became novitiates at Buddhist mon- asteries, gaining an education before returning to lay life. All told, the boundaries between the sangha and the laity were somewhat fluid. While it is difficult to talk about the history of Buddhism without discussing the sangha and the laity, it must always be remembered that neither group was completely coherent or completely distinct.

Similar to the division between the laity and the sangha, I also often speak of the differences between lay Buddhism and monastic Buddhism. Again, these terms likely obscure significant variation within, and simi- larities between, the forms of Buddhism practiced by the laity and the sangha. As used here, these terms are meant to draw on recent approaches to Indian religions that differentiate the popular religion of the laity from

the more abstract and esoteric religion of the sangha (Fuller 1992; Lopez 1995). As stated by Lopez (1995:11),

Buddhism has a vast literature dealing with what we term logic, epistemology, and ontology—works that are (depending on one’s perspective) as profound or as impenetrable, as rich or as arid, as anything produced in the West. However, like philosophical works in other cultures, Buddhist treatises are the products of a tiny, highly educated elite (largely composed of monks in the Buddhist case) and rarely touch the ground where the vast majority of Buddhists lived their lives.

The standard history of Buddhism posits a gradual corruption of the more meditative and esoteric aspect of monastic Buddhism with the more vulgar practices of lay Buddhism. In subsequent chapters, I argue that something like the opposite occurred. Whether following the standard history or the revisionist history presented in this book, the arguments require an analytical division of the laity and the sangha, lay Buddhism and monastic Buddhism, if only to allow for the patterns of interaction between different Buddhists to be fruitfully examined.

Colonial and Postcolonial History

A massive body of literature focuses on colonialism and postcolonial- ism in India.1 With each year, these investigations become more nuanced and subtle, tracking the effects of colonialism and postcolonialism with increasing detail. Although my focus is not on the history of colonialism itself, it is important to examine the effects that colonialism has had on the study of Indian archaeology (Chakrabarti 1997, 2001; Lahiri 2005; Singh 2004; Thapar 2000). These effects color both the colonial and post- colonial sources that I rely on. For this reason it is necessary to approach historical sources with an understanding of the context in which they were created. Many of the historical themes I explore here are based upon one simple “fact”—Sanskrit is an Indo-European language related to Persian, Greek, Latin, and most other European languages. Further, the Rig Veda, the earliest readable text from India, is written in Sanskrit. For the British and other Europeans, this established that some ancient link existed between the European and Indian civilizations.

  1. For example, see Dirks 1993, 2001; Guha and Spivak 1988; Breckenridge and van der Veer 1993; Inden 1990; see also Said 1978.

    With the recognition of the common origin of Sanskrit and European languages by Sir William Jones near the end of the eighteenth century (Jones 1824), modern European narratives of Indian history began to emerge. Thomas Trautmann (1997) provides a nuanced investigation of these narratives, arguing that colonial history emerged as a syn- thesis of the work of European Sanskritists and the racial science of nineteenth-century anthropology. In this synthesis, the introduction of Indo-European languages was assumed to coincide with the migration of Indo-European people from Central Asia—the Aryans—into India dur- ing the second millennium bce. Colonial historians credited the Aryans with introducing “civilization” to India.2 They were considered a race supe- rior to the indigenous South Asians, who were forced southward as the Aryans assumed control over northern India. For colonial historians, this served to explain the modern distribution of languages and races found in British India.

    By the mid-nineteenth century, colonial historians discovered that the Dravidian and Munda language families of peninsular India were differ- ent from the Indo-European languages of North India (Caldwell 1856; Campbell 1816; Ellis 1816; see Trautmann 1997 for a review of the iden- tification of the Dravidian language family). European historians postu- lated that the people who spoke these languages were the descendants of the people displaced by the Aryans. Thus, a linguistic divide between the North and South was explained as the product of the historical replace- ment and migration of different races. This racial explanation was but- tressed by the observation that people in the South were dark skinned in comparison to the lighter skin of the descendents of the Aryans in the North. The one “fact” left to account for was the apparent sophistication of the Dravidian speakers of the South. This was explained by a process of “Aryanization” or “Sanskritization,” in which northern Aryans were pos- tulated to have moved south to become the political and religious elite over the Dravidian speakers of the South (see Srinivas 1966, 1989). Just as the British employed local languages in their own colonial endeavor,

  2. The recognition of pre-Aryan cities at Harappa and Mohenjo-Daro in the early twentieth century muddied this picture but did not cause significant changes to the Aryan framework promoted in colonial histories. Rather, the Aryans were cred- ited with destroying the pre-existing civilizations of India, creating a strong break between pre-Aryan and post-Aryan civilizations. The pre-existing civilizations of India were then reduced to an interesting historical oddity, without any direct impact upon the later historical developments. More recent archaeological research (see Schaffer 1984; Kenoyer 1997) has asserted much greater continuity between the Indus Valley civilization and later historical periods, rejecting many of the assumptions of the Aryan explanation.

    the Aryans who moved south adopted the local languages of their new homes, slowly losing their original Aryan tongue. The presence of numer- ous Sanskrit loan words in Dravidian languages and common religious elements between the North and the South were seen as survivals of the process of Aryanization.

    Over the course of two centuries of colonial scholarship, a simple story of Indian history was constructed. The origins of Indian civilization were viewed as the product of an earlier influx of colonizers, who were themselves the progenitors of the Europeans who now were in the pro- cess of re-colonizing India. Today it is almost impossible not to recognize that this historical narrative legitimized the colonial practices of the British in India (see Chakrabarti 1997, 2001; Lahiri 2000, 2005; Singh

    2004; Trautmann and Sinopoli 2002; Trautmann 1997). Nevertheless, it should also be remembered that much of the colonial scholarship remains valuable to this day. That this scholarship existed within the context of British colonialism does not necessarily limit its usefulness in studying pre-colonial Indian history and archaeology.

    If colonial histories of India emphasized the civilizing affects of for- eigners, postcolonial histories of India emphasized the indigenous devel- opment of Indian traditions. Where colonial archaeologists saw social and technological advances in India as the result of successive waves of invad- ers bringing new ideas to India, postcolonial archaeologists were more likely to see these advances as homegrown. This does not mean that post- colonial historians rejected the claim that Sanskrit was an Indo-European language. Rather, postcolonial historians argued that Sanskrit was adopted by Indians rather than imposed by foreign invaders. More so, simply because Indians adopted Sanskrit, it does not mean that every other advance in India was a result of foreign innovation. Where British colonial archaeologists constantly looked outside India for the origin of cultural practices, architectural and artistic traditions, and systems of governance, postcolonial archaeologists emphasized the organic develop- ment of these practices and traditions within India.

    Overall, I tend to follow a more postcolonial view of Indian history. While India was located at the center of trade routes that linked East Asia, Southeast Asia, Central Asia, Africa, the Near East, and the Mediterranean, India was not simply a container into which foreign ideas were poured. This is not to say that all developments in India were entirely independent. They weren’t. With trade contacts came new commodities and ideas that Indians adapted to their own use. India, particularly Northwest India, was repeatedly invaded and occupied by Central Asians. Indians benefited and borrowed from the foreigners they came into contact with, but foreigners

    also benefited and borrowed from Indians—with the spread of Buddhism across Asia, perhaps, the clearest example of the latter.

    Anthropological Archaeology and South Asian Studies

    There is one final disclaimer that must be addressed before proceeding. Archaeologists who study South Asia are divided between two overlap- ping camps. The first, and larger, camp are those who take a South Asian studies perspective. This perspective takes South Asia as the focus of research, with different scholars employing varied methods to further elucidate whatever subject they believe will add to the knowledge of South Asia. The second perspective, and a far rarer one, is anthropological archaeology. Unlike most of Europe and South Asia, where archaeology is considered a part the discipline of history, in the United States archae- ology is a sub-discipline of anthropology. The placement of archaeology within anthropology in the United States is mostly due to a simple his- torical accident. In the late nineteenth and early twentieth centuries, archaeologists in the United States focused on the study of the ancient inhabitants of the New World as an extension of the ethnographic stud- ies of living Native Americans. Where European and South Asian archae- ologists studied their own history to understand where they came from, American archaeologists studied other people’s history to find out what makes people tick. These differences in the origin and academic affilia- tion continue to inform the practice of American archaeology to the pres- ent day. In the anthropological perspective, the goal of archaeological research is to explain how and why societies change in the broadest sense. The anthropological perspective relies on cross-cultural comparisons and the application of broader social theory to specific cases with the goal of explicating fundamental social processes. The South Asian studies per- spective, in contrast, is more particularistic, focusing on specific archaeo- logical sequences within South Asia. In general, archaeologists following a South Asian studies perspective distrust the application of broader social theories from one context to another. As a result, their research on South Asian archaeology tends to be empirically sound, but often has an atomis- tic and descriptive orientation.

    I am, unapologetically, an anthropological archaeologist, and the anthropological perspective pervades my studies of Indian Buddhism. The goal of An Archaeological History of Indian Buddhism is not only to describe the archaeological remains of Buddhism in India, but also to explain how and why Buddhism emerged, developed, and ultimately collapsed in

    mainland South Asia. This does not mean that I employ a social evolution- ary perspective that ranks diverse and unrelated societies into abstract categories of complexity or sophistication. While the social evolutionary perspective was the hallmark of American anthropological archaeology between 1960 and 1980, both before and after this period social evolu- tionary thinking made up only a small, highly contested, portion of the discipline. While I freely borrow theoretical and methodological insights of social theorists from diverse disciplines and archaeologists working in places as varied as coastal Peru and the British Neolithic, my approach is not based on a foundation of social evolution. Rather, I simply recognize that other archaeologists often think up innovative and interesting ways to study things, and sometimes these innovative approaches can inform how I study Buddhism in India. All that said, I do not believe that my anthropological proclivities preclude the possibility of also providing a comprehensive survey of Buddhist archaeological remains. Throughout this book I present archaeological data in as straightforward and acces- sible manner as possible, and I analyze that material to illuminate larger social processes. I value, and employ, the insights of archaeologists and historians who take a South Asian studies perspective. There are benefits and weaknesses to both approaches, and while I may lean more toward anthropology, and others lean more toward South Asian studies, any seri- ous South Asian archaeologist must blend the strengths of both.


    There are many different textual sources for the study of Indian Buddhism. These include Buddhist texts themselves, but also non-Buddhist texts from rival religious orders in India, secular texts from India, and accounts by Greeks, Persians, Central Asians, Chinese, and, later, Europeans who visited India. In terms of creating a history of Indian Buddhism, each of these sources has different strengths and weaknesses. Indian texts, both Buddhist and non-Buddhist, have the benefit of being written by Indians themselves, but it is often difficult to date their composition. The composition of foreign sources are usually more firmly dated, but the authors of these sources often seem confused—projecting their own biases on what they observed in India. A final source for textual histo- ries is inscriptions. Throughout India, both kings and commoners carved inscriptions in temples, columns, and rock faces. Among the main benefits of inscriptions is that they are often easily dated and, in many cases, the

    dates precede the composition of the earliest textual sources. On the down- side, most inscriptions are very short—providing only limited insight into the lives of the people who had them carved. All of these sources are criti- cal for constructing an archaeological history of Buddhism, but they must be incorporated carefully, accommodating the different problems and potentialities of each source.

    It should be noted at the outset that, other than inscriptions, most of the literary sources are written versions of what were oral and written histories that preceded them. For this reason, texts have two important dates. The first is the date that scholars believe the existing texts were written. The second is the period these texts refer to and from which they are presumed to have originated as oral histories. For example, many scholars argue that the Rig Veda, the earliest readable text in India, records oral histories referring to the mid-second millennium bce. However, the Rig Veda was not written down until sometime in the early to mid-first millennium bce, at the earliest (Basham 1967; Thapar 2002). It can be expected that, just as modern texts are constructed within a particular social context, the versions of the primary texts available today were also written in specific social contexts. The result is a mediation between what was written and the source material they were based upon, either written or oral (see Thapar 2000 for an excellent discussion of this point). These texts can be used for investigating periods to which they refer, but not uncritically.

    Buddhist Sources

    It is not possible to assign a date when Buddhist textual sources accu- rately testify to the nature of early Buddhism. The available sources are found in a variety of languages and Buddhist sects from throughout Asia. Within each tradition, each text has its own history and context. The frag- mentation of Buddhism and its dissemination throughout Asia resulted in numerous Buddhist sects. Though they shared many texts, each sect added its own interpretations and modifications. The result is that almost all texts can be found in multiple versions. In some cases the differences are minor. In other cases they are wholly different works. In and of itself, this progressive fragmentation of Buddhism is neither surprising nor problematic. The difficulty emerges because the originals upon which all of these texts are based are no longer available. This may be due to the natural decay of early manuscripts and the eventual collapse of Buddhism in India. It is also likely that for many generations after the Buddha, texts

    were only preserved orally, suggesting that the existing documents are among the earliest written sources.

    While Buddhism emerged and initially developed in the Gangetic Plain, the earliest textual sources are all found outside India (Schopen 1995:475–476). Sri Lankan texts available to modern scholars date to roughly the fifth century ce. The earliest Chinese sources also appear to date to the fifth century ce. Still later are Tibetan (seventh century ce) and Sanskrit (fifth–seventh centuries ce) sources. It is important to note that the latter are found, for the most part, in Central Asia rather than India. Just as the Pali Canon was modified to fit the Sri Lankan social context, we can assume that sources from Tibet, China, and Central Asia were similarly manipulated.

    Given this variety in sources, and the lack of original texts, many Buddhist scholars have employed what they call “higher criticism” to reconstruct early Buddhism (Bareau 1974; Frauwallner 1956; Lamotte 1988). The method, on the surface, is both simple and compelling. Those textual and doctrinal elements that are shared in the most disparate exist- ing sources are most likely to have the greatest antiquity. Thus, if a specific account is found in both the Chinese and Sri Lankan texts, for example, it is likely to have a common origin in India. Given that these sources date to the fifth century ce, and are believed to have been current for at least a few centuries before that, proponents of higher criticism postulate that their reconstructed Buddhism dates to the first few centuries bce. Some proponents even claim that they are reconstructing the form of Buddhism preached by the Buddha himself (see Bareau 1974).

    The actual practice of higher criticism is much more complicated than the simple outline presented above (critiques of higher criticism can be found in Schopen 1997 and Trainor 1997). First, since the existing sources have all been translated into different languages, the identifica- tion of similar phrasing or concepts requires reverse translation of the texts back into their original Sanskrit. This is not an easy or self-evident task. Second, higher criticism does not determine an actual date for the common, ancestral text that underlies its offspring. Despite claims by its proponents, commonalities in Chinese and Sri Lankan texts only demon- strate that the common text existed at an unspecified time prior to the existing texts in the fifth century ce. There is no reason to believe that this reconstructed Buddhism resembled anything propounded by the Buddha. The final difficulty in higher criticism lies in its assumption of indepen- dence among the different Buddhist traditions. If there was contact and borrowing among Buddhist traditions in China, Sri Lanka, Southeast Asia, and the Himalayas, many of the assumptions of higher criticism crumble.

    For example, if one sect developed a new monastic code or interpreta- tion of a sermon and another sect simply copied it, no earlier common text is implicated. Commonalities can be the product of communication and contact, rather than common origin (Schopen 1997:25–29). As will be discussed in later chapters of this book, Buddhist monks from across Asia studied in prominent monasteries in the Gangetic Plain well into the start of the second millennium ce. As such, Buddhists from widely dispersed traditions would have had the opportunity to share and adopt each other’s innovations, violating the assumption of independence that higher criticism demands. With the abandonment of Buddhist monaster- ies in the Gangetic Plain in the beginning of the second millennium ce, contact between monks from distant regions would have been greatly reduced. Thus, regional independence of Buddhist traditions is a very late development, and the assumptions of higher criticism are problematic for reconstructing early Buddhism.

    Buddhist texts are traditionally divided into three distinct categories, collectively known as the tripitaka. The first category contains sutras, the sermons and biography of the Buddha. The second are the vinayas, con- taining the rules governing monastic life. The third, the abhidharmas, consist of later commentaries on earlier Buddhist texts. As far as under- standing Buddhist history, it is important to note whether accounts are derived from the more theological and abstract sutras or the more practi- cal and quotidian vinayas. If nothing else, the varying placement of these accounts shows whether the sangha considered particular activities a theological or practical matter. The development of Buddhist studies over the last two centuries can, perhaps over-simplistically, be understood as a shift from studies focusing on the sutras to newer studies focusing on the vinayas and abhidharmas—a shift from a more abstract and theologi- cal understanding of Buddhism to a more quotidian understanding of the lives of the Buddhist sangha. The mistake is to read the sutras as accu- rately depicting the activities of the early sangha, and the vinayas as the activities of the later sangha. This point will orient much of the discussion of textual evidence throughout this work. Wherever textual evidence is discussed, I will note whether it is derived from the sutras, vinayas, or the abhidharmas.

    Indian, Non-Buddhist Sources

    Of the non-Buddhist textural sources originating in India, the earliest, by far, are the Vedas, with the Rig Veda being the earliest. The four Vedas

    contain descriptions of early religious beliefs, rituals, medical practices, and heroic stories of kings. Together the Vedas are often seen as the foun- dation for the rest of Indian history, referred to in the subsequent reli- gious literature of Brahmanism, Hinduism, Jainism, and Buddhism. The rest of the primary source material from India was written in the first millennium ce. These sources include the Hindu epics, the Mahabharata and Ramayana, a more scholastic text on statecraft (the Arthashastra), and a collection of histories and religious stories called the Puranas. In the peninsula, Tamil Sangam literature consists of collections of poems. All of these literary sources are believed to refer to the first millennium bce but were composed in the first millennium ce. The earliest available versions of the Ramayana and the Mahabharata date to the second or third centu- ries ce (Thapar 2002). Trautmann (1971) and Thapar (2002) argue that the Arthashastra, though typically ascribed to Kautilya, a third-century bce royal advisor, was likely collated only in the second or third century ce. The Sangam literature of Tamil Nadu is more difficult to date but most likely was composed in the first half of the first millennium ce (Basham 1967; Thapar 2002).

    The Puranas, a collection of histories, myths, stories, and poems, are

    particularly difficult to date and interpret. They appear to have been col- lated in the mid-first millennium ce (Basham 1967; Singh 2008; Thapar 2000, 2002). As historical documents, the Puranas are notoriously dif- ficult to use (Kosambi 1965; Ray 1986; Thapar 2000; Trautmann and Sinopoli 2002). The historical and mythical components are mixed to such a degree that determining strict chronological and historical data is often impossible. In several key areas—the dynastic lists of the Satavahana state, for example—the Puranas present multiple versions of the same event. While problematic for developing chronologies, the Puranas do pro- vide a window into the social patterns of the early India, as long as it is remembered that this view is filtered through the mid-first millennium ce context in which the Puranas were written (see Thapar 2000 for a more detailed investigation of the historical value of the Puranas). In this way, the Puranas are no better, or worse, than the other early literary sources from India.

    A final source for the study of Indian Buddhism are mostly Persian accounts recording the military campaigns of Central Asians who began raiding northwest India in the beginning of the second millennium ce. By the end of the twelfth century ce, Central Asian Turks established the first Indo-Muslim state, the Delhi Sultanate, in North India. As will be discussed at length in Chapter 7, under the threat of the Turkish armies, most of the Buddhist monasteries in the Gangetic Plain were abandoned

    in the early thirteenth century ce. As with other historical accounts, the Persian sources must be read carefully (Eaton 2000). While some were written at the time of the conquests, many others are retrospective accounts written centuries later. Even those Persian accounts that date to the periods they refer were often hagiographies of particular rulers, valo- rizing their military accomplishments and overstating their Islamic faith. That said, when read carefully, Persian accounts provide genuine insight on the decline and collapse of monastic Buddhism in India.

    Travelers’ Accounts

    Foreign travelers’ accounts provide a different perspective on the Indian history. While they often contain obvious confusions on Indian prac- tices, in many cases they are well dated. The earliest of these sources record the campaign of Alexander the Great into northwestern India in 326 bce (Briant 2002; McCrindle [1926] 2000; Thapar 2002). Upon his departure from India, Alexander left governors to rule over the territo- ries he conquered. Many of the most useful Western accounts of early India are derived from court officials posted in these territories. The best of these sources is the Indika of Megasthenes, a Greek ambassador who traveled in India during the fourth century bce (McCrindle [1926] 2000). Unfortunately, Megasthenes’ writings are only preserved as long quota- tions in later, secondary works. Other accounts of India are found in the writings of numerous Greek and Roman authors (e.g., Ptolemy, Arrian, Strabo, Pliny, and Herodotos), but are generally based upon secondhand accounts and are of lesser value (see McCrindle [1901] 1979, [1927] 2000). Among the most important Greek sources is the Periplus of the Erythraean Sea (Casson 1989; Schoff 1974). The Periplus was a guide for Greek sea cap- tains trading on the west coast of India in the first century ce. It provides valuable information on the geography of the west coast, as well as politi- cal, economic, and social issues (see Ray 1999). While there is some dis- cussion of the main ports and geography of the east coast, it appears that

    most of this was reported secondhand.

    In addition to the Greek and Persian accounts are several important Chinese sources. These record pilgrimages to India by Chinese Buddhists beginning in the fifth century ce. Among the most important of these pilgrims were Faxian, Xuanzang, and Yijing. For the most part, Faxian’s travels in India (c. 399–412 ce) were limited to the Gangetic Plain in the north (Giles 1923; Liu 1988; Rongxi and Dalia 2002). A couple of centuries later (c. 629–645), Xuanzang extensively traveled through much of India,

    providing careful descriptions of numerous Buddhist establishments he visited along the way (Beal 1908; Rongxi 1997). Together, Faxian and Xuanzang’s accounts guided much of the archaeological research conducted in India in the nineteenth century (Trautmann and Sinopoli 2002). A few decades after Xuanzang visited India, Yijing spent about a decade (c. 675–685 ce) studying at the great monastery of Nalanda in North India (Rongxi 2000). While Yijing’s account does not provide the same broad-scale information found in Faxian’s and Xuanzang’s accounts, his extensive account of Nalanda provides impressive detail on one of the most important monasteries of later Indian Buddhism.

    While important, the Chinese pilgrim accounts have their own prob- lems. First, particularly in the case of Yijing, the accounts were written with a reformist agenda. Indian Buddhism, as presented in Chinese pil- grimage accounts, is treated as the true version of Buddhism that Chinese Buddhism had fallen from. In places, these accounts also seem to contain the inevitable cultural misunderstandings of any foreigner traveling to a distant land, wondering at seemingly mundane details. For example, Yijing’s account of life at Nalanda includes an extensive—and somewhat baffling—discussion of toilet practices that included detailed descriptions of fourteen balls of earth and a board or brick on which to carry them (Rongxi 2000:88).

    By the end of the first millennium and the beginning of the second millennium ce, pilgrims from China, Sri Lanka, Southeast Asia, and the Himalayas increasingly note the abandonment of Buddhist monasteries and pilgrimage sites. By the mid-second millennium ce, foreign accounts document attempts to refurbish and restore abandoned Buddhist pil- grimage sites in India, restorations that only began to succeed in the late nineteenth century ce.


    Higher criticism and the study of later Buddhist texts provide possibilities for the investigation of early Buddhism—possibilities of how and why cer- tain practices were followed, and possibilities of historical development that require independent confirmation. For some Buddhist scholars, early inscriptions have cast doubt on many of the interpretations of higher crit- icism. The value of inscriptions is well summarized by Schopen (1997:30):

    The inscriptional material has at least two advantages. First, much of it pre- dates what we can definitely know from literary sources. Second, and perhaps

    greater importance, this material tells us not what some literate, educated Indian Buddhist wrote, but what a fairly large number of practicing Buddhists actually did.

    The earliest readable inscriptions in India were carved at the direction of Ashoka in the third century bce. While they attest to the existence of early Buddhism, they do not provide significant detail on the forms that early Buddhism took at that time. Buddhist inscription become far more common in the first and second centuries bce. These inscriptions record the donations of money and architectural pieces to Buddhist monasteries and pilgrimage complexes by devout laity, guilds, monks, nuns, and royal families. Over the succeeding centuries, Buddhist donation inscriptions were carved at monasteries and pilgrimage sites throughout India. While often short, inscriptions provide a great deal of information on Buddhism that dates well before the surviving textual sources. In subsequent dis- cussions of Buddhism, I will employ information gathered from these inscriptions whenever possible. Typically, inscriptions will be discussed where they diverge from the textual sources. However, for the most part, textual and inscriptional sources agree on many of the basics of Buddhist thought. Often, inscriptions show that practices once thought to be later corruptions of Buddhism had greater antiquity than previously thought.


    The various textual sources for the study of Buddhism provide different insights and evidence for the study of Indian Buddhism. It is not neces- sary, or even desirable, to privilege any one source over others. The same goes for archaeological evidence of Indian Buddhism. Archaeological evidence for Buddhism shares some of the same problems that textual sources have, and some problems unique to the material context in which archaeology operates. An archaeological history of Indian Buddhism requires just as much concern with the archaeological evidence as it does with textual evidence.


    For the most part, archaeological research on Indian Buddhism has focused on religious architecture, to the detriment of studies of smaller, more mundane aspects of the archaeological record. Where archaeologists

    in many other parts of the world often focus on ceramics, lithics, fau- nal remains, and other more quotidian aspects of past societies, in the archaeological study of Indian Buddhism, architecture takes center stage. While I make use of smaller artifacts in my analyses, for the most part, this archaeological history primarily relies on the abundant information concerning the layout and construction of Buddhist religious institutions. Archaeological research in India has been conducted for more than two centuries (see Singh 2008 for a good introduction to Indian archaeology). In the nineteenth and early twentieth centuries, much of this research focused on early Buddhist monasteries and pilgrimage sites. Among the more important of the early researchers was Alexander Cunningham ([1854] 1997, [1876] 1962, [1892] 1998). As the first director general of the Archaeological Survey of India, he helped establish the systematic study of archaeology in India. Excavations under his direction at Sanchi, Bharhut, Bodh Gaya, and numerous other locations established archaeo- logical investigation of early Buddhism. Other researchers (e.g., Fergusson and Burgess [1880] 1988) continued these studies, revealing large num-

    bers of early Buddhist monuments and monasteries.

    In the first half of the twentieth century, research continued on early Buddhist sites, but a new emphasis on cities and large settlements began to develop. Major excavations were carried out at Arikamedu (Wheeler 1946), Taxila (Marshall [1951] 1975), and several other cities.3 Together, these allowed for a more nuanced understanding of the early Indian urbanism to emerge. In this research Buddhism was understood as part of a process of increasing urbanism, the development of regional exchange systems, and the establishment of craft guilds throughout India. At this time, a strong interest in the importance of trade between India and Rome developed, particularly after Wheeler’s excavations at Arikamedu revealed numer- ous artifacts with clear Mediterranean origin (Wheeler 1954). Wheeler’s interpretation of these remains as indicating a Roman trading colony on the east coast of India became central to understandings of the develop- ment of Indian history. While Wheeler’s claim that Arikamedu housed a Roman colony has now been cast into doubt (Begley 1983), a focus on the role of India in long-distance trade has continued to be among the central concerns of archaeological inquiry (see Ray 1994, 1999).

  3. Smaller excavations were also carried out at Kausambi (Sharma 1960, 1969), Hastinapura (Lal 1955), Rajghat (Narain and Roy 1976; Narain and Singh 1977), Nagarjunakonda (Ramachandran 1953), and numerous other locations. See Chakrabarti 1995 and Allchin 1995 for an overview of archaeological studies of the Early Historic Period.

In the latter half of the twentieth century, archaeological study of Indian Buddhism continued with the investigation of numerous sites across India, but particularly in Andhra Pradesh (Prasad 1994; Sastry et al. 1992; Subrahmanyam 1964) and Orissa (Chauley 2000; Nigam 2000). Much of this work has followed closely the work of Wheeler. Both Wheeler’s field methodologies and historical interpretations have, until recently, been relatively unchallenged. In particular, his excavations at Brahmagiri (Wheeler 1948) have served as the baseline for interpreting the ceramic sequence of most sites in peninsular India. Despite the advent of carbon-14 dating in the 1960s, most sites are still dated using Wheeler’s ceramic chronology. A recent reanalysis of the ceramics from Brahmagiri by Kathleen Morrison suggests that the periods assigned by Wheeler were too short (Morrison 2005). Morrison argues that, in most cases, wares persisted for several centuries longer than previously believed. If correct—and her use of carbon-14 dating makes this likely—many of the sites once thought to be firmly dated in India must be re-evaluated. Chronology, once thought to be more or less resolved in India, has re-emerged as a central concern among archaeologists. As such, I will pay inordinate attention to issues of dating throughout this book.

In addition to the archaeological excavations, the study of ancient Indian Buddhism has been significantly advanced by the work of several Indian historians, most notably Romila Thapar (1966, 2000, 2002) and Himanshu Ray (1986). Since about the year 2000, there has been renewed interest in the archaeology of Indian Buddhism by foreign scholars, including Robin Coningham at Lumbini (Coningham et al. 2013), Julia Shaw at Sanchi (2000, 2007, 2013b; Shaw and Sutcliffe 2001), Akira Shimada at Amaravati (2012; Shimada and Hawkes 2009), Jason Hawkes at Bharhut (2008, 2009), and my own work at Thotlakonda in Andhra Pradesh (Fogelin 2006).

Pilgrimage Centers

The material remains of Indian Buddhism are found, with very few excep- tions, in two general types of sites—pilgrimage centers and monasteries. From the outset, it should be noted that many pilgrimage centers con- tained monasteries, and many monasteries became the foci of pilgrim- age. Further, the distinctiveness of pilgrimage centers and monasteries varied significantly in different periods of Indian history and even region- ally during individual periods. These temporal and geographic trends are an important part of the archaeological history I present in this book.

The simple division between pilgrimage centers and monasteries pre- sented here is only intended to facilitate an initial understanding of the range of archaeological evidence available for the study of ancient Indian Buddhism.

The earliest Buddhist pilgrimage sites are presumed to have been built in the Gangetic Plain in the centuries after the Buddha’s death. The four most important pilgrimage sites mark the key locations in the biogra- phy of the Buddha—his birth at Lumbini, his enlightenment at Bodh Gaya, his first sermon at Sarnath, and his death (and release from the cycle of rebirth) at Kushinagar. After the rediscovery and excavation of these sites in the nineteenth century, these four sites have become the focus of pilgrimage by Buddhists from around the world. While it is presumed that all mark genuine locations of the events they commemo- rate, Coningham (2001) has convincingly argued that, with the possible exception of Lumbini, none has Buddhist archaeological materials dat- ing earlier than the third century bce, two centuries after the Buddha’s death. This does not mean that Bodh Gaya, Sarnath, and Kushinagar were not pilgrimage centers prior to the third century bce, only that there is no archaeological evidence for it. As for Lumbini, as will be discussed in Chapter 3, recent excavations led by Coningham and Acharya have revealed a possible early Buddhist temple dating to the sixth century bce (Coningham et al. 2013).

The Sri Lankan Mahaparinibbana-sutta (Davids and Davids [1910] 2007) and other Buddhist textual sources record that after the Buddha’s death, his disciples cremated his body and gave his ashes to eight kings to place in large mounds of earth erected near crossroads. These earthen mounds were called stupas, and burial in stupas was the typical manner for interring the remains of important individuals of many religions at the time (Shimada and Hawkes 2009). Over the subsequent centuries, stu- pas became the principal foci of Buddhist ritual (see Figure 1.1), though Buddhists also worshiped at trees and other objects in some instances (e.g., the Bodhi tree at Bodh Gaya). Textual sources record that in the third century bce the Mauryan King Ashoka disinterred and redistrib- uted relics from seven of the eight prototypical stupas, creating thousands of more elaborate stupas made of brick, stone, and stucco. Relying on these accounts, archaeologists have long sought to identify the original eight stupas in which the Buddha’s remains were interred, particularly the fabled eighth intact stupa. As will be discussed in Chapter 3, none of the alleged ancestral stupas identified by archaeologists is reliable, since none can be clearly dated to the times of the Buddha. What can be said, how- ever, is that after the third century bce, stupas became the primary focus

Figure 1.1: Buddhist pilgrimage center at Sanchi (c. third century bce–thirteenth century ce)

Courtesy of the Digital South Asia Library and American Institute of Indian Studies (Accession No. 27).

of Buddhist pilgrimage in India, with large pilgrimage centers built sur- rounding them (Shimada and Hawkes 2009).

Mitra (1971:21–22) has identified four broad types of stupas in India. The first were those that contained the cremated remains of either the Buddha or one of his principal disciples. The second contained the Buddha’s material possessions (e.g., begging bowl or robes). The third type of stupa marked the location of key moments in the Buddha’s biography (birth, first sermon, death, etc.). Finally, surrounding many primary stupas (of the above three types) were numerous votive stupas—small stupas con- taining the cremated remains of devotees. Votive stupas allowed devotees to engage in perpetual worship of the Buddha, even after death (Schopen 1997:ch. 7). While valuable, Mitra’s division of stupas into four distinct categories is also problematic. In practice, individual stupas often blended elements of different categories. Stupas erected on locations associated with the Buddha’s life often contained relics (e.g., Sarnath; Cunningham [1854] 1997). The stupas of key disciples were sometimes erected near Buddha stupas, with the disciples’ stupas serving simultaneously as votive stupas to the Buddha and as the focus of worship themselves. Finally, it must be noted that stupas were not always the central focus of pilgrimage.

At Bodh Gaya, ritual focused on the Bodhi tree, a descendant of the tree under which the Buddha gained enlightenment. At Lumbini, pilgrims focused their attention on the location where the Buddha was born, today located inside the Maya Devi Temple. While less common than pilgrim- age sites centered on stupas, some sites outside the Gangetic Plain were also constructed around trees, rocks, or other ritual foci. For example, a descendant of the Bodhi tree from Bodh Gaya is one, of several, foci of pilgrimage to Anuradhapura in Sri Lanka.

Whether trees, rocks, or stupas, the primary form of ritual at Buddhist pilgrimage centers consisted of individuals walking around (circumam- bulating) the foci, accumulating merit and ultimately speeding their path to nirvana. While circumambulatory paths (pradakshina) were often large enough to hold many worshipers simultaneously, circumambula- tion was ultimately an act of worship between an individual and the Buddha—often represented by the Buddha’s relics contained within the stupa. As such, circumambulation was one of the individualizing elements of early Buddhism (Fogelin 2003). Over time, Buddhists began to alter the architectural layout of ritual complexes to allow both individual and group worship of stupas. Group ritual consisted of collective worship by multiple people in courtyards surrounding the stupas and pradakshina (circumambulatory paths).

A pilgrimage center, almost by definition, required the presence of a stupa or similar ritual foci. But two other types of structures were also commonly built at pilgrimage centers. The first were chaityas, apsidal halls with a stupa placed at one end (see Figure 1.2). Chaityas served as subsidiary shrines to the central ritual foci. The second were buildings (usually square) with small rooms arrayed around a central courtyard (see Figure 1.2). These were Buddhist monasteries, or viharas. As will be discussed in Chapters 5 and 6, it is likely viharas were later additions to pilgrimage centers, with most constructed in the early to mid-first mil- lennium CE. Prior to this, viharas, for the most part, were restricted to Buddhist monasteries.


If the defining element of a pilgrimage center is the central stupa or ritual foci, the defining element of a monastery is the vihara. Without a place to house the sangha, a monastery is simply not a monastery. But, just as pilgrimage centers often have secondary structures built around them, the sangha often constructed stupas—both within chaityas and

Figure 1.2: Monastery 45 (c. tenth century ce) and Temple 18 (seventh century ce) at Sanchi (not to scale; after Marshall and Foucher 1983)

as freestanding structures—adjacent to their viharas. In earlier periods, these subsidiary stupas were likely intended to draw the Buddhist laity to worship at the monasteries, though in later periods worship at monastic stupas was likely limited to the sangha themselves. In general, Buddhist monasteries became larger and their layout more homogenous over time. Where the earliest monasteries likely housed no more than 100 monks, later monasteries like Nalanda in the Gangetic Plain had resident popula- tions in the thousands.

Buddhist textual sources suggest that the Buddha established monas- teries and nunneries during his lifetime to serve as rainy season retreats. Ashokan inscriptions from the third century bce indicate that at least some Buddhist monasteries existed at that time. Two possible Buddhist monasteries, Lomas Rishi and Sudama, date to this period (Fergusson and Burgess [1880] 1988; Mitra 1971). Unfortunately, it is unclear if Buddhists or the Ajivikas built Lomas Rishi and Sudama, or even if the Ajivikas were a distinct or a divergent sect of Buddhism or Jainism (Singh 2008). With the possible exception of Lomas Rishi and Sudama, then, the earliest archaeologically known Buddhist monasteries in India date only to the first century bce. The best preserved of these monasteries are found in the Western Ghats of peninsular India (see Figure 1.3). Their preservation is a product of the medium of their construction—the sangha had these monasteries carved into cliff faces. These rock-cut monasteries provide

Figure 1.3: Bhaja rock-cut monastery in the Western Ghats (c. first century bce–seventh century ce)

Photo by author.

an excellent venue to examine the spatial arrangement of early Buddhist monasteries due to their almost perfect preservation. Visiting these sites feels almost like walking into a fossilized monastery. Monastic cells within the viharas still contain rock-cut beds and wall niches the former residents would have used. The nearby chaityas are also almost completely intact. Given this, the Buddhist monasteries of the Western Ghats have long served as the primary foci for scholars seeking to understand the nature of early Buddhist monasticism.

In the late nineteenth century, the rock-cut temples of the Western Ghats were excavated and mapped by Fergusson and Burgess ([1880] 1988). To this day, their precise elevations and ground plans inform much of the scholarship on early Buddhist monasticism. Fergusson and Burgess’s view of the monasteries, however, was primarily informed by readings of Buddhist monastic literature that depicted monasteries as rainy season retreats—locations where the Buddhist sangha could isolate themselves and engage in extended meditation. Like other early British archaeologists (e.g., Brown 1965; Cunningham [1854] 1997), Fergusson and Burgess believed that Buddhist monasteries were supported by royal donations. Over the twentieth century, the perception of isolation and royal patronage has gradually shifted to a view of Buddhist monas- teries as more active and engaged with worldly concerns. Thapar (2002)

and Ray (1986) noted that much of the support for Buddhist monaster- ies was derived from craft guilds rather than the royalty. Thapar, while continuing to see monasteries as isolated retreats, interpreted these gifts as evidence that monasteries were the recipients of donations resulting from competitive giving between guilds, lesser elite, and the royalty. In contrast, Ray saw Buddhist monasteries as more actively engaged in eco- nomic activities, serving as nodes on long-distance trade networks (see also Lahiri 1992; Morrison 1995) and managing agricultural production in the peripheries of developing states (see also Heitzman 1997). Recently, Shaw (2000, 2007, 2013b) has argued that Buddhist monasteries fostered and promoted the construction of irrigation systems to establish patron- age systems with local populations. In my own work, I have presented Buddhist monasteries as precariously balancing their public and private religious obligations with their need to provision and sustain the resident population (Fogelin 2006).

In subsequent chapters I will address the various and changing roles of monasteries, both in terms of Indian Buddhism and in terms of broader social and economic concerns. Rather than view any one of these expla- nations of Indian Buddhism as the best, I will argue that at different times, and in different places, Buddhist monasteries fulfilled all of these roles and more. I will also address the ways that the design and layout of Buddhist monasteries impacted the sangha themselves. Whether living in retreats or trade centers, the Buddhist sangha required mechanisms to reduce tensions and foster group cohesion within the close confines of the monasteries.

Buddha Images

Beginning in the second and third centuries bce, Buddhist iconography becomes available for archaeological study. The earliest Buddhist imagery is found on the railings defining the pradakshina at second century bce pilgrimage centers and, perhaps, columns erected by Mauryan kings in the third century bce. Early Buddhist iconography is rich and detailed, with numerous depictions of stupas, people worshiping the Buddha, and events in the Buddha’s life. What is not depicted, however, is the Buddha himself. Prior to the beginning of the first millennium ce, depictions of the Buddha were limited to standardized symbols of his presence, includ- ing footprints, empty thrones, trees, and parasols. The earliest Buddha images in India were sculpted in Gandhara and Mathura in the first and second centuries ce. From this location in the northwestern periphery

Figure 1.4: First or second century bce relief from the North Gate of Sanchi and fourth–seventh century Gandharan Buddha image

Both images courtesy of the Digital South Asia Library and the American Institute of Indian Studies (Accession Nos. 40173 and 20510).

of India, the tradition of Buddha images spread across other portions of India and beyond. By the mid-first millennium ce, Buddha images were common at both pilgrimage centers and monasteries throughout India (see Figure 1.4).

Beginning in Chapter 5, I extensively examine the origin and devel- opment of Buddha images in India. Buddha images were first carved at roughly the same time as Mahayana Buddhism began to develop. The earliest Buddha images, however, did not depict Mahayana figures—they were depictions of biographical events in the life of the Buddha consis- tent with the theology and doctrines of early Buddhism (Leidy 2008; Schopen 2005:11–12). That is, where earlier Buddhist iconography would depict the Buddha through symbols of his presence, the earliest Buddha images retained the earlier style, but replaced the symbols of the Buddha with images of the Buddha. While Buddha images demonstrate that the earlier taboo on depictions of the Buddha was being challenged, they do not signify the advent of Mahayana Buddhism. The earliest images with Mahayana elements might date to the second or third century ce, with Mahayana images becoming common only in the fifth and sixth centuries ce (Bareau 1985; Leidy 2008; Schopen 2005:11–12). This is not to say that the origin of Buddha images and Mahayana Buddhism are unrelated. They both testify to a shift in the practice of Buddhism between the first and

fifth centuries ce. Both Buddha images and Mahayana Buddhism signify a greater emphasis on the person of the Buddha and the beginning of a new ascetic ideal within Buddhism.

The Limits of Archaeological Evidence

While archaeology has much to offer the study of Indian Buddhism, there are some issues on which it is mostly mute. Most important, currently there are no Buddhist remains that date to the time of the Buddha, and very few that date to the two centuries after his death. In fact, with the exception of Lumbini (Coningham et al. 2013), no archaeological materials can be unequivocally shown to both be Buddhist and date to a period prior to the third century bce. Given this, archaeology cannot be used to study Buddhism in the time of the Buddha or the centuries immediately follow- ing his death. As I have argued above, the same can be said of the Buddhist textual sources, which date to the beginning of the first millennium ce. Without significant textual, inscriptional, or archaeological evidence to employ, an archaeological history of Indian Buddhism can only begin in the third century bce, some two centuries after the Buddha’s death. This does not mean that archaeology has nothing to say about the fourth, fifth, and sixth centuries bce. There are numerous archaeological sites dat- ing to this period. It’s just that very little in these sites is unequivocally Buddhist. Archaeology can inform on the general social context in which Buddhism emerged, but it cannot inform on the specifics of Buddhism during that period.

Another problem with the archaeological sources is the lack of research on domestic settings. With few exceptions (e.g., Marshall [1951] 1975; Smith 2001), archaeological research has been limited to the pilgrimage centers and monasteries of India. Buddhist textual sources frequently note that the sangha would perform rituals in the homes and villages of the laity. Without excavations of domestic contexts, archaeology cannot contribute to a study of domestic Buddhist ritual. It would undoubtedly be interesting and informative to compare the ritual practices of the Buddhist sangha and the laity, to see the points of similarity and disjuncture. With luck, future excavations can focus on the more quotidian lives of ancient Indians, but for now this hole in Buddhist history cannot be filled.

There is one final absence in the archaeological record that is more dif- ficult to ignore. At present, no archaeological evidence for Buddhist ascet- ics has been found in India. There are many caves that are alleged to have been used by the Buddha, his disciples, or other prominent ascetics, but

none has been properly excavated or has inscriptions dating to the time of their use. For example, Faxian notes in his fifth-century ce chronicle of his travels in India that several natural caves near the city Rajgir were used by ancient ascetics, and that the caves were the focus of pilgrimage at the time he visited them. But no archaeological or epigraphic evidence confirms Faxian’s claims. While it is possible that the evidence of Buddhist asceticism has not yet been found, or that the structures used by Buddhist ascetics were too perishable to have survived to the present day, at some point negative evidence begins to actually mean something.

The absence of evidence for Buddhist ascetics in India becomes more problematic when viewed in terms of the ample evidence for early asceti- cism in the rest of the Buddhist world. In Sri Lanka, for example, caves were used by early Buddhist ascetics from, perhaps, as early as the third century bce (Carrithers 1983:6; Coningham 2001). In Thailand, the archaeological signature of Buddhist asceticism is similarly easy to find, with the caves and other material remains of forest monks dating as early as the sixth and seventh centuries ce (Munier 1998:34). Given the rela- tive ease with which Buddhist asceticism has been found in other parts of the Buddhist world, the lack of evidence of Buddhist asceticism in India becomes more notable. Among the central arguments of this book is that, contrary to the standard history of Buddhism, asceticism was a late devel- opment in Indian Buddhism, and existed more as an ideal to think about than as an actual thing the sangha did. Lacking any archaeological evi- dence of Buddhism prior to the third century bce, and arguing from nega- tive evidence for the rest of the sequence, this will be difficult.


In 1983, Eric Hobsbawm and Terrence Ranger published an edited vol- ume titled The Invention of Tradition. In the volume, Hobsbawm, Ranger, and the other contributors argued that much of what is presented as authentic history consists of “responses to novel situations which take the form of reference to old situations, or which establish their own past by quasi-obligatory repetition” (Hobsbawm 1983:2). For example, where the coronation ceremonies of the British royalty are presented as the time- less repetition of ceremonies of old, in reality these ceremonies date to only the late nineteenth and early twentieth centuries—by which time the British royalty served as little more, perhaps even less, than figure- heads in the British political system. In earlier times, when British mon- archs actually had power, coronation ceremonies were more informal and

ad hoc. In this sense, the more recent “invented” coronation ceremonies had almost no precedent in history. Critically, however, the invention of tradition does not require that a tradition be wholly novel, but can also include the appropriation of a genuine tradition in novel ways. For exam- ple, the neoclassical design of the US Capitol building borrowed design elements of Greek and Roman architecture to invoke a tradition of clas- sical architecture and governance. In reality, of course, both Europeans and the Euro-Americans who cherished neoclassical architecture were not the descendants of the Romans, but rather the descendants of the people who sacked Rome, bringing the Western Roman Empire to an end. The key here is that the US Capitol Building could have referenced many other his- torical precedents—it could have been designed to resemble Stonehenge, for example. The choice of classical traditions over other traditions signi- fies an invention of tradition as much as creating a new tradition out of whole cloth.

By the mid-1990s, archaeologists had taken up the invention of tradi- tion, though often in terms of the broader concept of “memory” (e.g., Mills and Walker 2008; Van Dyke and Alcock 2003):

People remember or forget the past according to the needs of the present, and social memory is an active and ongoing process. The construction of social memory can involve direct connections to ancestors in a remembered past, or it can involve more general links to a vague mythological antiquity. (Van Dyke and Alcock 2003:3)

The value of the invention of tradition or memory for archaeologists lies in its clear material manifestations. Whether the regalia and crown of British coronation ceremonies or the stone columns of the US Capitol building, memory is inscribed in material objects, monuments, and land- scapes. This is not to say that all elements of memory are material. They aren’t. Rather, it is only possible to say that memory and the invention of tradition often have material manifestations, and this materiality does not simply reflect inventions of tradition, but actively facilitates them.

Among the central arguments I make in this book is that from the third century bce—the earliest period for which archaeological, epigraphic, and textual evidence is available—the Buddhist sangha was fully domesti- cated, living year-round in well-established monasteries. Buddhist asceti- cism was invented only in the early to mid-first millennium ce through the writing of new Buddhist texts, the building of new Buddhist mon- asteries, and the carving of new Buddha images. The earliest Buddhist texts were written in the beginning of the first millennium ce to help

invent the ascetic ideal and project that ideal onto the Buddhist past. At the same time, Buddhist monasteries were redesigned and Buddha images were created to construct new forms of social memory that justified a new emphasis on monastic isolation. This does not mean that Buddhist monks all became solitary ascetics. Rather, the invention of the ascetic ideal was part of a collective withdrawal by the sangha from day-to-day interaction with the Buddhist laity—a withdrawal facilitated by monasteries’ sub- stantial wealth and the ownership of large tracts of arable land. By cele- brating an invented history of solitary asceticism, the sangha legitimized a newfound collective asceticism practiced within the walls of monasteries that were becoming isolated from the laity for the first time. In the end, however, the collective asceticism of the sangha planted the seeds that led to the collapse of monastic Buddhism in the second millennium ce. When the sangha withdrew from day-to-day interaction with the laity, the laity returned the favor by converting to Hinduism and other faiths. With the loss of monastic lands and royal support at the hands of Central Asian Turks in the second millennium ce, monasteries no longer had a laity to fall back on. Without monasteries, Buddhist monasticism failed in India, with the last vestiges of lay Buddhism and Buddhist pilgrimage centers following soon thereafter.

In the rest of this book I attempt to provide the necessary evidence and

detail to support the revisionist history just described. It is a complicated argument, relying on widely divergent lines of evidence from textual, epi- graphic, archaeological, and art historical sources. It also relies on archae- ological methodologies, both well established and newly created for this project. Given the importance of these methods, all of Chapter 2 is taken up by introducing and explaining them. I return to India in Chapter 3, using the tools developed in Chapter 2 to begin the process of building an archaeological history of Indian Buddhism.




The Material of Religion


nthropologists have long complained of the faddishness of anthropo- logical theory. Rather than a slow accumulation of knowledge with a corresponding increase in explanatory potential, it seems that each new generation is rewarded for creating new theories, new questions, and new critiques of past work—whether or not the new theories are particu- larly new and whether or not the older theories had any real deficiencies.

Anthropologists dismiss older theories out of hand, often for little more reason than that they are old. As argued by Sahlins (2002:73–74),

In the social sciences, paradigms are not outmoded because they explain less and less, but rather because they explain more and more—until, all too soon, they are explaining just about everything Paradigms change in the social

sciences because, their persuasiveness really being more political than empiri- cal, they become commonplace universals. People get tired of them. They get bored.

Anthropologists do not reject older theories because they are wrong; anthropologists embrace newer theories because they allow for the study of interesting new questions and concerns. By this view, anthropology has not progressed by replacing bad theories with good theories, but rather by replacing boring theories appropriate for the study of one aspect of human life with interesting theories for the study of a different aspect of human life.

To the degree that subsequent generations of anthropologists limit themselves to the study of whatever new questions and concerns are in vogue, there is no real harm. There are far worse sins than anthropologists

( 34 )

studying things they find interesting, using theoretical perspectives well suited to their investigation. The problems arise when anthropologists attempt to use the latest theory du jour to explain “just about everything,” to study questions and concerns best approached with older, boring theo- ries well suited to those older, boring issues. Where an anthropologist is investigating a complex, multifaceted question, the reliance on a single theoretical approach is an impediment to understanding. The long-term development of Indian Buddhism is one of those complex, multifaceted questions that cannot be approached using only the latest theory du jour. It must be approached using multiple theoretical perspectives—old and new—each applied to the specific contexts to which they are best suited.

Whatever the theoretical perspective, to a large extent, the modern study of religion has consisted of a long conversation with Karl Marx. It is not just that many of the modern understandings of religion can be traced back to Marx, but rather that even the inconsistencies and prob- lems within Marx’s work continue to be the points of contention within modern debates about the nature of religion. In his writings, Marx some- times portrayed religion as the product of the material and economic conditions of people within any given society or, in other writings, he described religion as a strategy by which the elite legitimize their privi- leged social position. In the first view, religion is the product of human agency in relation to people’s real material circumstances. In the latter view, religion is imposed from above, with those people following religious practices losing a significant portion of their agency. This same tension is found in most modern theoretical approaches to religion. Where some theorists emphasize the creative agency of people in creating or manipu- lating religious practices, others take a more structural approach, view- ing abstract religious principles or cosmologies as more determinative of human behavior. Marx attempted to resolve this tension through calls to the dialectic, arguing that the contradictions in the material foundation of society create imbalances and the possibility of radical social change. In this sense, Marx shaped most modern debates concerning religion, with different theorists employing dialectical thinking to place themselves along a continuum between structure and agency, stasis and change, and tradition and revolution.

In this chapter I introduce the theoretical perspectives that I employ in my analyses of Indian Buddhism. As I introduce these theoretical perspectives, I also present the specific material indicators that allow these theories to be applied to archaeological contexts. Given his cen- trality to almost all subsequent approaches, I begin with a discussion of Marx and the neo-Marxists of the twentieth century. I then move to a

discussion of Max Weber and Emile Durkheim. In some sense, this is the point where the two halves of Marxist theory began to diverge. Today, those theorists who emphasize practice theory and materiality trace their lineage back to Marx through Weber and the neo-Marxists of the Frankfurt School. Anthropologists with a more structural bent trace their Marxist lineage through Durkheim. I complete my discussion of previous theorists with the semiotic approach of Charles Sanders Peirce. In some sense, Peirce is the one theorist I discuss who lays outside the dialectical framework that Marx established, though I argue that his theories can be productively incorporated within it.

I conclude the chapter with an extended discussion of the fundamental incoherence of religion. By this view, religions are shot through with theo- logical contradictions, social disjunctures, and periodic ruptures that con- stitute the history of any specific religion. It is this incoherence, I argue, that necessitates the use of multiple theories that can address the discor- dant elements. If Indian Buddhism is incoherent, then the study of Indian Buddhism can only benefit from the incorporation of multiple theoretical perspectives, each capable of explicating contradictory religious practices and beliefs.


In studying ancient religion, archaeologists have appropriated many of their foundational understandings from nineteenth- and early twentieth-century anthropology, religious studies, and sociology. In the process, archaeologists have been forced to identify the material implications of what are, for the most part, more intangible theoreti- cal perspectives. In Chapter 1, I discussed the recent revival of interest in the archaeological study of religion. Here, in conjunction with dis- cussions of specific theories of religion, I begin to discuss how religion can be investigated in an archaeological context—a context in which material remains are the primary source of evidence. The following is not intended to be an exhaustive account of all of the methods I will employ in my investigations of Indian Buddhism. For the most part, I will introduce interpretive methods at the first instance I employ them. Here I will only discuss the most ubiquitous and fundamental methods that I employ throughout the analysis. Prior to this, however, one criti- cal assumption underlying the archaeological study of religion must be addressed—religion is something that people do, not just something that people think about. Archaeological remains, of any kind, are the

product of human actions. While these actions may be guided by reli- gious beliefs, human action initially puts an artifact in one place rather than another, erects a temple in one form rather than another. While not all human actions leave material traces, many do. After more than a century of developing and refining approaches to studying that limited subset of human actions that leave material traces, archaeologists have become very good at using fragmentary material evidence to flesh out the lives of past people.

The goal of the archaeological study of religion, then, is the identifi- cation of specific material indicators of religious actions in the past. As discussed more extensively below, often these actions are ritual in nature. Once established, ritual and other religiously motivated actions can be interpreted through the use of more general theoretical understandings of religion. The trick is to determine methodological approaches that con- cord with varying theoretical approaches—that illuminate human actions consistent with varying theoretical understandings. Rather than discuss all of this in the abstract, it is best to illustrate this in reference to spe- cific theories of religion. As most of these understandings are ultimately derived from the work of Karl Marx, I begin there.

Marx, Marxism, and Neo-Marxism

Marx’s fundamental subject of inquiry was the economic basis of nineteenth-century European society. At the core of Marx’s conception of capitalism are the dual concepts of infrastructure and superstruc- ture. In A Contribution to the Critique of Political Economy ([1859] in Elster 1986:187), Marx argued that “[t]he sum total of these relations of produc- tion constitutes the economic structure of society, the real foundation, on which rises a legal and political superstructure and which correspond to definite forms of social consciousness.” In other texts (e.g., The Eighteenth Brumaire of Louis Bonaparte [1852] 1964), Marx places greater emphasis upon the dialectic between the economic base of society and the super- structure. However, while Marx does note the possibility of the super- structure conditioning the economic foundation, his writings on religion were generally more unidirectional (Elster 1986:504–510).

The class that has the means of material production at its disposal, conse- quently also controls the means of mental production, so that the ideas of those who lack the means of mental production are on the whole subject to it.

The ruling ideas are nothing more than the ideal expression of the dominant material relations. (The German Ideology [1845–1846] 1998 in Elster 1986:302)

In a strict Marxist framework, religion is superstructural, and there- fore secondary. By this view, religion only served to legitimize elite power and domination. While Marx was consistent in his belief that religious ideology developed from the material realities of social relations, his writ- ings are often contradictory about the mechanisms for the emergence and development of the ideologies. In The German Ideology ([1845–1846] 1998), he sees ideology as forced upon the working classes. In other writ- ings (e.g., the introduction to A Critique of Hegel’s Philosophy of Law [1844] 1970), Marx argues that the working class generates religious ideology as a means of coping with their subordinate social position. In either case, Marx saw religion, particularly Christianity, as supporting the interests of the elite, legitimizing their privileged status. After Marx’s death, in 1893, Engels coined the term false consciousness to describe the ideology of capitalism.

The concept of false consciousness was refined and extended in the early twentieth century in the work of Gramsci (2011), Adorno (Horkheimer and Adorno 2007), and other Frankfurt School scholars who sought to explain how modern capitalism had become more entrenched, rather than less, in the twentieth century. These scholars saw false consciousness (reconceived as hegemony by Gramsci) as a powerful force that natural- ized the ideology of the elite among the working class. To some degree, the work of Gramsci, Adorno, and other neo-Marxists demonstrated that ideology was stronger and more determinative than Marx had realized. Rather than viewing religion as derived from of the economic conditions of society, neo-Marxists recognized that religion could create and pre- serve economic realities.

The continued centrality of the Marxist and especially neo-Marxist views of religion in subsequent theoretical approaches cannot be over- stated. For example, Bloch (1989:45) states that ritual is “the exercise of a particular form of power . . . [that makes] a power situation appear a fact in the nature of the world.” Similarly, much of the literature on prac- tice theory emphasizes the ways that rituals promote the development of relationships of power (Bell 1992; Comaroff 1985; Ortner 1989). Bell (1997:82, italics in original), in a summary of practice theory, argues that “rather than ritual as the vehicle for the expression of authority, practice theorists tend to explore how ritual is a vehicle for the construction of rela- tionships of authority and submission.”

Max Weber

While similar to and clearly influenced by Marx, from the start Max Weber recognized that religious practices could condition economic foundation of society and vice versa. In The Protestant Ethic and the Spirit of Capitalism ([1905] 1958), Weber argued that the emergence of Protestantism in the sixteenth century was a necessary precondition for the rise of modern rational capitalism in the seventeenth century, but that modern ratio- nal capitalism was only one of many possible economic systems that could have arisen. Rather than Protestantism causing modern rational capitalism, Weber argued that there was an “elective affinity” between Protestantism and modern rational capitalism. Thus, The Protestant Ethic and the Spirit of Capitalism illustrates two of the most characteristic ele- ments of Weber’s research. First, rather than make general proclamations about religion, Weber was more concerned with the origin and function of specific religions in specific historical contexts. Second, Weber’s approach is less determinative than Marx’s, instead emphasizing the vagaries of historical processes. As such, Weber’s approaches are not as easily gener- alizable as Marx’s. Weber’s approaches inspire rather than strictly codify research on religion. Where Marxist and neo-Marxist scholars argue that religion does this or that, Weber argues that religion may do this or that, depending on the specific historical context in which it is practiced.

Perhaps the most emblematic and, in some sense, simple element of

Weber’s approach is his recognition that attempts at legitimacy may or may not succeed (Weber 1978:31–38, 212–215). Following Weber, it is impor- tant to distinguish legitimations (attempts at legitimacy) from domination or legitimacy (the state in which a broader population accepts one group’s legitimations as valid). In contrast to Weber, Marxists and neo-Marxists often assume, prima facie, that religion successfully legitimizes domina- tion, promotes false consciousness, and reinforces hegemony. Today, we know that this simple understanding of domination is false. In numer- ous cases—liberation theology in South America, for example, or the civil rights movement in the United States—religion legitimized resistance. When examining other cases—roadside shrines memorializing traf- fic fatalities, for instance—it does not seem that legitimization of elite authority plays any significant role at all. As such, Weber’s concepts of legitimations and legitimacy are more dynamic and useful than false con- sciousness, domination, and hegemony—if only because legitimations are actions that people do, actions that can fail. The difference in perspective between Weber and Marx can best be illustrated through an examination

of how Weber explained the establishment of new religions and political movements.

For Weber, the routinization of charisma is a process by which a char- ismatic leader of a new social movement, such as the Buddha, is trans- formed into a legitimizing figure of domination. Initially, charismatic leaders challenge the existing social norms and power structures of the society by creating relationships with his or her followers that are “strictly personal” (Weber 1978:246). These relationships, however, are unstable in the long run. With the death of the charismatic leader, the relationships inevitably deteriorate. In order for the community to persist, something must be done to reintegrate the radical social movement into broader social traditions. This, of course, necessitates changes in the community of people who follow the original leader, as well as modifications to the social order from which it originally diverged.

As conceived by Weber (1978), the routinization of charisma is a pro- cess by which a subversive ideology and leader are transformed into a legitimizing ideology of authority. The followers of the charismatic leader can establish new rules of legitimate succession (e.g., hereditary). They can codify the teachings and appoint certain individuals to legitimately adju- dicate disputes over these teachings. They can identify locations associ- ated with the charismatic leader and control access to these locations. In this sense, all of the techniques that people employ to routinize charisma are legitimations—attempts by certain groups of followers to attain legit- imacy through association with the original charismatic leader. The act of transferring the authority of a charismatic leader, however, cannot be assumed to be either simple or uncontested. Groups who oppose the rou- tinization of a charismatic leader and divergent groups within the leader’s followers compete and contest different attempts at routinization. As with any legitimation, attempts to routinize charisma can fail.

Weber’s understanding of routinization concords well with what is known of the history of Indian Buddhism.1 In Indian Buddhism, kings, commoners, and divergent Buddhist sects all competed to be the right- ful heirs to the charisma of the Buddha. At different times and in differ- ent places, different groups of Buddhists gained or lost legitimacy, gained or lost the authority that association with the Buddha could confer. For this reason, I find Weber’s approach to legitimization more valuable than

  1. It should be noted that Weber developed his thoughts about the routinization of charisma in part through studies of Buddhist, Muslim, and Christian history (Weber 1978). In this sense, it is not surprising that the sectarian struggles within Indian Buddhism concord with Weber’s theories.

Marxist or neo-Marxist approaches. Throughout the following analysis, I use legitimations to refer to the attempts by some to legitimize their position vis-à-vis the Buddha. I use legitimacy to describe those times when specific legitimations were successful.

If archaeologists decide to study legitimization, what sorts of meth- ods and material evidence should they employ? What sorts of evidence can they use to determine how or if legitimization was employed in the past? First, following Weber, I argue that the archaeologists should ini- tially seek to identify legitimations—attempts at legitimacy, rather than legitimacy itself. The construction of a temple is a legitimation by whoever builds it. Whether that legitimation succeeds is another question, requir- ing further evidence and interpretation. But what sorts of evidence can archaeologists use to initially identify whether a specific material trace is a legitimation in archaeological contexts?

In an extended study of ancient ceremonial architecture in coastal Peru, Jerry Moore (1996:2) argued that “public buildings are physical testimonies of the use of power.” Following a practice perspective, Moore suggests that ceremonial architecture did not simply reflect authority; it was instrumental in creating authority. Though Moore does not use the term himself, archaeologists can productively view public build- ings as legitimations. Moore employed five material criteria to evalu- ate how, and if, ceremonial structures legitimized elite authority (Moore 1996:139–141):

Permanence: Is the structure made of durable (e.g., stone) or less durable (e.g., wood) materials?

Centrality (Proximity): In order for legitimization to occur, peo- ple should locate a religious structure in a position associated with the people being legitimized—adjacent to a palace or in the capital city, for example. Moore refers to this as centrality, though I prefer proximity (see discussion in text).

Ubiquity: The construction of buildings in multiple locations throughout a territory can also serve to legitimize domination (e.g., parish churches in every village).

Scale: The larger the structure, the more it will legitimize the per- son or people who constructed it.

Visibility: A religious structure should be visible to those people from whom acquiescence to authority is desired.

Each of Moore’s criteria could be analyzed and unpacked. For example, visibility does not mean that everyone must be able to enter and view the

entire structure. The elite can reinforce authority by restricting access to sacred places—but only if everyone knows it and can see it. It might even be the case, paradoxically, that as rituals of legitimization become more exclusive and secret, religious structures would require larger, more prom- inent façades (Fogelin 2008c). Centrality, to my mind, is somewhat of a misnomer. While building a temple next to a king’s palace in the capital would associate the king with the divine, a king might also build promi- nent structures adjacent to an isolated cult center. Here a king attempts to legitimate himself through association with a pre-existing center of reli- gious importance. For this reason, I prefer to substitute the term “proxim- ity” for “centrality” in Moore’s criteria. That is, legitimations should serve to associate an authority with the divine.

Moore never intended his criteria to be exhaustive, nor were all criteria necessary for a past structure to be considered a legitimation. In Japan, for example, the royalty often constructed important imperial temples of wood. Similarly, other criteria could easily be added. Exclusivity, for exam- ple, seems to be a critical element of Buddhist monastic legitimations. The point here is that legitimations leave material traces in the archaeo- logical record—material traces that archaeologists can employ to study legitimations in the past.

Emile Durkheim

Emile Durkheim was a contemporary of Max Weber. Where Weber extended and refined Marxist understandings of legitimate domination, Durkheim drew on Marxist views concerning the division of labor and false consciousness to investigate the ways that religion could serve to unify communities. In Durkheim’s terms, religion promoted feelings of social solidarity among members of a community. Before I can discuss social solidarity, however, I must note one of the major flaws in Durkheim’s theories—his mostly mistaken distinction between the sacred and the profane ([1915] 1995:44).

All known religious beliefs . . . present one common characteristic: they pre- suppose a classification of all things, real and ideal, of which men think, into two classes or opposed groups, generally designated by . . . the words sacred and profane.

Ever since Durkheim first penned these words, scholars have made a sport of attacking them. Each new approach to the study of religion seems

to begin with a critique of the sacred and profane—often crediting this division as a fundamental misunderstanding of religion that has per- vaded all previous scholarship. Like pretty much everyone else, I reject Durkheim’s strict division of the sacred and the profane. I do not, how- ever, believe that this is a particularly novel position or that Durkheim’s distinction represents some fundamental misunderstanding of religion by Western scholars. It seems clear that, with the exception of Durkheim and perhaps a few others (e.g., Eliade 1961), almost all scholars realize that religion is difficult to untangle from other aspects of human life and that religion permeates many, if not all, human endeavors. This is not to say that the definition of religion is unimportant or trivial. It isn’t. The point is that scholars of religion have explicitly struggled with this ques- tion for at least a century.

Despite Durkheim’s mistaken arguments concerning the sacred and the profane, many of his other arguments concerning religion are valu- able. For the purposes of my work, Durkheim’s work on social solidarity is particularly important. In descriptions startlingly similar to Marx, Durkheim argued that a division of labor underwrote modern European societies. That is, different members of society were engaged in different activities (e.g., tinker, tailor, soldier, sailor, etc.), each necessary for the perpetuation of society as a whole. Different functional parts of society needed to be integrated with other parts in the same way that different organs of the body perform specific functions that keep the body as a whole functioning.

This is what constitutes the moral value to the division of labor. Through it the individual is once more made aware of his dependent state vis-à-vis society. It is from society that proceed those forces that hold him in check and keep him within bounds. In short, since the division of labor becomes the predominant source of social solidarity, at the same time it becomes the foundation of the moral order. (Durkheim [1893] 1984:333)

For Durkheim, the chief way that people were “made aware of [their] dependent state vis-à-vis society” was religion. In the late nineteenth cen- tury, Durkheim suggested that a rise in the rate of suicides in France was the result of a diminishment of attendance in religious ceremonies ([1897] 1951). He argued that participation in religious ritual promoted a sense of belonging to a community. The loss of social solidarity, Durkheim argued, resulted in feelings of isolation, leading toward an increase in suicides. Though recognizing the legitimizing role of religion in perpetuating social inequality, in some sense Durkheim focused on the sunny side of false

consciousness. This does not mean that Durkheim believed that people should continue attending religious ceremonies that perpetuated social inequalities—like Marx, Durkheim was a social activist who sought to refashion society in more equitable ways. Durkheim argued that in order for people to avoid a feeling of isolation that leads to despondency and suicide, people must engage in collective rites that reaffirm their codepen- dence with other members of society.

[R]eligion is something eminently social. Religious representations are collec- tive representations which express collective realities; the rites are a manner of acting which are designed to excite, maintain, or recreate certain mental states in these groups. (Durkheim [1915] 1995:9)

That is why all parties political, economic or confessional, are careful to have periodical reunions where members may revivify their common faith by manifesting it in common. (Durkheim [1915] 1995:212)

For Durkheim, collective rites—whether religious or secular—served a psychological need, promoting group identity and affiliation.

As formulated by Durkheim, social solidarity was a relatively unprob- lematic and uncontested social process. Durkheim convincingly demon- strated the importance of collective rites, but we need to consider exactly who gets to participate in these rites and the manner in which these rites are performed. Different rites can promote and perpetuate differ- ent conceptions of who is, and who is not, in the group and what sorts of inequalities are perpetuated. In some cases, social solidarity may be based on common affiliation with a charismatic leader. In others cases, solidar- ity may be more egalitarian or communal. Following Weber, attempts to establish particular forms of social solidarity are legitimations that confer legitimacy only when successful.

As described by Durkheim, the performance of collective rites main- tains and fosters social solidarity. Rites are actions that people do in spe- cific places—actions that may leave material traces of their practice. As in the distinction between legitimations and the state of legitimacy, the material record directly reveals only attempts at solidarity. The degree to which these material attempts were successful requires additional evidence and interpretation. While it is possible for people to engage in collective rites in unmodified natural spaces, people often construct elaborate facilities for the performance of collective rites. An analysis of the design and layout of these facilities can reveal both attempts to promote solidarity and the particular form of solidarity that was being attempted.

The identification of attempts at solidarity comes primarily through an analysis of the ways that spaces are designed to foster or impede relationships between people engaged in collective rites (Fogelin 2003, 2006). Many Quakers, for example, believe that all are equal before the Lord and that God speaks directly to each person. Quakers who hold this belief reject ritual leaders, preferring instead to sit silently until any member of the denomination feels moved to speak. For this reason, Quaker meetinghouses often consist of little more than a room filled with concentric rings of chairs or benches. This layout allows all mem- bers to see each other while limiting the possibility of any one person leading worship. As such, the design of Quaker meeting halls fosters a particular form of solidarity consistent with an egalitarian relationship among participants.

In contrast to Quaker meetinghouses, the layout of a Catholic church promotes a different form of solidarity, a hierarchically organized com- munity based on the common identification with officiating priests who have special skills (e.g., transubstantiation) that are unavailable to lay Catholics. Rather than concentric rings, pews in Catholic Churches are typically arranged in rows, all facing an elevated platform on which a priest leads worship. Parishioners can clearly see the priest, but cannot see people behind them or the faces of those in front. While still housing collective rites that promote solidarity, Catholic churches promote a more hierarchical form of solidarity than Quaker meetinghouses.

Catholic churches and Quaker meetinghouses are, in part, attempts to promote particular forms of solidarity by those who designed and built them. In both cases, it seems that these attempts succeeded. Both Catholic churches and Quaker meetinghouses are long-lasting and ubiqui- tous architectural forms. It would seem unlikely that Catholics or Quakers would continue to construct these buildings if these attempts at solidarity had failed. It should never be forgotten, however, that there is no guaran- tee that attempts at solidarity will succeed. Throughout history, many, if not most, attempts at solidarity by aspiring religious orders have failed.

Rites of Passage and Liminality

Drawing from Durkheim’s concept of social solidarity, Van Gennep (1960) and Turner (1967) examined the common elements of ritual that mark the movement of individuals from one social status to another (i.e., child/adult, unmarried/married, alive/dead). That is, if society consists of people existing in relation to one another, shifts in the social status

of one of its members become both important and problematic. Societies require mechanisms that can move individuals from one location within the social structure to another, without creating a rupture in the fabric of social life. Van Gennep identified three cross-cultural phases of ritual that affect the change from one social status to another. These three periods (separation, margin, and aggregation) make up what Van Gennep labeled rites of passage. The first phase of rites of passage removes an individual, both physically and metaphorically, from his or her existing social role. In the second phase, the liminal period, individuals have their social roles transformed. Finally, the third phase reintegrates the individual back into society, with his or her new social role. Important to my research on Indian Buddhism is the recognition by Turner and Van Gennep that spa- tial movement and material symbols often mark movement through the stages of a rite of passage.

In Turner’s (1967:ch. 7) classic discussion of boys’ initiation ceremo- nies among the Ndembu of central Africa, movement through the phases of the rite of passage were accompanied by the physical movement to a specially constructed camp in the bush. This camp, the physical location of the liminal phase, was marked by the construction of a fence blocking from view all activities that occurred within. Passage through a gate into the enclosure symbolically marked the passage into the liminal phase, complete with a symbolic rebirth as the men leading the ceremony forced the boys to pass through their legs.

As defined by Turner (1967:93), the liminal period is an “interstruc- tural situation” that lies at the heart of rites of passage. Individuals within the liminal period are neither here nor there, or as Turner described it, “betwixt and between.” Lying between two culturally sanctioned catego- ries, people in a liminal period are unclassifiable, and thus not part of their society. With this foundation, Turner investigated the perceptions of liminality. Relying heavily upon Douglas’s (1966) Purity and Danger, Turner argued that disorder and contradiction are considered unclean, polluting, and dangerous. Since people in the liminal period are outside any socially sanctioned category, they are themselves unclean and danger- ous. This explains, in part, the typical pattern of separation and isolation of people going through rites of passage. Segregation limits the expo- sure of initiates to society and vice versa. Reintegration marks initiates’ return to a classifiable social category. But, passage through these peri- ods of disorder and contradiction are necessary, and the transformative power of the liminal period is undeniable. Pollution, then, is not some- thing to simply be avoided, but rather a necessary and dangerous source of transformative power.

Since the pioneering work of Van Gennep and Turner, anthropologists have come to recognize that liminality has much broader application than simply as the middle stage of a rite of passage. Certain people and places exist in a perpetual state of liminality—that is, certain people and places are perpetually betwixt and between. Religious figures, for example, can exist in the indefinable middle ground between the heavens and earth. Proximity to the liminal provides shamans, priests, and other religious figures with special skills and transformative power (e.g., soul flight or transubstantiation), but often results in a degree of isolation from normal social processes (e.g., living separately or sexual abstinence). Similarly, certain places can also be perpetually liminal due to their association with the divine. Those locations that blur the boundaries between the heavens and earth are powerful (e.g., churches, graveyards, stupas), but potentially dangerous. Swearing in church is a greater transgression than swearing in the course of daily life. The power and pollution of liminality must be contained in certain people and locations, or else the bulk of the population would be forced to live in a state of perpetual anxiety. Just as the Ndembu placed initiation camps in distant locations behind fences, religious figures and sacred locations in many societies are often isolated. Liminality, then, serves to partially explain the source of Durkheim’s ([1915] 1995:44) mistaken belief that “[a] religion is a unified system of beliefs and practices relative to sacred things, that is to say, things set apart and forbidden.”

Durkheim was wrong to suggest that the sacred is always set apart

and forbidden, but as studies of liminality show, religious people, prac- tices, and locations are often set apart and forbidden. For an archaeolo- gist this is important because the methods people employ to separate and isolate the liminal can potentially leave material traces (e.g., walls, gates, geographic isolation). Spatial analyses can reveal liminal locations, illuminating specific conceptions of the sacred in the process. As will be discussed in later chapters, Buddhists often isolated monasteries and pil- grimage locations behind large walls or in remote locations, separating the sangha and the practice of Buddhist ritual from mundane, worldly concerns. This is not to say that Buddhist monasteries and pilgrimage centers were completely set apart and forbidden—they weren’t. Indian Buddhists were constantly forced to strike a balance between their desire for isolation and asceticism with their desire to establish and provision a powerful, coherent community of Buddhists. As discussed in later chapters, much of the history of Indian Buddhism can be understood by examining how subsequent generations sought to balance and exploit these competing desires.


In the last fifty years, anthropologists and fellow travelers in related disciplines have reformulated Marx’s dialectic and have expanded on the pioneering work of Durkheim and Weber. Where Marx ([1859], in Elster 1986:187) examined the dialectic between “the real [economic] foundation” of society and the “legal and political superstructure,” the modern dialectic juxtaposes the real actions of agents—economic and otherwise—in relation to more disembodied structural principles. That is, the modern dialectic contrasts the actions of agents altering and pre- serving the social order with the pre-existing social order that constrains agents’ actions. Like older theorists, modern theorists vary by which part of the dialectic they emphasize, with some seeing structure as more determinative and others emphasizing the creative power of agents. In the modern study of religion, these differences are expressed in varying understandings of the relationship between religion and ritual.

Structural Approaches to Religion and Ritual

Clifford Geertz (1973:90) provides a succinct definition of religion from a structural perspective.

Religion is a system of symbols which acts to establish powerful, pervasive, and long lasting moods and motivations in men by formulating conceptions of a general order of existence and clothing these conceptions with such an aura of factuality that the moods and motivations seem uniquely realistic.

The emphasis here lies on belief and the meaning of symbols—the manner in which belief serves to instill in people a sense of where they belong in society and the universe. In this sense, Geertz’s definition of religion is strongly informed by Durkheim’s conception of social solidar- ity. Rituals, in this conception, serve to enact or promote symbolic mean- ings in a format that the masses can easily understand. As phrased by Wallace (1966:102), “ritual is religion in action; it is the cutting edge of the tool It is ritual that accomplishes what religion sets out to do.” In

this formulation, ritual is a form of human action determined or shaped by underlying religious views.

Of particular importance to a structural perspective is the idea that religion is a particularly stable and long-lasting cultural phenomenon. If religion is a relatively stable phenomenon, and ritual is the enactment of

religious principles, then rituals must also be relatively stable over time. Thus, many scholars view rituals as a particularly anachronistic element of human societies (Bloch 1977, 1986; Connerton 1989). Just as many Christians continue to use the King James Bible, many societies continue to engage in rituals that employ archaic speech or actions. This is perhaps best illustrated in Bloch’s (1986) classic discussions of boys’ circumcision ritual among the Merina in Madagascar. Bloch demonstrates that the con- tent and form of the ritual has been unchanged for almost two centuries, despite its shift from a small-scale village puberty rite to a large-scale ritual central to Merina identity.

The anachronistic and invariant elements of ritual fit well within archaeological approaches that employ historical and ethnohistorical sources. If religion is among the most stable and long-lasting cultural phe- nomena, then ethnographic, ethnohistoric, and historic accounts—even those that postdate the archaeological period being studied by several centuries—are a legitimate source for the study of ancient religious practices. Furthermore, given the richness of the symbolic, mythic, and doctrinal information available in these sources, it is not surprising that archaeologists who have a more structural understanding of religion tend to focus on these sources. Given the difficulty of determining the meaning of symbols in purely prehistoric contexts (Fogelin 2007a; Hayes 1993), assuming the stability of religion over the long term is a convenient research strategy. Recent trends in the anthropology of religion, however, have questioned these more simple formulations of historical memory.

Ritual Practice

Some modern anthropologists and religious historians advocate for the primacy of ritual practice in the study of religion. These scholars empha- size the creative or revolutionary aspects of ritual. Rituals do not enact stable sets of religious beliefs, but rather rituals construct, create, mod- ify, or preserve religious beliefs (Bell 1992, 1997; Humphrey and Laidlaw 1994). As discussed in Chapter 1, people constantly choose to remem- ber, forget, or recreate elements of their religion through ritual practices (Connerton 1989; Hobsbawm and Ranger 1983). In archaeology this per- spective has been employed in the works of Bradley (1991, 1998, 2000), Rowlands (1993), and others (Chesson 2001; Mills and Walker 2008; Pauketat 2001; Van Dyke and Alcock 2003). Although specific rituals may remain the same over long periods of time, their meaning for society is constantly recontextualized. People transform and change underlying

religious beliefs through the creation and practice of rituals. Rather than focus on the stable meanings of ritual actions, practice theorists empha- size the experiential aspects of ritual and the effects of ritual on the social relations between ritual participants. Practice approaches tend to focus on ritual change and what ritual does, rather than what it means, although it is important not to overplay this last point.

From a practice perspective, Bell (1997:ch. 5) identifies six characteris- tics that rituals and ritual-like activities exhibit to varying degrees. Bell is clear that these characteristics are not exhaustive, nor are the characteris- tics limited to religious ritual. The characteristics are as follows.

Formalism: Rituals often employ more formal, or restricted, codes of speech and action than people use in everyday life.

Traditionalism: Rituals often employ archaic or anachronistic elements.

Invariance: Rituals often follow strict, often repetitive, patterns.

Rule-governance: Rituals are often governed by a strict code of rules that determine appropriate behavior.

Sacral symbolism: Rituals often make reference to, or employ, sacred symbolism.

Performance: Rituals often involve the public display of ritual actions.

Ritual, from a practice perspective, is more a process than an event (Bell 1992; Humphrey and Laidlaw 1994). Certain actions are, or become, ritu- alized; they become more formal, traditional, invariant, and so on. That is, ordinary actions assume greater meaning and significance as they become ritualized and alter religious belief.

Does Religion Exist?

While most modern theorists of religion accept that a dialectic exists between ritual and religion, recently Talal Asad has questioned the util- ity, if not the existence, of the very concept of religion. Like almost all scholars who preceded him, Asad begins his analysis by noting the impos- sibility of defining religion in terms of Durkheim’s dichotomy of the sacred and the profane (2003:30–37). Asad, however, adds an important twist to the standard argument. Drawing heavily from Said’s Orientalism (1978), Asad (2003:31) argues that the concept of religion is derived from nineteenth-century European understandings of secularism.

It was late nineteenth-century anthropological and theological thought that rendered a variety of overlapping social usages rooted in changing and

heterogeneous forms of life into a single immutable essence, and claimed it to be the object of universal experience called “religious.”

For Asad, the conceptual category of religion emerged in the nineteenth century as the foil to the concept of secularism. By this view, religion was artificially marginalized as the province of irrational personal belief, existing in contrast to the rationality and secularism of modern Western states. By creating a category of human experience called religion, Asad argues, anthropologists made two fundamental mistakes. First, they removed power and contestation from the analysis of religion; religion became a disembodied structural phenomena. Second, anthropologists universalized this problematic category onto all societies in the world. In so doing, they ascribed a problematic category for understanding Western religion into contexts in which the concept of religion, or often even the word “religion,” was completely lacking.

The value of Asad’s critique is that it provides an underlying mechanism to explain the peculiar intellectual history of the study of non-Western religions. Before studying non-Western religions, colonial scholars first attempted to identify the religious elements of non-Western societies. In the case of Buddhism, this consisted of identifying the primary, or canon- ical, texts from the bewildering quantity and diversity of early Buddhist texts. Drawing on the Western notion that religion was primarily of the mind and esoteric, early colonial scholars of Buddhism identified a limited number of the most scholastic texts in Buddhism—those texts with the most philosophical or esoteric elements—as the canonical texts of ancient Buddhism (Lopez 2001). It is no surprise then, that the subsequent his- tories derived from these texts were highly scholastic. Often ignored or marginalized in this early research were the full range of scholastic texts (sutras), the subsequent commentaries on these texts (abhidharma), and mundane texts concerning monastic law (vinayas). Colonial scholarship on Buddhism created a canon of the Buddhist religion that excluded the true diversity of political and doctrinal debates contained within the full range of Buddhist literature.

Recently, Severin Fowles (2013) has used Asad’s insights to reinterpret the Pueblo religion in the southwestern United States. Fowles argues that the concept of religion has harmed archaeological, historical, and ethno- graphic studies of Pueblo life. Rather than religion, Fowles (2013:240 ital- ics in original) uses “doings,” a term that Pueblo people use to describe what both they and others call religion.

Pueblo doings are altogether different. At Taos, they simply are the wider social framework within which Christian and non-Christian religions are tolerated.

Pueblo doings, in other words, are the very warp and weft of Pueblo society, not one patch among many within some other and more encompassing insti- tutional patchwork. Doings are less a matter of individual choice than of obli- gation and duty, a necessary consequence of having been born in a particular landscape and a particular community.

By relying on doings rather than more universal considerations of reli- gion, Fowles is able to untangle more localized actions and to see Pueblo doings as using place as a substitute (or proxy for) time. As such, Fowles’s interpretation is highly—and intentionally—particularistic.

Archaeology of Religion and Ritual

In contrast to Fowles’s rejection of the concept of religion, some archaeolo- gists following a more structural perspective do the opposite; they expand the concept of religion to include almost everything (Brück 1999; Insoll 2004). By rejecting the Durkheimian (1915 [1995]) distinction between the sacred and the profane, they begin to equate religion and culture. This perspective is clearly articulated in Insoll’s recent discussions of religion (Insoll 2004:22):

The more we look, the more we can see religion as a critical element in many areas of life above and beyond those usually considered—technology, diet, refuse patterning, housing. All can be influenced by religion; they are today, why not in the past? Religion can be of primary importance in structuring life into which secular concerns are fitted, the reverse of the often-posited framework.

There is clearly an element of truth to this argument. Simply because some action is economically rational does not necessarily mean that it is not also religiously motivated. However, from this perspective, religion seems to be everywhere and all pervasive in non-Western societies. The problem with this perspective is the growing recognition (noted by Insoll 2004:17) that some societies, even “traditional” societies, have only a limited inter- est in things religious (Asad 2003; Barth 1961; Fowles 2013; Kemp 1995). The separation of church and state may be a modern, Western notion, but it is a mistake to assume that people in other societies were, or are, neces- sarily ruled by their religious beliefs any more, or less, then Europeans or Americans. Following Asad, the centrality of religion in human society is highly variable and assumptions of its universal importance highly suspect.

Other archaeologists who are interested in domestic or other small-scale rituals celebrate the idea that religious and secular rituals are not distinct or clearly identifiable (Bradley 2005; Walker 1999). Rather than see- ing this as a problem, these archaeologists are interested in the process whereby a seemingly ordinary action becomes ritualized (Humphrey and Laidlaw 1994). Rather than seeing rituals as either religious or secular, archaeologists of this sort view the gray area between religion, ritual, and everyday life as worthy of study.

Once we reject the idea that the only function of ritual is to communicate reli- gious beliefs, it becomes unnecessary to separate this kind of activity from the patterns of daily life. In fact, ritual extends from the local, informal and ephemeral to the public and highly organized, and their social contexts vary accordingly. (Bradley 2005:33)

In a detailed study of ritual in prehistoric Europe, Bradley (2005) exam- ines the role of a type of earthen enclosure (Vierekschanze) in Neolithic Central Europe. These structures blend ritual, domestic, and even indus- trial characteristics in ways that have frustrated attempts at understand- ing them. One of these structures, Mšecké Zehrovice in Bohemia, blends ritual structures and metalworking (Bradley 2005:21–23; Venclová 1998). Relying on ethnographic accounts of metalworking, Bradley argues that the transformative practices of metalworking typically rely on magic, rit- ual, and restricted knowledge (see Dobres 2000; Schmidt and Mapunda 1997). Traditionally, metalworking was a ritual act, even if it resulted in the production of utilitarian objects. Thus, the question “is it religious?” is viewed as fundamentally flawed. Metalworking is both sacred and pro- fane. Sacredness does not adhere to any object or phenomena in particular, but is created through an object’s use or performance in specific contexts (Appadurai 1986; Walker 1998, 1999, 2002), often tied metaphorically to an object’s mundane or domestic role (Bradley 2005; see also Ortman 2000; Tilley 1999; Plunket 2002).

While seemingly contradictory, I see value in all of these different mod- ern perspectives on what religion is or is not. While Asad may be right that local spiritual practices should not be subsumed within a larger category of religion, it is also a mistake to say that every local practice is entirely unique. Many Pueblo doings, for example, look a whole lot like Christian, Buddhist, or other religious doings. Pueblo initiations like the vision quest do seem to exhibit the classic tripartite division of rites of passage. It would be a mistake to completely reject anthropological advances con- cerning religion, even if these advances are tied to the peculiar history of

European secularism. On the other hand, Western academic understand- ing of religion should never be applied uncritically. The application of any anthropological theory to a new context requires a demonstration that the theory is applicable to that context, and if so, what modifications are required to the theory to make it appropriate for that context. We should always be wary of anthropologists who employ any single theory “to explain just about everything.” It is not just that this tends to be boring; it also tends to be wrong.


Archaeologists are usually more comfortable studying ritual than they are studying religion. The reason for this is straightforward. Drawing from modern anthropological understandings of the dialectic, most archaeologists view ritual as a form of human action that leaves mate- rial traces, whereas they view religion as a more abstract symbolic system consisting of beliefs, myths, and doctrines. Since symbols are intellectual constructs rather than material realities, archaeologists feel that claims about them are always less certain. Many archaeologists, myself included, have criticized those archaeologists who focus on sym- bolic and structural analyses of religion for this reason (Fogelin 2007a; Howey and O’Shea 2006; Insoll 2004; Renfrew 1994; Walker 1998). I now believe that I was mistaken in many of my earlier criticisms. Rituals do, at times, enact structural religious beliefs through complex symbolic actions. If religion and ritual truly inform one another, archae- ologists can only benefit from investigations into both sides of the dia- lectic. Where written sources describe religious beliefs, mythologies, and symbolism, a structural approach to ritual is fairly straightforward. However, there is no reason, a priori, to assume that archaeologists can- not study ancient symbolism and belief in purely material contexts. The question remains, however, how best to study symbolism within the material confines of archaeological research.

If symbols are organized in complex systems, as Geertz (1973:90) argues, then knowledge of some aspects of the symbolic system could be used to infer other parts. Archaeologists might even identify a key symbol (Ortner 1973), one that serves as a central element of ancient ritual prac- tices. Furthermore, studies of ancient symbolism need not focus solely on the symbols themselves but can infer the meaning of symbols from their material context. If iconography depicting a ruler is consistently associ- ated with an image of an eagle, for example, it is likely that the eagle is

itself a symbol of royalty. With that insight, it might be possible to exam- ine other contexts in which eagle iconography exists to further elucidate the symbolic meaning of eagles and royalty. In the same way, it is possible to examine the material context of Buddhist symbolism (e.g., stupas and Buddha images) to gain insight into their meaning and function that goes well beyond what is reported in Buddhist texts.

There are many excellent approaches appropriate to the anthropologi- cal and archaeological study of religious symbolism (e.g., Lévi-Strauss 1963, 1983; Eliade 1961). With the exception of the insights of Victor Turner (1967), Mary Douglas (1966), and Clifford Geertz (1973), I do not employ these symbolic or structural approaches in my analysis of Indian Buddhism to any significant degree. I have no doubt that the application of these approaches could work, but in my symbolic analyses of Buddhism I primarily rely upon the semiotic perspectives of Charles Sanders Peirce (1931–1958, 1992, 1998). While Peirce was a contemporary of Durkheim and Weber, sociocultural anthropologists only rediscovered his work in the last thirty years (Mertz and Parmentier 1985; Silverstein 1976; Singer 1978) and archaeologists in the last decade (Bauer 2002; Fogelin 2012; Preucel 2006; Preucel and Bauer 2001).

Charles Sanders Peirce and Semiotics

The scope of Peirce’s philosophy is vast, with critical contributions to epistemology (see Fogelin 2007c), formal logic, and semiotics (see Preucel 2006). In his work on semiotics, Peirce developed a complex theory con- cerning the relationship between people, signs, and the objects to which signs refer (e.g., the relationship between the word “stupa,” the sensual perception of the structure to which the word “stupa” refers, and the per- son thinking about a “stupa”). Peirce defined signs as “something that stands to somebody for something in some respect or capacity” (Peirce 1931–1958:vol. 2:228). Among the key elements of this definition is the emphasis placed on the role of people thinking (interpretants) about sign/object relations. For many anthropologists and linguists, the inclu- sion of interpretants in Peirce’s formulation of semiotics is critical to its value for anthropological interpretation (Bauer 2002; Daniel 1984; Keane 1997; Parmentier 1997). Whereas the structural anthropology (e.g., Lévi-Strauss 1963, 1983) inspired by Saussure’s (1983) semiotics has been criticized for downplaying social agency (Bourdieu 1977; Giddens 1984), Peirce’s semiotics allows for human agency, even if it does not explain it particularly well.

In Peirce’s semiotics, interpretants play an active role in interpreting signs vis-à-vis objects. This does not mean that objects and interpretants are synonymous with physical objects and people. Peirce’s categories are abstract and of the mind. An object is not physical, but rather the sen- sual perceptions of the physical mediated by how a person thinks about it. Similarly, interpretants are not people. Rather, the ways people think in relation to signs and objects are interpretants. It is the relation between signs, objects, and interpretants that Peirce emphasizes, rather than any inherent qualities of signs, objects, or interpretants.

Peirce’s tripartite division of signs, objects, and interpretants was only one of many tripartite typologies he created. Among the most important of these is his “doctrine of categories.” This doctrine states that there are three different overarching categories of human experience: firstness, secondness, and thirdness. Peirce defined firstness as the conception of irreducible being or existence—something is. Peirce describes “redness” as an example of firstness (Peirce 1931–1958:vol. 1:25). The experience of redness consists of “unanalyzed, instantaneous, raw feeling” (Preucel 2006:52). Secondness refers to our experience of otherness, our perception of signs and their interdependence with other signs. For example, recogni- tion of a father requires the existence of a child. Finally, thirdness involves intellectual thought or argument—for Peirce, prediction—about the world around ourselves. Broadly speaking, these three categories prog- ress from the immediate and necessary to the intellectual and abstract. Socially speaking, there is power and importance in each category. Raw emotion has its own value, distinct and different from the value of intel- lectual insight.

The concepts of firstness, secondness, and thirdness matter for archaeology because Peirce directly relates them to a typology of signs consisting of icons, indexes, and symbols (see Table 2.1). The underlying principle emphasized in this latter typology is the manner in which an interpretant links signs to the objects to which signs refer. Icons, which



Doctrine of Categories

Typology of Signs

Social Impact of Signs









Peirce largely associates with firstness, promote an immediate feeling or emotion. Icons achieve this by sharing an innate connection to an object. A portrait, for example, is an icon of the person being depicted. Icons are often identified in terms of visual resemblance between a sign and its object, but the resemblance can also be exhibited in other senses (smell, taste, etc.) or qualities of the icon. Indexes, in contrast, emphasize secondness or otherness by their very construction. Indexes indicate the status of a sign by a necessary relation to another sign. Classic examples of secondness include a weathervane indexing wind, or a bullet hole indexing the passage of a bullet. A bullet hole is not a bullet, but the existence of a bullet hole necessarily implicates the passage of a bullet sometime in the past. Symbols, in Peirce’s terminology, denote an object through convention. A stop sign (a red octagon and the word “stop”) is a symbol, understood only because people have been taught its signif- icance. Symbols are most closely related to the category of thirdness. In terms of Peirce’s doctrine of categories, icons, indexes, and symbols exhibit greater and lesser degrees of firstness, secondness, and third- ness, greater and lesser degrees of emotional immediacy and intellectual abstraction.

It is important to note that Peirce did not see any of his categories or typologies as clearly distinct. Rather, he viewed the world from the per- spective of synechism—that the world is characterized by continuous variation. Within a semiotic perspective, signs are never simply icons, indexes, or symbols. Signs have elements of each, and the potential for multiple meanings. The same sign can be an icon, index, or symbol depending on the interpretant, but that is only part of the multivalent nature of signs. Even for a single individual, at a single moment in time, signs have multiple valences. The American flag, for instance, is a symbol of the United States, but could also be an index leading soldiers to safe ground during wartime. Peirce is explicit that signs are constantly in flux. Semioticians see the malleability and multivalency of signs as central to their investigations of language, art, and human cognition.

Signs have real social impacts on the people who use them. Semioticians can employ Peirce’s theories to identify the social significances of differ- ent signs in different social contexts. The most powerful or socially mean- ingful signs would be those that forcefully project all elements of all of these categories simultaneously. That said, different cultures might place different value on icons, indexes, and symbols. Daniel (1984), for example, argues that in Tamil society in India icons are considered to be the most fundamental or powerful type of sign. This stands in contrast, Daniel argues, to the Euro-American emphasis on indexes and symbols. When

examining the social implications of Peirce’s typology of signs, we must always be aware that different cultures may assign value differently.

The most developed aspect of Peirce’s semiotics is his sophisticated understanding of signs—his tripartite typology of signs and their rela- tionship to firstness, secondness, and thirdness. He pays less attention to objects and interpretants. Peirce’s semiotics contains no theory to explain the independent motivations of interpretants or their actions, except in reaction to signs and objects. Peirce’s semiotics also ignores much that can be said of objects. By defining objects as of the mind, Peirce under-theorized the ramifications of the materiality of objects. In many cases, signs are material objects. The medium of their construction affects their meaning (Eco 1976:267). Different media have distinct physical properties and are governed by physical laws that cannot be changed. Smoke may be a potent sign for an interpretant, but it cannot be made into a statue in the same way stone can. Different media have different potentialities for expres- sion. Artists can do things with oil paints (e.g., color) that they cannot do with charcoal, and vice versa. Further, interpretants commonly ascribe different meanings to different media (e.g., charcoal is more proletarian, while oils are more bourgeois), but portraits in both media could be icons of the same individual.

While not all signs are material, for archaeologists material signs are the focus of research. As such, semiotic approaches must be tempered with theories of materiality. While semiotics is more structural and materiality more agent oriented, nothing in either theory necessitates the complete rejection of the other. Semiotics and materiality are not contradictory, but rather complementary. Both are necessary for a comprehensive under- standing of the long-term symbolic development of Indian Buddhism. Where semiotics can inform on the significance and meaning of Buddha images, for example, materiality provides a way to understand Buddha images as material things constructed in specific places, at specific times for specific purposes.


Over the last two decades, some anthropologists and archaeologists have increasingly extended the concept of agency from people to mate- rial objects. This new perspective is called materiality. By this view, mate- rial culture is not a passive reflection of cultural practices, but rather a dynamic part of the construction of ideologies (Appadurai 1986; Boivin 2008; Gell 1998; Meskell 2005; Miller 2005; this theoretical tack is

presaged by Schiffer 1976, 1995). For example, the creation of pottery is not strictly utilitarian, and the variation in the form of pottery among different groups is not only due to social tradition (Miller 1985, 1987). Rather, the decision to make pottery in specific ways is a part of the cul- tural process itself. Each time a person chooses to make a piece of material culture, she decides if she will follow traditional forms or make some- thing wholly new. In this light, the decision by people to make something involves human agency, rather than simply the enactment of structural rules. Each time people choose to build or use something, they are forced to decide whether they will engage in dominant discourses or reject them. This dynamic perspective on material culture is commonly referred to as materiality. With the development of this perspective has come recogni- tion that the creation of material culture is part of the dynamic creation of self and identity, if not culture itself.

Building on theories of materiality, DeMarrais, Castillo, and Earle (1996; DeMarrais 2004) propose that material culture can be investi- gated as the “materialization of ideology.” Here, ideology is understood in a Marxist sense—that ideology serves to promote relations of power between the elite and commoners. DeMarrais, Castillo, and Earle argue that the control of production and consumption of prestige goods can be understood as part of the production of authority. They suggest, for example, that in Bronze Age Europe, elite control over the production and possession of weapons materialized an ideology that supported a devel- oping class of warrior chiefs. More recently, Johansen and Bauer (2011) have taken less of a top-down approach to the materiality of power, see- ing power as derived from a more dialectical relationship between people and things.

In the last decade, materiality researchers have increasingly questioned whether a clear division exists between people and the objects people cre- ate or, as materiality researchers phrase it, between subjects and objects (Boivin 2008; Gell 1998; Latour 1993, 1999, 2005; Miller 2005). New theories of materiality note that many objects are anthropomorphized and take on attributes of people. Following Bourdieu (1977), materiality researchers argue that the material world shapes human behavior and the acquisition of cultural rules. The type of house in which a person is raised will condition the way that a person sees the world as an adult. In this sense, objects can be said to have agency. The question for material- ity researchers is just how much agency to ascribe to objects. For Latour (1999, 2005), objects have agency equivalent to human agents in almost every way. Gell (1998), in contrast, sees objects as having agency only inasmuch as people ascribe or imbue objects with agency.

In an insightful evaluation of materiality studies, Boivin (2008) argued that even within materiality studies, the materiality of objects has often been downplayed.

[Studies of material culture] often continue to overlook the actual material- ity of the material world. What we frequently find instead is a far from novel emphasis on ideas, on human thought, and on representation. What we often find is a model, either implicit or explicit, of material culture as a text or as a language, as something that represents something else, and that is there to be interpreted. Thus, despite frequent claims to the effect that such studies are breaking down inherited Western dichotomies between ideal and material, between subject and object, and between culture and nature, many of them in fact simply re-impose traditional dualistic frameworks. The physicality of the material world continues to be ignored, as does the way that engagement with that materiality is at the crux of the human enterprise.

Boivin argues that the solution to this problem is to “undertake an exploration of the ways in which culture, society, and mind, the things we think of as most abstract and transcendent in our lives, are in fact far more material, visceral, and sensual than most of our academic models acknowledge” (Boivin 2008:23). Like Latour, Boivin argues that schol- ars must thoroughly collapse the dichotomies between mind and matter, form and substance, eliminating the tendency to elevate the abstract and cultural over the concrete and material (Boivin 2008:23).

Though I agree wholeheartedly with Boivin’s claim that neither mind nor matter should be considered more fundamental or important, I ques- tion the value of collapsing the distinction between mind and matter, subjects and objects, or people and things. To say that mind and matter are different does not demand that one be elevated over the other. But collapsing the dichotomy limits the ability to recognize and interpret the relationships between ideas and things. Things are not ideas and ideas are not things, however hopelessly interrelated the two might be. From my view, Gell’s position is more useful. While some objects are clearly viewed as having agency by the people who create and use them, others are con- sidered lifeless. In practice, the perception of an object can be manipulated by a skilled craftsperson. A bronze sculpture of a ballerina, for instance, may appear to a viewer to be light on her feet or moving. The bronze statue is, of course, heavy and motionless. To focus only on the perception of the object, or to collapse the distinction between the object and subject, would be to deny the medium of the statue, the skill of the artist who modeled the statue, and even the intent of artist in the first place. My complaint

with this view is that it downplays or denies the materiality of material objects—like Peircean semiotics, it privileges the perception of an object over the material form of the object itself. Following Gell (1998:ch. 7), craftspeople imbue their agency into objects, creating a form of “second- ary agency” derived from the craftspeople who initially created an object with purpose or intent. That is, people perceive the intent of the crafts- person as expressed in the form of an object, and ascribe that intent, that agency, to the object itself. This is not say that people will always perceive the specific intent of the craftsperson. Rather, the perception of an object is always an act of interpretation.

Like Gell, I hold that there is a real value in maintaining a duality between material objects and the perception of material objects, if only because the disjuncture between the two can be informative about people’s interaction with the material world (Fogelin 2012). Following the perspec- tive of Peirce, the meaning of a semiotic object is not straightforwardly imbued in the material object itself. In semiotics, objects are abstract and of the mind. Semiotic objects are linked to material objects through an interpretant’s sensual perception of material objects. It is possible, then, to alter a material object in subtle ways to alter the perception of that material object, and ultimately the semiotic object in the mind of the interpretant. Similarly, changes in the understanding of semiotic objects can inform physical manipulations of material objects. This is the point of articulation between theories of materiality and semiotics. By examining the relationship between material objects and semiotic objects—between objects of the world and objects of the mind—archaeologists can explore the meaning of objects without abandoning human agency.

Semiotics and materiality/practice theory each have strengths and weaknesses. Practice theory is particularly good at explaining the motiva- tions of agents (both human and non-human) in the past, and the impact of agents on structure. Semiotics, on the other hand, provides a frame- work for analyzing signs, and the relative impact of signs on the people who use and construct them. As stated by Marx ([1869] 1963:15), Men make their own history, but they do not make it just as they please, they do not make it under circumstances chosen by themselves, but under cir- cumstances directly encountered, given and transmitted from the past.”

Practice theory explains how “[people] make their own history,” while semiotics explains the meaning and social impact of signs “given and transmitted from the past.” Semiotics identifies structure in the world of signs that people find themselves in. Practice theory and materiality provide mechanisms to explain the goals of people materializing signs. Semiotics explains the consequences of these actions on the people who

inherit and interact with signs previously materialized. Taken together, semiotics, practice theory, and materiality explain the relationship between the material world and the world of signs better than any of these theories can alone. When combined with the older theories of reli- gion, archaeologists can draw on the insights of numerous scholars of religion. While the resulting analyses will contain more contradictions and will lack the coherence of an investigation relying on only one the- oretical approach, I argue that the interpretive messiness of a blended approach more accurately reflects the actual practice of religion. This blended approach, however, is not without difficulties—chief among them is coherence.


Academics often employ coherence to evaluate the quality of a proposed explanation (Fogelin 2007c). That is, any explanation should be inter- nally consistent, and specific elements of the argument should not lead to contradictory results. While I recognize the value of coherence—that an incoherent argument should raise concerns—I believe that, by itself, coherence is not necessarily the best way to evaluate a potential explana- tion. An excessive reliance on coherence can lead to two distinct prob- lems. The first problem is that an excessive reliance on coherence leads to artificially narrow investigations. There is no doubt that a study employ- ing only the concept of solidarity and only investigating those aspects of human life related to solidarity will be more coherent than a study that combines theories of solidarity with agency to investigate wider cultural practices. In a sense, narrow studies promote false coherence by ignoring confusing or contradictory elements.

The second problem with an over-reliance on coherency is that it is only valuable if anthropologists assume that the cultural phenomena are them- selves coherent. Yet, if there is one thing anthropologists have learned in the last fifty years, it is that cultures are incoherent, dysfunctional, and infused with power inequalities (Dirks, Eley, and Ortner 1993; Fogelin 2011; Foucault 1972 [1969]; Geertz 1957; Ortner 1984).

[A] central aspect of the concept of culture has been the claim of relative coher- ence and internal consistency—a “system of symbols,” a “structure of rela- tions.” But an intriguing line of discussion in contemporary critical theory has now posed a major alternative view: culture as multiple discourses, occasion- ally coming together in large systemic configuration, but more often coexisting

within dynamic fields of interaction and conflict. (Dirks, Eley, and Ortner 1994:3–4)

The older anthropological understandings of homeostasis, balance, and feedback have been replaced by discussions of contestation, negotiation, rupture, and collapse. Since culture is not coherent, since cultures are infused with intractable, irresolvable social contradictions and disjunc- tures, we should be suspicious of coherent explanations of human culture. Rather than theories that seek to resolve apparent social contradictions and disjunctures, we need theories that explain the persistence and social impact of contradictions and disjunctures.

Social contradictions occur when people encounter a situation in which two or more social rules demand opposite courses of action (Giddens 1979:141). A disjuncture, as I use it, is a lesser form of contradiction. Disjunctures create confusion about the proper course of action, though the different social rules may not entirely contradict each other. If con- sciously experienced, social contradictions and disjunctures force people to examine their implicit beliefs, laying bare the precarious foundations on which people base their actions. For this reason, social contradictions seem to demand some form of resolution; they seem to demand the rejection, replacement, or synthesis of the contradictory beliefs. As such, the experi- ence of social contradiction is intimately tied to questions of power, defined here as the transformative capacity of agents to alter or preserve the mate- rial or social order (Giddens 1984:14–16). Where different agents have different views on the best way to transform the social order—different views on how best to resolve contradiction—moments of social contradic- tion can become a source for persistent social disjunctures. This point can be best explained through a review of an old debate in anthropology.

In 1957, Clifford Geertz published his second article, “Ritual and Social Change: A Javanese Example.” In it, he challenged the function- alist notions of religion advocated by Bronislaw Malinowski and pro- posed a new interpretation that anticipated symbolic anthropology and later anthropological insights on power. In Magic, Science and Religion, Malinowski ([1925] 1948) had argued that Trobriand Islanders took every practical step to protect themselves from harm, but certain activities were inherently unpredictable and dangerous. For example, Trobriand Islanders were excellent sailors and were careful about the construction and care of their boats, but even the best sailors would occasionally die at sea despite their precautions. For this reason, Trobriand Islanders engaged in protective magic before fishing in open water. Trobriand Islanders used magic to cope with the remaining uncertainty once practical options were

exhausted. Extrapolating from his observations in the Trobriand Islands, Malinowski argued that the primary function of religion was to reduce people’s anxiety about death.

In “Ritual and Social Change,” Geertz (1957) used an extended case study of a failed funeral in a small town in Java to critique Malinowski’s functionalism. In most villages in Java prior to the twentieth century, religion was highly syncretic, combining local animistic practices with Hindu and Islamic elements. Funerals were expedient affairs, with a quick cleansing of the body and a few Islamic chants, followed quickly by burial. They were simple affairs that discouraged overt signs of emo- tional distress. As stated by Geertz (1957:40), “the whole momentum of the Javanese ritual system is supposed to carry one through grief without severe emotional disturbance.”

By the 1950s, however, many former village-dwellers lived in urban areas, and new political parties centered on religious affiliation were form- ing. Some political parties dominated by Muslims advocated for a more strict application of Islamic law in Indonesia, while other political par- ties considered Islam a foreign religion that needed to be expunged from Javanese ritual practice. Newly urban villagers who formerly practiced syncretic religions began dividing themselves along political/religious lines. In the context of these competing political/religious factions, one local official banned local Islamic religious figures (modin) from saying Islamic chants at the funerals of non-Muslim people. It is in this context in which a ten-year-old boy in a non-Muslim family died.

Upon hearing of the boy’s death, the men from nearby households began assembling at the boy’s house to prepare the body for burial. When the local modin arrived and realized the family was non-Muslim, he refused to lead the ceremony or perform the required Islamic chants. Hours of inaction followed, as those assembled for the funeral divided themselves into rival political/religious groups. Eventually the female relatives of the boy began openly weeping and prostrating themselves over the boy’s body, contrary to the accepted practices of a Javanese funeral. Finally, through a variety of negotiations, a hasty ad hoc funeral was performed. Rather than easing the family’s grief, the funeral “brought on an extended period of pronounced social strain and severe psychological tension” (Geertz 1957:35).

For Geertz, the breakdown of the funeral was the product of migration from rural, homogenous villages to more urban, heterogeneous settings. Where previously people living in close proximity within a village would have shared a common system of religious beliefs, immediate neighbors in urban areas often had opposing political/religious affiliations. The

continued desire to perform traditional village ceremonies was contra- dicted by the newly urban lifestyle of the participants in the ceremony. Without any way to reconcile contradictory beliefs, the participants in the funeral became paralyzed, unable to resolve their grief. Based on his experience with this failed funeral, and contrary to Malinowski, Geertz (1957:48) argued that

as a matter of fact, it is around religious beliefs and practices—[funerals], holi- days, curing, sorcery, cult groups and so on—that the most seriously disrup- tive events seem to cluster. Religion here is somehow the center and source for stress, not merely the reflection of stress elsewhere in society.

More broadly, Geertz argued that functionalism tended to “stress the harmonizing, integrating, and psychologically supportive aspects of reli- gious patterns rather than the disruptive, disintegrative, and psycho- logically disturbing aspects” (Geertz 1957:32). Geertz argued that the disruptive elements of politics and religion could inform an understand- ing of religious change that functionalism failed to address.

While both Geertz and Malinowski recognized that uncertainty can cause significant social stress, each had a different conception of the source. For Malinowski, the stress of uncertainty is derived from the fear of injury or death, the unavoidable and unpredictable danger that people periodically face in their lives. In a sense, religion and magic ameliorate the stress of uncertainty by making the unknown, known. While Geertz clearly demonstrates that this view of religion and magic is not always correct, it would be a mistake to say it never is. Religion can salve the fears of those approaching death, even while it creates and heightens anxieties of others.

Geertz’s conception of uncertainty is somewhat different. For Geertz, uncertainty occurs when people hold two or more beliefs that demand contradictory courses of action. Rather than too little information, the participants in the failed Indonesian funeral knew too much. Geertz does not discuss how people ameliorate the stresses of social contradictions. Rather, Geertz makes productive use of the ways these contradictions play out in cultural conflict and negotiation. This insight—that culture is shot through with contradictions and disjunctures—has become the basis of much of the best anthropological research of the last fifty years.

The new emphasis on social contradictions and disjuncture is, perhaps, most clearly shown in the work of Michel Foucault. Beginning with The Archaeology of Knowledge (1972), Foucault became progressively more interested in the role of discontinuity and rupture in European history.

Foucault’s subsequent histories of prisons (1977) and sexuality (1990a, 1990b, 1990c) focused on the transformative potential of, and contes- tations over, social contradictions and disjuncture. In The Archaeology of Knowledge, Foucault (Foucault 1972:10) argued that a focus on discontinu- ity was the fundamental distinction between traditional history and what Foucault called “new history.”

One of the most essential features of the new history is probably this displace- ment of the discontinuous: its transference from the obstacle to the work itself; its integration into the discourse of the historian, where it no longer plays the role of an external condition that must be reduced, but that of a working concept; and therefore the inversion of signs by which it is no longer the negative of the historical reading (its underside, its failure, the limit of its power), but the positive element that determines its object and validates its analysis.

Since Foucault, anthropologists and scholars working in related dis- ciplines have identified numerous instances where social contradictions concerning race, class, citizenship, and sexuality are implicated in persis- tent struggles over domination and resistance. Drawing from Geertz’s and Foucault’s insights on the critical importance of social contradictions, his- torical ruptures, and disjuncture, anthropologists have come to recognize that religion, like any other domain of cultural experience, is incoherent.

  1. here is a growing tendency to move culture out of the realm of the exotic custom, the festival, the ritual, etc., and into the center of the historical prob- lematic, or rather to recognize that the rituals and festivals are sites in which larger and more dynamic fields of discourse, larger and more powerful hege- monies, are being constituted, contested, and transformed. (Dirks, Eley, and Ortner 1994:6)

    But one question lingers in all of this recent research on religion. If reli- gion does not ameliorate the psychological stresses caused by uncertainty, what, if anything, does? This question leads to a theoretical path not taken. What was lost in Geertz’s critique is that, at times, people do need to ameliorate their fears and anxiety about uncertainty—the failed funeral Geertz described shows just how destructive and debilitating uncertainty can be. There is another way that people can address social contradictions and disjunctures, one that avoids transformative changes to the social order or the creation of social fissures. People can ignore them.

    Ignoring Social Contradictions and Disjunctures

    I propose that people often ignore social contradictions that cannot be solved, suppressed, or synthesized (Fogelin 2011). But ignoring the problem is not a simple process, particularly when the contradiction is constantly refreshed or represented to members of the community. Ignorance may be bliss, but it requires strategies and effort to achieve. Moreover, these strategies cannot require constant vigilance, as the act of vigilance would serve to remind people anew of the problem. There are, however, communal strategies for ignoring intractable social contradic- tions that permeate culture. These strategies do nothing to resolve contra- dictions; they only serve to hide them so that communities can go about their lives without the constant reminder of them. As such, ignoring the problem is a conservative act. Ignorance helps preserve the status quo by allowing people to ignore problematic contradictions inherent to the sta- tus quo. That said, the mechanisms of ignorance are not always beneficial to one group more than others. Where no ready solution can be found and no real advantage gained, ignorance can, in some instances, serve the interests of all involved. In other instances, the selection of what contra- dictions should be ignored and the methods used to ignore them can be contested. Like legitimations, attempts to ignore a contradiction or dis- juncture can fail.

    People can ignore social contradictions by avoiding or preventing situ- ations where contradictory beliefs are brought into sharp relief—people can reduce the perception of contradictions through the spatial and tem- poral separation of contradictory acts. Among the more straightforward ways to separate the experience of social contradiction is to literally sepa- rate the experience of the two halves of the contradiction temporally, spa- tially, or both. Temporalizing contradiction entails doing contradictory acts at different times. Spatializing contradiction entails doing contra- dictory acts in different places, though it also has an important temporal element since one person cannot be in two places at once. Either strategy can foster a degree of beneficial ignorance by limiting the simultaneous experience of two halves of a social contradiction. In this light, the prob- lem with the funeral described by Geertz was that contradictions in the political/religious life of newly urban villagers were encountered simulta- neously during the practice of the funeral. Had participants in the funeral chosen to engage in their contradictory political and religious activities at different times or in different places, the trauma of the funeral could have been ameliorated for all involved.

    To some extent, my perspective was anticipated by a brief discussion of contradictions by Giddens (1979:141–145). In a larger work that examined historical materialism, Giddens (1979:144–145) argued that

    the tendency of contradiction to involve conflict is weakened to the degree to which contradictions are kept separate from one another. Conversely, the more there is fusion or ‘overlap’ of contradictions, the greater the likelihood of conflict, and greater likelihood that the conflict will be intense.

    Phrased more crudely, Giddens suggested that we should expect more conflict over contradictions in those cases where agents are most cogni- zant of the contradictions. Giddens used the degree of separation as an analytical tool to predict the likelihood that conflict would arise over a given social contradiction. In contrast to Giddens, I see separation as a creative strategy that agents employ to reduce the conflict and trauma of social contradictions. Agents can employ their transformative capacity to separate the experience of contradictory beliefs, thereby preserving the existing social order while reducing the conflict and trauma that social contradictions can engender.

    Fifty years ago Geertz rightly critiqued Malinowski’s idea that reli- gion reduced anxieties about death and uncertainty. But something important was lost in Geertz’s critique. Religion may create as many social contradictions as it resolves, but the experience of contradic- tion can often be traumatic for those forced to experience it. While many social contradictions are contested, synthesized, negotiated, and exploited—sometimes it is better to simply ignore them. The methods used to paper over cracks in the social order are no less complex than those used to exploit or resolve them, and the investigations of inten- tional ignorance no less valid.

    Archaeologists and cultural anthropologists have long recognized that certain activities are performed in discreet places or at discreet times. Explanations for the separation of these activities include efficiency, exclusivity, secrecy, marginalization, liminality, pollution, or even simple convenience. To these I add ignorance—the amelioration of the anxieties generated by social contradictions through the temporal and spatial sepa- ration of contradictory acts and beliefs. While used here to examine the long-term development of Indian Buddhism, this analytical framework is potentially applicable to the study of contradictory beliefs in many soci- eties. Whatever the context, the study of ignorance employs the age-old maxim—out of sight, out of mind.


    Modern anthropologists have long debated the relative strengths and weaknesses of structure and agency. Structural approaches illuminate the similarities between cultures and the persistence of long-lasting cultural phenomena. Agent-based approaches, in contrast, lend themselves to more particularistic and contingent explanations of historical change. In prac- tice, few anthropologists strictly follow either a structural or agent-based approach. Rather, most ricochet between the two—often justifying their theoretical inconsistency through vague incantations of the dialectic.

    I no longer believe that my theoretical inconsistency has anything to do with the dialectic. Rather than justify my theoretical inconsistency through calls to dialectical reasoning, I now justify my theoretical dal- liances by noting the inconsistency, incoherence, and contradiction of human culture. Theoretical coherency is only a virtue in a world that is coherent. As will become clear in the following pages, Indian Buddhism was anything but coherent. From the start, Indian Buddhism was shot through with contradictions and disjunctures—chief among them the competing desires for asceticism and communalism. The history of Indian Buddhism, then, is the study of how Buddhists sought to resolve, exploit, and ignore this and other persistent contradictions. Over time, some disjunctures were resolved, though often at the expense of creating new disjunctures. Other disjunctures persisted, with subsequent generations addressing persistent disjunctures in novel ways.

    A full understanding of the long-term development of Indian Buddhism requires theories that can address the diverse, contradictory, and inco- herent elements of Buddhism. While I value the latest anthropological insights on practice, materiality, and semiotics, I persist in seeing value in the work of Marx, Weber, Durkheim, Turner, and others. Rather than rejecting older theories for the novelty of newer theories, I find it more informative—and interesting—to use them all.




    From The Buddha to Ashoka

    c. 600–200 bce


    ith the possible exception of the Buddha’s relics said to reside within stupas across Asia and some recent archaeological finds at Lumbini (Coningham et al. 2013), there are few Buddhist archaeological remains dating to the life of the Buddha (traditionally 563 bce to 483 bce; but see Bechert 1995), or the centuries immediately following his death.1 Critically, this lack of specifically Buddhist remains is not due to a general lack of archaeological remains from the sixth through fourth centuries bce. Archaeologists have excavated numerous sites from this period. Despite the large number of excavations, archaeologists have mostly failed to find remains that can either be definitively identified as Buddhist or, if identified as Buddhist, that unequivocally date to this early period. In some sense, this should not be surprising. During the life of the Buddha and the centuries immediately after his death, Buddhism was a small and developing religious order. It is unlikely that archaeologists

    1. Over the last two and a half millennia, Buddhists have repeatedly excavated, divided, moved, and reinterred the relics of the Buddha. With the possible excep- tion of those interred in a stupa at Vaisali, no relics have been found in stupas that date to the centuries immediately following the Buddha’s death. It would be pos- sible to perform various destructive dating techniques (e.g., carbon-14 dating) on the Buddha’s relics, but this would be highly inappropriate and, in the end, uninforma- tive. I am not interested in questioning whether specific relics of the Buddha are, or are not, authentic. Dating the Buddha’s relics would not alter or inform the beliefs and actions of those who venerated the Buddha’s relics in the past.

      ( 70 )

      would stumble upon the few material remains created by the Buddha or his followers, let alone stumble upon the small subset of those remains that can be unambiguously identified as Buddhist.

      Despite a common view of archaeology by non-archaeologists, archae- ologists are remarkably ineffective at identifying singular events of ori- gin. Given the fragmentary and limited nature of archaeological evidence, archaeologists can usually only identify the periods in which social move- ments or technological advances become common—the periods in which the material remains of these movements or technologies become ubiqui- tous. The lack of Buddhist archaeological remains dating to the life of the Buddha or the centuries immediately after his death is neither surpris- ing nor an indication that Buddhism was not practiced at this early date. None of this should be taken to mean that archaeology has nothing to offer the study of Buddhism in the sixth through fourth centuries bce. What archaeology can provide is an analysis of the larger social context in which Buddhism emerged.

      In some sense, an archaeological history of Buddhism and a textual study of Buddhism must begin at different times. Textual histories can rely on later Buddhist accounts to discuss the biography of the Buddha. An archaeological history, on the other hand, can begin only when archae- ological traces of Buddhism first become ubiquitous. This occurs only in the third century bce with inscriptions by the Mauryan King Ashoka and a few other archaeological remains. Even here, however, what can be said of Buddhism through archaeology is limited. It is only in the second century bce that the archaeological remains of Buddhism become suf- ficiently numerous as to be approached in a rigorous way. Despite these difficulties, in this chapter I will examine the archaeology of the sixth through third centuries bce in order to tease out what can, and cannot, be learned about the ways early Buddhists established and perpetuated their religion. For the most part, I hold off on discussing the material remains of Buddhism from the second century bce until Chapter 4. At times, however, an understanding of the third century bce depends on archaeological and epigraphic information from the second century bce or later.

      Given the paucity of Buddhist archaeological remains from the sixth through third centuries bce, an archaeological investigation of these early remains must rely on the retrospective use of more recent archaeological and historical evidence. For example, in the third century bce the Mauryan King Ashoka had columns erected at several locations that we know were Buddhist pilgrimage centers in later periods (e.g., Sanchi, Sarnath, Bodh Gaya). Unfortunately, later constructions and modifications of these

      centers have obscured all of the material traces prior to the second century bce. Viewed retrospectively, it is likely that Ashoka had these columns erected at locations that were already established pilgrimage centers in the third century bce, if not earlier, but without material evidence this claim must remain somewhat conjectural. Throughout this chapter, I ref- erence more recent archaeological sites and historical sources that are more thoroughly discussed in later chapters.

      I begin this chapter with a discussion of the broader social context of early Buddhism from the sixth through third centuries bce, with informa- tion drawn from archaeological and historical sources. This is followed by an examination of the limited archaeological evidence for early Buddhism in the sixth through fourth centuries bce. The bulk of this chapter exam- ines the archaeological evidence for Buddhism in the third century bce. While limited, this evidence suggests that in the third century bce, Buddhists were attempting, with only partial success, to establish a collec- tive Buddhist identity within a highly individualistic religious tradition. At the same time, rulers, like Ashoka, were attempting to legitimize their authority through association with the Buddha, the sangha, and other developing religious orders.


      Despite the limited number of archaeological remains of Buddhism from the sixth through third centuries bce, a great deal is known of the general archaeology and history of this period (see Allchin 1995; Basham 1967; Ray 1986; Singh 2008; Thapar 1997, 2002; see Figure 3.1). In archaeologi- cal and historical studies in India, the sixth through third centuries bce roughly correspond to the beginning of the Early Historic Period (c. 600 bce–200 ce). There is no clear break between the Early Historic Period and the preceding Vedic Period in North India. The selection of 600 bce as the start of the Early Historic Period in the Gangetic Plain has more to do with an increase in the available historical sources than with any clear archaeo- logical boundary. It is worth noting, however, that no existing Buddhist textual sources date to the life of the Buddha or the centuries immediately after his death. As discussed in Chapter 1, Buddhist and non-Buddhist textual sources that report on the Early Historic Period were collated as much as a millennium later. Since the extant versions were written in periods with much greater political centralization and stratification, it is likely that they project later political and religious attitudes into earlier time periods (see Thapar 2000).







      Chunnar Bharhut Quarries



      Rajgir Sudama, Lomas Rishi




      Arabian Sea

      Bay of Bengal

      Figure 3.1: Archaeological sites discussed in Chapter 3

      Thapar (1966, 1984, 2002) has argued that at the start of the Early Historic Period, polities of the Gangetic Plain included both kingdoms and republics. Polities with more autocratic and prominent kings at their centers dominated the heart of the Gangetic Plain. In the peripheries sur- rounding these kingdoms, polities retained the communal leadership that characterized the Vedic period. Here assemblies were created of village

      and tribal leaders who made communal decisions for these early republics. The Buddha was born to the chief of one of these Early Historic Period republics (the Shakyas) in what is now southern Nepal.

      More recently, several scholars have argued that these early republics and kingdoms can be subsumed within a broader rubric of city-states (see Chakrabarti 1995, 2001; Erdosy 1995; Kenoyer 1997). In this view, small states centered at individual cities were in constant communication and conflict with one another. Periodically one city-state might gain advantage over others, but territorial expansion was just as often followed by with- drawal. Particularly in the plains, cities began to build elaborate defensive walls. Merchants, traders, craftspeople, and religious specialists aug- mented the bulk of the population, who remained agriculturalists. Vedic rituals from earlier periods continued to be practiced, but during the Early Historic Period, a wide variety of religious traditions emerged to challenge the orthodoxy of Brahmanism that continued the practice of Vedic rituals. Textual sources that report on the Early Historic Period mention numerous wandering ascetics propounding a bewildering variety of reli- gious views. Over time, some of these wandering ascetics began to attract followers and settle in small monasteries throughout the India. Of these sects, Jainism and Buddhism became major religions that persist to this day. Originating in the sixth century bce, they share many common doc- trinal and ritual elements. Both saw the world in which we live as an illu- sion, both sought to end the cycle of rebirth, and both saw the cessation of rebirth as achievable through meditation and ritual acts. Both also downplayed the importance of caste in their teachings. Thapar (2002) has suggested that this allowed Jains and Buddhists to engage in trade more easily than Brahmans, who were forced to follow strict caste prohibitions. The result was that both Buddhism and Jainism thrived in the developing cities of the Gangetic Plain and spread rapidly from city to city from the

      fifth through third centuries bce.

      In sum, the picture that emerges of religious practice in the beginning of the Early Historic Period is of multiple overlapping religious sects, all competing for the support of the laity. Of these, Brahmanism, Jainism, and Buddhism have survived to this day in some form. Other sects (e.g., Ajivikas and Charvakas) have not (Basham 1967; Thapar 2002). While this book focuses on Buddhism, Buddhism was only one of many religious tra- ditions in simultaneous practice throughout India. Given the similarities in the doctrinal beliefs and ritual practices of these different religions, it is often difficult to separate the material remains of different religions. For example, while today stupas are associated with Buddhism, Jains also built and worshiped at stupas. Thus, in addition to all the standard

      problems of archaeological identification (e.g., dating, preservation, etc.) when evaluating the material evidence of early Buddhism, archaeologists must always be aware that specific remains could be the product of mul- tiple religious sects.

      Ashoka and the Mauryas

      In 326 bce Alexander the Great led his armies into northwestern India. Alexander was not pushing into unknown territory. He was consolidating his victory over the Persians who, for at least two centuries prior to the arrival of Alexander, had received tribute and support from the north- western portions of India (Briant 2002; McCrindle [1926] 2000; Thapar 2002). Alexander’s campaign in India lasted only a few years. With his death, following shortly after his departure from India, the provinces he had established slowly collapsed. Alexander himself had little impact on India. Rather than a singular moment of contact, Alexander’s campaign in India was only part of a long-standing connection between India, Persia, and the Mediterranean. The movement of both goods and ideas between these distant regions did not begin with Alexander, nor did they end with his departure.

      Alexander’s campaign in northwestern India did have one important consequence—it cleared away rivals, allowing Chandragupta Maurya to establish the Mauryan Empire in the area recently held by the Persians and Greeks (Thapar 2002). Over the next two centuries the Mauryan Empire expanded under several subsequent kings, consolidating power in the Gangetic Plain, with areas of significant influence in the peninsula. The southward expansion of the Mauryas is critical to the spread of north- ern religions and cultural practices into the peninsula beginning in the third century bce (Srinivas 1966).

      By 321 bce Chandragupta Maurya had defeated his main rivals, the Nandas. He continued his conquest of the Gangetic Plain until 297 bce, when he abdicated the throne to his son, Bindusara, and became a Jain monk. Bindusara continued his father’s military campaigns, principally in South India. Between them, Chandragupta and Bindusara success- fully established the first large empire in India. Despite this, the specific borders and forms of imperial control established by Chandragupta and Bindusara are not clear. For the most part, what is known of the Mauryan Empire must be inferred from the legacy of their successor, Ashoka.

      Ashoka became king in c. 272 bce and ruled for roughly 40 years. His historical importance is derived from the large number of inscriptions he

      had incised on rock faces and pillars he had erected throughout India (see Figure 3.2) and detailed accounts of Ashoka’s life in Sri Lankan and other texts (e.g., Ashokavadana [Strong 1983]). Critically, while the epigraphic and textual accounts agree on some elements of Ashoka’s biography, they differ on many others (see Strong 1983:ch. 1). The textual accounts of Ashoka provide a great deal of specific biographical detail about Ashoka’s

      Arabian Sea

      Bay of Bengal

      Figure 3.2: Distribution of Ashokan inscriptions in India

      life but, like historical accounts of the Buddha, they were collated and redacted several centuries after Ashoka’s death. Overall, the textual sources are most useful for understanding how Ashoka was remembered in later times, while his inscriptions provide a record of his actions in the third century bce (Strong 1983; Thapar 1997, 2002). This does not mean that Ashokan inscriptions should be read as a simple history of Ashoka’s life. With the exception of those inscriptions carved in South India, Ashokan inscriptions were written in the local languages of the area in which they are located. For this reason, it is likely that Ashoka intended them to communicate his proclamations to local audiences rather than simply to Mauryan officials (Sugandhi 2003). This suggests that Ashokan inscriptions are best viewed as legitimations (see Chapter 2)—statements intended to legitimize Ashoka’s authority in specific ways.

      In 260 bce Ashoka conquered the Kalinga, a kingdom in what is now coastal Orissa. In the 13th Major Rock Inscription, Ashoka states that “[o]n conquering the Kalinga the Beloved of the Gods felt remorse, for, when an independent country is conquered the slaughter, death, and deportation of the people is extremely grievous to the Beloved of the Gods, and weighs heavily on his mind” (Thapar 1997:255; see Box 3.1).2 Other inscriptions and textual sources document that the human suffering caused by this campaign led Ashoka to enter a period of religious contemplation and, at least according to the textual sources, convert to Buddhism. For the rest of his life, Ashoka is said to have engaged in numerous beneficial acts, including constructing hospitals, wells, and inns for the betterment of his subjects, redistributing relics of the Buddha from seven ancestral stupas to thousands of newly constructed stupas, and making large donations to the Buddhist sangha.

      The Pali Canon of Sri Lanka also records that Ashoka convened a large

      council of Buddhist monks designed to cleanse the sangha of disreputa- ble monks, to clarify and unify divergent interpretations of the Dharma, and to reduce the schisms that were developing between rival Buddhist monastic orders. An inscription found at several Buddhist sites sup- ports this textual account. In the Schism inscription, Ashoka proclaimed that, “[w]hoever creates a schism in the Order, whether monk or nun, is to be dressed in white garments, and to be put in a place not inhabited by monks or nuns” (Thapar 1997:262; see Box 3.1). According to the Pali Canon, the council concluded by dispatching missionaries to preach the

    2. Thapar’s translations of Ashokan edicts were intentionally loose—they were intended to be more easily accessible to non-specialists. There are substantive debates over the best way to translate some terms and phrases.

      Box 3.1


      1 2 T H M A J O R R O C K E D I C T

      The Beloved of the Gods, the king Piyadassi [Ashoka], honors all sects and both ascetics and laymen, with gifts and various forms of recogni- tion. But the Beloved of the Gods does not consider gifts or honor to be as important as the advancement of the essential doctrine of all sects. This progress of the essential doctrine takes many forms, but its basis is the control of one’s speech, so as not to extol one’s own sect or dispar- age another’s on unsuitable occasions, or at least to do so only mildly on certain occasions. On each occasion one should honor another man’s sect, for by doing so one increases the influence of one’s own sect and benefits that of the other man; while by doing otherwise one diminishes the influence of one’s own sect and harms the other man’s. Again, whoso- ever honors his own sect or disparages that of another man, wholly out of devotion to his own, with a view to showing it in a favorable light, harms his own sect even more seriously. Therefore, concord is to be commended, so that men may hear one another’s principles and obey them. This is the desire of the Beloved of the Gods, that all sects should be well-informed, and should teach that which is good, and that everywhere their adher- ents should be told, ‘The Beloved of the Gods does not consider gifts or honor to be as important as the progress of the essential doctrine of all sects.’ Many are concerned with this matter—the officers of Dhamma, the women’s officers, the managers of the state farms, and other classes of officers. The result of this is the increased influence of one’s own sect and glory to Dhamma.

      1 3 T H M A J O R R O C K E D I C T

      When he had been consecrated eight years the Beloved of the Gods, the king Piyadassi, conquered Kalinga. A hundred and fifty thousand people were deported, a hundred thousand were killed and many times that number perished. Afterwards, now that Kalinga was annexed, the Beloved of the Gods very earnestly practiced Dhamma, desired Dhamma, and taught Dhamma. On conquering Kalinga the Beloved of the Gods felt remorse, for, when an independent country is conquered the slaughter, death, and deportation of the people is extremely griev- ous to the Beloved of the Gods, and weighs heavily on his mind. What is even more deplorable to the Beloved of the Gods, is that those who dwell there, whether Brahmans, sravanas, or those of other sects, or householders who show obedience to their superiors, obedience to mother and father, obedience to their teachers and behave well and

      devotedly towards their friends, acquaintances, colleagues, relatives, slaves, and servants—all suffer violence, murder, and separation from their loved ones. Even those who are fortunate to have escaped, and whose love is undiminished [by the brutalizing effect of war], suffer from the misfortunes of their friends, acquaintances, colleagues and relatives. This participation of all men in suffering, weighs heavily on the mind of the Beloved of the Gods. Except among the Greeks, there is no land where the religious orders of Brahmans and sramanas are not to be found, and there is no land anywhere where men do not support one sect or another. Today if a hundredth or a thousandth part of those people who were killed or died or were deported when Kalinga was annexed were to suffer similarly, it would weigh heavily on the mind of the Beloved of the Gods.

      The Beloved of the Gods believes that one who does wrong should be forgiven as far as it is possible to forgive him. And the Beloved of the Gods conciliates the forest tribes of his empire, but he warns them that he has power even in his remorse, and he asks them to repent, lest they be killed. For the Beloved of the Gods wishes that all beings should be unharmed, self-controlled, calm in mind, and gentle.

      The Beloved of the Gods considers victory by Dhamma to be the fore-

      most victory. And moreover the Beloved of the Gods has gained this victory on all his frontiers to a distance of six hundred yojanas [i.e., about 1,500 miles], where reigns the Greek king named Antiochus, and beyond the realm of that Antiochus in the lands of the four kings named Ptolemy, Antigonus, Magas, and Alexander; and in the south over the Cholas and Pandyas as far as Ceylon. Likewise here in the imperial territories among the Greeks and the Kambojas, Nabhakas and Nabhapanktis, Bhojas and Pitinikas, Andhras and Parindas, everywhere the people follow the Beloved of the Gods’ instructions in Dhamma. Even where the envoys of the Beloved of the Gods have not gone, people hear of his conduct according to Dhamma, his precepts and his instruction in Dhamma, and they follow Dhamma and will continue to follow it.

      What is obtained by this is victory everywhere, and everywhere vic- tory is pleasant. This pleasure has been obtained through victory by Dhamma—yet it is but a slight pleasure, for the Beloved of the Gods only looks upon that as important in its results which pertains to the next world. This inscription of Dhamma has been engraved so that any sons or great grandsons that I may have should not think of gaining new conquests, and in whatever victories they may gain should be sat- isfied with patience and light punishment. They should only consider conquest by Dhamma to be a true conquest, and delight in Dhamma should be their whole delight, for this is of value in both this world and the next.


      B H A B R A I N S C R I P T I ON

      The king of Magadha, Piyadassi, greets the Order and wishes it prosper- ity and freedom from care. You know Sirs, how deep is my respect for and faith in the Buddha, the Dhamma and the Sangha [i.e., the Buddhist creed]. Sirs, whatever was spoken by the Lord Buddha was well spoken. And Sirs, allow me to tell you what I believe contributes to the long survival of the Buddhist Dhamma. These sermons on Dhamma, Sirs—the Excellence of the Discipline, the Lineage of the Noble One, the Future Fears, the Verses of the Sage, the Sutra of Silence, the Questions of Upatissa, and the Admonition spoken by the Lord Buddha to Rihula on the subject of false speech—these sermons on the Dhamma, Sirs, I desire that many monks and nuns should hear frequently and meditate upon, and likewise laymen and laywomen. I am having this engraved Sirs, so that you may know what I desire.

      R U MM I N D E I P IL L A R I N S CR I P T I O N

      The Beloved of the Gods, the king Piyadassi, when he had been conse- crated twenty years, came in person and reverenced the place where Buddha Sakyamuni was born. He caused a stone enclosure to be made and a stone pillar to be erected. As the Lord was born here in the village of Lumbini, he has exempted it from tax, and fixed its contribution [i.e., of grain] at one-eighth.

      N I G A L I S A G A R P I L L A R I N S C R I P T I O N ( PA R T I A L L Y D A M A G E D )

      The Beloved of the Gods, the king Piyadassi, when he had been conse- crated fourteen years, increased the stupa of Buddha Konakamana to double [its former size] . . . And when he had been consecrated . . . years he came in person, worshipped . . . brought . . .

      S C H I S M E D I C T

      The Beloved of the Gods orders the officers of Kausambi/Pata[liputra] thus:

      No one is to cause dissention in the Order. The Order of monks and nuns has been united, and this unity should last for as long as my sons and great grandsons, and the moon and the sun. Whoever creates a schism in the Order, whether monk or nun, is to be dressed in white garments, and to be put in a place not inhabited by monks or nuns. For it is my wish that the Order should remain united and endure for long. This is to be made known to the Order of monks and the Order of nuns. Thus says the Beloved of the Gods: You must keep one copy of this document and place it in your meet- ing hall, and give one copy to the laity. The laymen must come on every uposatha day [day of confession and penance] to endorse this order. The same applies to special officers who must also regularly attend the uposatha, and endorse this order, and make it known. Throughout your district you must circulate it exactly according to this text. You must also have this precise text circulated in all the fortress districts [under military control].

      Dharma throughout the world. It is through this final act that Buddhism reached Sri Lanka.

      Whatever the specific content of the varying textual accounts of his life, Ashoka is remembered as a prototypical divine king (chakravartin). Several scholars, however, have challenged the veracity of the textual accounts of Ashoka and the traditional histories based on them (Strong 1983; Thapar 1997). As argued by Strong, the textual accounts of Ashoka were not intended to be historical in the common sense of the word, but rather are best understood as legends. The authors of these legends intended them to illustrate the principles of Buddhism more than they intended them to accurately portray Ashoka’s biography. This insight leads to an alternative way to read Ashokan inscriptions. Rather than assume that Ashoka was a Buddhist, and that all his proclamations were intended to promote Buddhism, it is better to step back and let the content of the inscriptions speak for themselves (Thapar 1997). In this light, it appears that Ashoka was not solely a proponent of Buddhism. While a small body of his inscriptions (e.g., the Schism Inscription, the Bhabra Inscription, the Rummindei Pillar Inscription, and the Nigalisagar Pillar Inscription; see Box 3.1) may have been intended to promote Buddhism, the greater number are directed toward the creation of an imperial ideology that fostered a more inclusive empire (Strong 1983, 1994; Thapar 1997).

      In the 12th Major Rock Inscription, Ashoka makes clear his view that a ruler should support all of the religions within his domain, stating “The Beloved of the Gods, the king Piyadassi, honors all sects and both ascet- ics and laymen, with gifts and various forms of recognition” (Thapar 1997:255; see Box 3.1). While a preference for Buddhism may be pres- ent in a few inscriptions, Ashoka advanced a more generalized concep- tion of Dharma in the majority of his edicts. Specifically, he advocated pacifism, vegetarianism, and respect for family and authority (e.g., 13th Major Rock Inscription; see Box 3.1). Ashoka’s use of the term Dharma, while greatly influenced by Buddhist thought, had resonance throughout most of the religions of India (Strong 1983, 1994; Thapar 1997, 2002). In this sense, most Ashokan inscriptions were nonsectarian. It should also be noted that these broad concepts would have served imperial interests well. In essence, they promoted activities such as pacifism and respect for authority, which eased the process of governance. In this sense, Ashokan inscriptions can be productively viewed as legitimations. They were not, however, Ashoka’s only legitimations. Rather than discuss this further here, I will hold off until the full range of Ashoka’s legitimations has been introduced.

      Ashokan inscriptions also provide some information on the politi- cal structure and history of the Mauryan Empire. Several inscriptions record embassies and trade missions to Greece, Syria, and Egypt. The inscriptions also record the names of territories not under Mauryan control, principally several in the extreme south. They record the exis- tence of Mauryan provinces, tax collectors, and courts. Together all of these references suggest that the Mauryan Empire aspired to be a strong, centralized state, though it is likely that these aspirations were met only in North India, with Mauryan control of the South far more ephemeral. After the death of Ashoka, the Mauryan Empire gradu- ally diminished in power, with the final Mauryan king assassinated in 185 bce.


      Other than possibly the Buddha’s relics and some recent finds at Lumbini, there are no archaeological remains that unquestionably date to the life of the Buddha or the centuries immediately following his death. There are, however, several archaeological sites that various archaeologists have suggested, with varying degrees of certitude, date to the life of the Buddha. The difficulties faced with all of these sites are based on two prob- lems: dating and affiliation with Buddhism. For example, relying on a tex- tual account in the Sri Lankan Sutta Pitaka, a somewhat odd structure in Rajgir, has been identified as a monastery given to the Buddha by a phy- sician named Jivaka (Allchin 1995:246–247). Unfortunately, excavations at the Jivakarama monastery revealed no archaeological materials clearly datable to the Buddha’s lifetime. More problematically, the layout of the Jivakarama monastery, with its long elliptical rooms, resembles a simi- larly elliptical rock-cut cave in the nearby Nagarjuni hills (c. 40 km west of Jivakarama monastery). Inscriptions at this rock-cut cave and two others nearby record that they were dedicated by Dasaratha (Asoka’s grandson and immediate successor to the Mauryan throne) to the Ajivikas in c. 232 bce (Singh 2008:361). Given all of this, there is no compelling reason to identify the Jivakarama monastery as dating to the life of the Buddha, or even to identify the structure as being used by Buddhists rather than the Ajivikas. Similar problems plague the identification of the archaeological sites identified as dating to the time of the Buddha. For the most part, Buddhist sites said to date to the life of the Buddha have not been exca- vated, or if excavated, the historical accounts have been accepted without

      question. Thus, excavated sites like Shravasti and Kausambi are commonly said to date to the life of the Buddha, but no archaeological remains have been found that clearly demonstrate this. As for unexcavated sites, they are too numerous to mention.

      The two most convincing early Buddhist sites in India are Lumbini and a small stupa located in the ancient city of Vaisali, the capitol of the Licchavis. At Lumbini, recent excavations within the Maya Devi Temple revealed the foundations of a small wooden structure surrounding a small courtyard (Coningham et al. 2013). In the courtyard the excava- tors identified the traces of tree roots. Using a combination of optically stimulated luminescence (OSL) and carbon-14, the structure was dated to roughly the sixth century bce. The excavators argued that they found a small wooden temple that contained a tree as the focus of worship and pilgrimage. With their numerous radiocarbon and OSL dates, the exca- vators unequivocally demonstrated that the site was pre-Mauryan. The identification of the site as Buddhist is less certain, though extremely likely. It is possible that the temple was created by people other than Buddhists but, as the researchers convincingly argued, early depic- tions of tree-shrines in friezes on other early Buddhist sites make the Buddhist affiliation of the tree-shrine the most likely. Whatever the case, by at least the third century bce, Lumbini had become a Buddhist pilgrimage center. Overall, the recent excavations at Lumbini provide the best evidence of pre-Mauryan Buddhism to date. As will be dis- cussed further below, the finds at Lumbini suggest that early Buddhist temples were likely made of wood and the ritual foci may have been trees rather than stupas.

      The other possible archaeological evidence for pre-Mauryan Buddhism

      was found at Vaisali (Figure 3.3). Excavations in the 1950s revealed a brick stupa with a clay core that was enlarged three or four times (Allchin 1995; Mitra 1971; Sinha and Roy 1969:16–23). The excavators of this stupa believed that the final brick construction phase was contemporary with a nearby third century bce Mauryan column, thus making the earlier clay stupas pre-Mauryan. Critically, the excavators noted that the relic within the earlier clay stupa was disinterred at some later date. In the Sri Lankan Mahaparinibbana-sutta (Davids and Davids [1910] 2007), the Licchavis are listed as one of the eight groups that received a portion of the cremated remains of the Buddha and built a stupa to enshrine them in their capi- tal. The excavators argued that the small clay stupa they found at Vaisali was the original stupa of the Licchavis, and the subsequent removal of the relic evidence of Ashoka’s redistribution of the Buddha’s relics in the third century bce.

      Figure 3.3: Relic Stupa at Vaisali (c. third century bce)

      Courtesy of Hideyuki Kamon, Wikimedia Commons [Buddha’s ashes Stupa, Vaishali, Bihar.jpg] (converted from color to black and white)

      Unfortunately, the dating of the relic stupa at Vaisali is problematic. First, the Mauryan column referred to by the excavators was not located directly next to Vaisali, but rather next to another stupa located about

      2.4 kilometers away. The other primary evidence the excavators used to date the site was the presence of a specific type of pottery, Northern Black Polished ware. Though often interpreted as a time marker for the Mauryan period, even the most conservative modern analyses of Northern Black Polished ware find that it was present in North India between 550 and 100 bce (Allchin 1995:105). Given this, Northern Black Polished ware does not provide the necessary chronological resolution to determine whether the relic stupa at Vaisali dates to the time of the Buddha, the centuries after his death, or even the Mauryan period. Without more conclusive evidence, the antiquity of Vaisali remains in doubt (Allchin 1995:243; Coningham 2001:68; Mitra 1971:75). This small stupa at Vaisali may be one of the eight ancestral stupas in which the Buddha’s remains were ini- tially interred, or it may have been built in Mauryan times.

      Whether built at the time of the Buddha or in Mauryan times, the relic stupa at Vaisali is likely the earliest archaeologically known stupa. It is not, however, the only stupa that is known to have existed at this time. The partially damaged Nigalisagar pillar inscription in southern Nepal states that, “The Beloved of the Gods, the king Piyadassi, when he had been

      consecrated fourteen years, increased the stupa of Buddha Konakamana to double [its former size] . . . And when he had been consecrated . . . years he came in person, worshipped . . . brought . . .” (Thapar 1997:261). While the stupa that is mentioned in this inscription no longer exists, presum- ably, it existed for some time prior to its enlargement by Ashoka—though exactly how long it existed prior to its enlargement cannot be determined from the inscription. As such, Vaisali and the Nigalisagar pillar inscrip- tion demonstrate that stupas and ritual centered on stupas date to the Mauryan period, if not earlier.


      The quantity and quality of archaeological material available for study increases markedly in Mauryan times. This material includes sites like the relic stupa at Vaisali (discussed above) and other Buddhist sites, Mauryan pillars, and Ashokan inscriptions. It might seem odd to include Ashokan inscriptions in a list of archaeological materials. Following the perspec- tive of materiality (see Chapter 2); however, it must be remembered that inscriptions are not only texts, but also material things. The form, place- ment, and physical properties of Ashokan inscriptions were not simply a reflection of Ashoka’s power, but rather were instrumental in his asser- tions of power. This does not mean that the content of the inscriptions is unimportant, but rather that when combined with their physicality, a more nuanced understanding of their religious meaning and political function is possible.

      Inscriptions of Ashoka

      Traditionally, the locations of Ashokan inscriptions in India have been interpreted as evidence for direct Mauryan control over a substantial por- tion of the peninsula (see Figure 3.2). More recently, some have noted that the distributions of the southern inscriptions are tightly circumscribed in the Deccan Plateau and Krishna River valley (Fussman 1988; Habib and Habib 1990; Ray 1986; Sinopoli 2001; Sugandhi 2003, 2008; Thapar 2002). Several gold fields lie within this area and likely account for the Mauryan interest in the region. Other than this limited area, there is little evidence for Mauryan presence in the South. Further, unlike the Ashokan inscriptions found in North India, the southern inscriptions were written

      in Brahmi, rather than the local languages of the areas in which they were found. Unlike the inscriptions in the North that were intended to be read to broader populations, in the South it appears that the inscriptions were intended only for a smaller number of Mauryan officials fluent in Brahmi (Sugandhi 2003). Given this, it seems likely that Mauryan political control was weak in South India, focusing on areas of the greatest agricultural or mineral wealth (Sinopoli 2001; Sugandhi 2003, 2008). While there may have been a fairly large degree of political centralization in the core of the empire in the North, in the peripheries Mauryan presence was likely less dramatic.

      Ashokan inscriptions have been found on rock outcrops in North and South India and on stone pillars found only in the North (see Figure 3.2). In both cases, but particularly in the case of the pillars, inscriptions were usually located in prominent locations adjacent to important religious centers or settlements. The exceptions to this pattern occur in South India, where Sugandhi (2003, 2008) has shown that Ashokan inscriptions were often carved in peripheral locations, distant from major settlements. Ashoka had identical (or nearly identical) inscriptions carved in numerous locations throughout India. Using the criteria developed by Moore (1996; discussed in Chapter 2), Ashokan inscriptions can be productively viewed as material legitimations. That is, most Ashokan inscriptions are perma- nent, proximate to important places, ubiquitous, large, and highly visible. As for those less visible, peripheral inscriptions in South India, their loca- tions might suggest that Ashoka had less authority in these more distant peripheries of the Mauryan state.

      Mauryan Pillars

      In addition to the inscriptions, Ashoka and other Mauryan kings had pil- lars erected at important locations throughout North India. These pillars were typically 12–14 meters high and were highly polished (see Figure 3.4). Mauryan Pillars consisted of a gently tapered column with an elaborately carved capital. The capitals typically had three parts: at the bottom was an inverted lotus, above this was a round (sometimes square) drum, and finally an animal carving, usually lions. Save for a few constructed from stone quarried near Mathura, all pillars were made of sandstone from the Chunar quarries in the heart of the Gangetic Plain near modern Varanasi. Based on archaeological research at these quarries (Pant and Jayaswal 1990), it appears the pillars were initially quarried and roughed out at Chunar. The final carving and polishing occurred where they were erected,

      Figure 3.4: Mauryan pillar at Lauriya and Lion capital from Sarnath (c. third century bce) Courtesy of the Digital South Asia Library and the American Institute of Indian Studies (Accession Nos. 28816 and 25268)

      often several hundred kilometers from the quarries. The Mauryan pillar at Sanchi, for example, is roughly 550 kilometers from Chunar. The quarry- ing, transport, and finishing of the pillars signify a massive undertaking by the Mauryan state. All told, at least fifteen Mauryan pillars have been found in North India, though fragments of other, highly polished pillars could indicate several more. Ashokan inscriptions are found on ten of these pillars. This does not necessarily mean these pillars were erected on Ashoka’s orders. It is also possible that Ashoka had his inscriptions carved onto pillars erected by previous kings.

      The symbolic meaning of the capitals of Mauryan pillars have been debated for years (Agrawala 1965; Gupta 1980; Leidy 2008). Those who see Ashoka as primarily Buddhist interpret the capitals as symbols of Buddhism. Those who see Ashoka as promoting a nonsectarian concep- tion of Dharma interpret the capitals as more general symbols of ruler- ship. For the most part, there is no convincing evidence to decide between these two interpretations. That said, in some specific cases the symbolism

      does seem specifically Buddhist. The capital at Sarnath, for example, has four wheels carved on its drum (see Figure 3.4). Critically, this is the only Mauryan capital that includes wheel motifs. It seems unlikely that it is merely coincidental that the capital was located at Sarnath—the location of the Buddha’s first sermon, the place where the Buddha first turned the wheel of Dharma. Rather, it seems very likely that the wheel motif, at least at Sarnath, symbolized the wheel of Dharma in the specifically Buddhist sense of the term.

      The use of the wheel motif at Sarnath, and only Sarnath, might also shed light on Ashoka’s use of the term “Dharma” in his inscriptions. That is, while Ashoka might have used the term in a broadly nonsectarian sense, Buddhism likely informed his specific conception of Dharma. More so, a significant number of Mauryan pillars are located at sites known to be Buddhist pilgrimage centers in later times (e.g., Sanchi, Sarnath, Lumbini, Bodh Gaya), or have inscriptions that specifically mention Buddhist estab- lishments (e.g., Nigalisagar Pillar Inscription and the Rummindei Pillar Inscription at Lumbini [see Box 3.1]). Finally, the schism edict seems to have only been carved onto pillars located at Buddhist sites (e.g., Sanchi and Sarnath).

      The Legitimations of King Ashoka

      All told, Mauryan pillars and the Ashokan inscriptions carved onto the pillars constitute the clearest evidence that Ashoka, while a supporter of numerous religious sects, was principally a proponent of Buddhism. Faith aside, Ashoka’s pillars and inscriptions were also material legitima- tions. That is, their permanence, location proximate to important places, ubiquity, size, and visibility suggest that they were intended to assert his authority over the subjects of his empire. This interest is even shown in the stone used to construct the pillars. In an article on Buddhist land- scape perceptions, Barnes (1999) noted that East Asians often revere distant mountains and caves. In addition to constructing shrines in the mountains, East Asians transport stone from the mountains to create Buddha images in their cities, towns, and villages. “In so doing, they make every landscape of their occurrence a Buddhist landscape, facilitating a metaphysical awareness as persons go about their daily business” (Barnes 1999:121). In the terms of Peirce’s semiotics, the stone used for the con- struction of local images indexed the mountains that East Asians revered. That is, the use of non-local stone indexed the distant, sacred locations from which the stone was quarried.

      As in the movement of stone in East Asia, the transport of Chunar sand- stone from the center of the Gangetic Plain—the core of the Mauryan State—and its prominent display in the pillars erected at important sites indexed the core of the empire. As such, the movement of Chunar sand- stone was a legitimation, an attempt to project centralized authority over a geographically extensive and diverse empire. While no materials were transported, the ubiquity of Ashokan inscription (the carving of near iden- tical inscriptions throughout the empire) also served to index the centrality of the Ashoka in the peripheries of the Mauryan state. Critically, however, Ashoka was not simply erecting monuments that were to become important places; rather, he erected monuments and carved inscriptions at locations that were previously associated with the Buddha (e.g., Lumbini, Bodh Gaya, Sarnath) and other religious sects (e.g., the Ajivikas of the Barabar Caves). Ashoka’s legitimations were attempts to associate himself with Buddhist monasteries and pilgrimage sites that, to some degree, had their own importance and authority distinct and separate from Ashoka’s authority.

      Buddhist Monasteries and Chaityas in the Mauryan Period

      The earliest archaeologically known Buddhist monasteries in India were constructed in the Western Ghats in the first century bce (see Figure 3.1). This is not to say that these were the earliest Buddhist monasteries. As discussed above, Ashokan inscriptions testify to the existence of the Buddhist sangha from at least the third century bce. While earlier mon- asteries were constructed of perishable materials, the monasteries of the Western Ghats were carved directly into cliff faces. For this reason, the monasteries of the Western Ghats are preserved for archaeological inves- tigation, while earlier Buddhist monasteries are not. There are, however, telling clues in the construction style of some of the monasteries in the Western Ghats that suggest that second- and third-century Buddhist monasteries were simple structures made of wood and thatch.

      The earliest archaeologically known Buddhist monastery in the Western Ghats is Kondivte (c. 100 bce). The site consists of caves used as monastic living quarters (viharas) and one rock-cut worship hall (chaitya) containing a stupa (Mitra 1971; Nagaraju 1981). The chaitya at Kondivte is divided into two spaces: a small circular room containing a stupa and a larger assembly hall located in front of the circular room (see Figure 3.5). Importantly, the circular room was deliberately carved to resemble a thatch hut with wattle-and-daub walls. Windows, carved to resemble wooden grills, were even included. The assembly area was similarly carved to resemble a

      Figure 3.5: Kondivte (c. first century bce)

      Courtesy of the Digital South Asia Library and the American Institute of Indian Studies (Accession No. 84540)

      wooden hall. Because of this, many have argued that Kondivte was mod- eled on earlier, perishable chaityas that have not been preserved in the archaeological record (Brown 1965; Fergusson and Burgess [1880] 1988; Piggot 1943). While not a chaitya itself, the small sixth century tree-shrine found at Lumbini further testifies to a tradition of wooden construction in early Buddhist sites (Coningham et al. 2013).

      It is difficult to determine the antiquity of the presumed wood and thatch prototypes for Kondivte, but the layout and form of three archaeo- logical sites might suggest the prototypes date at least to Mauryan times (c. 250 bce). These sites are Bairat, Lomas Rishi, and Sudama. Lomas Rishi and Sudama are rock-cut caves located in the Barabar Hills in northeast- ern India (Brown 1965; Fergusson and Burgess [1880] 1988). In terms of design and layout, they are almost identical to Kondivte, save for the omis- sion of a stupa in the circular room (see Figure 3.6). An Ashokan inscrip- tion carved onto the door of Sudama indicates that these caves were not used by Buddhist monks, but rather by the Ajivikas—a contemporary reli- gious order similar, in many ways, to both Buddhism and Jainism (Dundas 1992). Thus, Lomas Rishi and Sudama cannot be used to specifically dem- onstrate that Buddhists placed stupas within thatch huts. Instead, Lomas Rishi and Sudama demonstrate that the architectural layout of Kondivte, in a general sense, was modeled on wood and thatch prototypes that were likely used by multiple religious sects for at least one or two centuries prior to the construction of Kondivte in c. 100 bce.

      Bairat is the only possible example of a freestanding Buddhist chaitya

      from the Mauryan period (see Figure 3.7). It sits on an isolated hilltop in the modern state of Rajasthan. Excavations by Sahni (1937) were

      Figure 3.6: Sudama (c. third century bce)

      Courtesy of the Digital South Asia Library and the American Institute of Indian Studies (Accession No. 28813)

      reinterpreted by Piggot (1943) as showing a brick stupa surrounded by

      (1) a row of wooden columns, which supported a domed roof and defined a circumambulatory path, and (2) a brick wall that defined the perimeter of the circular chaitya. Perhaps at a later date, a rectangular hall was erected to the east of the circular chaitya, creating a space remarkably similar in layout and form to Kondivte, Lomas Rishi, and Sudama. Further, like these other sites, the use of wooden columns and a domed roof recall ear- lier wood and thatch prototypes. Bairat was dated by the presence of pillar fragments of polished Chunar sandstone and a nearby Ashokan inscrip- tion. Given all of this, it seems likely that the circular chaitya at Bairat dates to Mauryan times, but this cannot be stated with certainty.

      Kondivte, Lomas Rishi, Sudama, and, perhaps, Bairat are the only sur- viving archaeological examples of the design and layout of the earliest Buddhist chaityas. Using the layout of these sites as a guide, the earliest monastic stupas were likely placed in small, circular huts (Brown 1965; Fergusson and Burgess [1880] 1988). In terms of ritual, these perishable huts would have allowed for the practice of circumambulation, but would not have provided space for any larger assemblages of monks and nuns.

      Figure 3.7: Bairat (c. third century bce)

      Piggot 1943:3; courtesy of Antiquity.

      Later, judging by the layout of Kondivte and Bairat, wooden halls were erected to one side of the perishable huts, allowing for larger assemblages of worshipers to gather and engage in collective rites while individuals could simultaneously circumambulate the stupa to accumulate personal merit. Judging from the layout of the early tree-shrine at Lumbini, the architecture of early Buddhist pilgrimage centers were also constructed of wood, though perhaps with circumambulation around trees rather than stupas.

      Another element of these early chaityas also bears mentioning. It appears that all were, in the terms of Durkheim ([1915] 1995:44) “set apart,” or in the terms of Turner (1967), liminal (see discussion in Chapter 2). All four of these sites were located on isolated hilltops or hillsides. Further, within Kondivte, Bairat, and, presumably, the wooden prototypes on which they were based, stupas were shielded from view by architectural façades. While modern approaches to the study of religion often downplay the separation of religious, political, and economic activities, the geographic placement and architectural layout of these sites suggest that early Buddhists sought,

      at least in some instances, to separate the experience of ritual and worship from more mundane, worldly concerns. This should not be taken to mean that these sites were entirely divorced from political considerations. The presence of Ashokan inscriptions at Sudama, Lomas Rishi, and Bairat all testify to the potential power that association with these isolated sites could confer. As will be discussed in later chapters, throughout the his- tory of Indian Buddhism, a persistent tension existed over the desire for isolation and separation on the one hand, and the need to integrate the sangha within the economic and political systems on the other. Relying on Kondivte and Bairat, this tension appears to be evident from the very beginning of the archaeological sequence.


      With the exception of Lumbini, and the possible exception of Vaisali, the material record of Buddhism available for archaeological study dates, at the earliest, to the third century bce (Coningham 2001). A semiotic examination of the earliest stupas, therefore, must be based on Buddhist textual sources and conjecture from later, archaeologically known, stupas. The Mahaparinibbana-sutta from Sri Lanka provides a detailed account of the death of the Buddha and the construction of the eight ancestral stupas in which his remains were initially interred (Davids and Davids [1910] 2007). There are several problems with using the Mahaparinibbana-sutta to analyze what the earliest Buddhist stupas would have looked like. First, the extantversions of the Mahaparinibbana-sutta dates to the first few centuries ce—500 to 800 years after the events they claim to record (Schopen 1997). Further, the Mahaparinibbana-sutta is not even preserved in Buddhism’s birthplace, but rather in the Buddhist chronicles of Sri Lanka. While many Buddhist scholars argue that the Mahaparinibbana-sutta can be used to infer early Buddhist theology (Bareau 1974; Frauwallner 1956; Lamotte 1988), other Buddhist scholars are highly critical of this claim (Schopen 1997; Trainor 1997). I cannot resolve this debate. Rather, it seems more realistic to consider any claims concerning the earliest forms of Buddhism derived from the Mahaparinibbana-sutta as provisional—pending the dis- covery of further archaeological and inscriptional evidence from the time periods in question (Coningham 2001; Fogelin 2007b; Schopen 1997:ch. 1; Trautmann and Sinopoli 2002).

      Neither archaeology nor textual sources provide reliable evidence for what the earliest stupas would have looked like. Even so, I believe an anal- ysis of the textual account presented in the Mahaparinibbana-sutta has

      value as a baseline for examining the known changes in stupas from third century bce onward. In some sense, the Mahaparinibbana-sutta describes an ideal, ancestral stupa. It presents an account of the origin of stupas that informed the shape, location, and contents of later archaeologically known stupas. The earliest archaeologically known stupas were designed to look like ancestral stupas and were often believed to contain relics of the Buddha. It is possible that the ancestral stupas and relics never existed. That said, the memories were real, even if the ancestral, pre-Mauryan stu- pas and relics may not have been. I begin, then, with a discussion of the earliest stupas based on the Mahaparinibbana-sutta.

      The eight ancestral stupas described in the Mahaparinibbana-sutta, perhaps evidenced in the clay stupa at Vaisali, consisted of two major elements: the cremated remains of the Buddha (corporeal relics) and the mound of earth or clay (anda) encasing them. Both the mound and the ashes have their own semiotic significance. While it is tempting to view the corporeal relics of the Buddha as indexes of the Buddha, later Buddhist inscriptions and textual sources demonstrate that they are best understood as icons of the Buddha. Their iconicity is shown through their innate relationship with the Buddha. Corporeal relics literally were the Buddha, and Buddhist monks, nuns, and devotees treated them as such.

      Relying primarily on the earliest extant inscriptional evidence from the first and second centuries bce, Schopen (1997:131–133; see also Strong 2004; Trainor 1997) has argued that early Buddhists viewed corporeal rel- ics as the living presence of the Buddha. This is shown, in part, through a second-century bce inscription on the broken lid of a relic casket from Shinkot in modern Afghanistan, which reads: “ . . . [on] the 14th day of the month Karttika, the relic of the Blessed One Sakyamuni which is endowed with life was established” (Schopen 1997:126 [italics added]; Lamotte 1988; Majumdar 1937). Other inscriptions and textual sources further demonstrate that, in the first few centuries bce, actions directed toward the Buddha’s relics were accorded identical standing as actions directed to the Buddha himself. Stupas were even accorded legal standing as a people with property rights (Schopen 1997:128–131). Schopen concludes that in early Buddhism, there was a “functional equivalence of the relic and the living Buddha” (Schopen 1997:131).

      An index, by definition, requires the recognition that one sign indi- cates the status of another sign. In early Buddhism, no such recognition seems to have occurred in regard to corporeal relics. While icons are often understood as having an innate relationship with their referent due to the sensual perception of a sign (e.g., sight, smell, taste, etc.), there are

      other ways in which an innate relationship can be created. In the case of early Buddhist corporeal relics, sensual perception was not possible since the relics were interred within earthen mounds. Rather, corporeal relics were icons of the Buddha through a material association with Buddha. Corporeal relics were the Buddha because they consisted of the same mat- ter as the Buddha. The same cannot be said of the stupas that contained possessions of the Buddha (e.g., begging bowl or robes) or that marked prominent locations in the Buddha’s life. These types of relics are best understood as indexes of the Buddha, though they were constructed in the same manner and followed the same religious sanctions as stupas containing the corporeal remains of the Buddha.

      The second semiotic element of ancestral stupas was the anda (large

      mound of earth) in which the relic was placed. This can be most produc- tively understood as an index. The anda served to indicate the presence of the relic—the icon—within. The scale of the anda would have made the stupa a prominent feature in the landscape. The scale of the anda would have also served as an index of the prominence or power of the unseen icon or index interred within.

      It is also possible that stupas as a whole were symbols, in Peirce’s sense of the term, of Buddhism in the third century bce. By the second century bce (see Chapter 4), it is clear that stupas served as abstract signs symbolizing Buddhism as a whole in the same way that crosses symbolize Christianity today (Dehejia 1997). Given the centrality of stupas in Buddhist ritual in the third century bce, it seems likely that they would have served as a symbol of Buddhism at that time. The difficulty lies in demonstrat- ing this. As discussed earlier, the interment of the cremated remains of important individuals was not limited to Buddhists in the Early Historic Period; Jains and others also venerated stupas. Rather than symbolizing Buddhism specifically, stupas may have had more generalized religious sig- nifications. Using the cross as an example, is it a symbol of Lutheranism (or any other specific sect of Christianity) specifically, Christianity gener- ally, or both? Given that this question is difficult to answer in modern contexts, it would seem impossible to answer in regard to stupas in the third century bce. There are, however, subtle differences in the depictions of crosses by different Christian sects. Catholic, Greek Orthodox, and Methodist crosses, for example, are identifiable to those who have been made aware of the differences. The same could be true of stupas in the third century bce and, as discussed in Chapter 4, subtle identifying differ- ences in Buddhist stupas are evident in the second century bce. However, given the limited remains of stupas from the third century bce, it is not possible to determine whether or not these subtle indicators existed at

      that time. Given all of this, it is not possible to determine whether stupas symbolized Buddhism in the third century bce or earlier. In sum, ances- tral stupas were icons of the Buddha encased within indexes of his pres- ence. Stupas may also have been more abstract symbols of Buddhism, but that cannot be conclusively demonstrated. Whatever the case, it appears that by the third century bce, stupas were the foci of Buddhist ritual at both monastic and pilgrimage centers. As will be discussed in later chap- ters, over time the design and form of stupas slowly changed. With these changes in design came changes in the semiotic significance of stupas that inform on the long-term development of Indian Buddhism.

      From at least the second century bce onward, ritual performed at stupas has taken two distinct forms (Fogelin 2003, 2006). The first, cir- cumambulation, consists of an individual walking around a stupa, accu- mulating merit to speed his or her path to nirvana. Circumambulation is often performed in clearly demarcated paths (pradakshina) immediately surrounding stupas. While many people may circumambulate simultane- ously within this path, circumambulation does not require the presence of other worshipers. Even when many are engaged in circumambulation, it is ultimately a form of ritual connecting a single devotee and the Buddha. Collective rites are also performed at stupas, often in large courtyards out- side the circumambulatory paths. Here devotees engage in group worship of a stupa. Collective rites may consist of quietly listening to a sermon by an officiating monk or more festive, communal worship without ritual leaders. Whichever form collective rites take, they serve an important function in terms of establishing and revivifying a common Buddhist identity. As discussed in Chapter 2, Durkheim ([1915] 1995) argued that social solidarity is fostered by the performance of collective rites. From this perspective, the collective rites performed at stupa complexes promote a common Buddhist identity among participants. Circumambulation, in contrast, promotes a more individual or personal relationship between a devotee and the Buddha.

      Given the limited archaeological evidence of Buddhism in the two centu- ries following the Buddha’s death, it is not possible to determine if circum- ambulation or collective rites were performed prior to the third century bce—though it seems likely that the tree-shrine at Lumbini would have been a focus of Buddhist pilgrimage. Archaeological evidence does sug- gest, however, that circumambulation and collective rites were performed at some stupas in the third century bce. At Kondivte and Bairat, circum- ambulatory paths encircle the small stupa (see Figures 3.5). Presumably the ancestral wood and thatch chaityas from the third century bce would have had a similar layout. Though no clearly demarcated circumambula- tory path was found at Vaisali, this does not mean that circumambulation

      was not practiced there—devotees could have circumambulated without the aid of a clearly demarcated path. Thus, it seems likely that Buddhists circumambulated stupas by at least the third century bce, at stupas within chaityas and, perhaps, at stupas at pilgrimage centers. A similar argument can be made concerning the performance of collective rites.

      At Vaisali, no information is available as to whether or not a defined courtyard was constructed to demarcate a space for collective rites. Even if no courtyard was created, early Buddhists may still have engaged in collective rites in the open space surrounding the stupa without leaving material traces of the performance of these rites. At Kondivte, Bairat, and the ancestral chaityas on which they were based, small halls were con- structed facing the stupa chamber (see Figures 3.5). While small, these spaces would have allowed for a limited number of worshipers to engage in collective rites. That said, given the small size of the hall at Kondivte and Bairat, it seems these collective rites would not have involved large numbers of people.

      There is one final attribute of the design of Kondivte, Bairat, and the ancestral chaityas on which they were based that bears mentioning—the layout privileged the practice of circumambulation over the practice of collective rites. Those engaging in collective rites in the hall had only obstructed views of the stupa through the narrow doors (see Figure 3.5). More so, this view would have been periodically interrupted by the pas- sage of those engaged in circumambulation. Since circumambulation was a more solitary act, the layout of Kondivte, Bairat, and their wooden ancestors were more conducive to the individual ritual and less conducive to the observance of collective rites. This detail is critical for understand- ing the difficulties that early Buddhists faced when attempting to routin- ize the charisma of the Buddha and establish a coherent community of Buddhists in the centuries after his death.


      As would be expected in any religion, early Buddhists in India were forced to contend with problematic social contradictions within their belief sys- tem (see discussion of incoherence in Chapter 2). Among the more impor- tant contradictions in early Buddhism was the emphasis on the individual attainment of nirvana versus the need to promote and maintain the com- munity of Buddhists (Bailey and Mabbett 2003; Fogelin 2011). Many of the beliefs, practices, and rites of early Buddhism were centered on the individual attainment of enlightenment through private meditation with little regard for the welfare of the broader Buddhist community. Even

      the biography of the Buddha reinforced these individualizing tenden- cies. While he served as an exemplar and teacher of Buddhist principles, the Buddha attained nirvana individually and left the sangha behind.3 Since the Buddha’s death, his followers have faced the difficulty of forg- ing a cohesive community from the individualistic teachings of early Buddhism. To a large degree, the disjuncture between individualism and communalism in Buddhism has persisted to the present day. Much of my analysis in subsequent chapters consists of an examination of how this long-lasting disjuncture played out in different periods, and how Indian Buddhists in different periods sought to resolve, exploit, and ameliorate this disjuncture.

      This approach to the study of Indian Buddhism is strongly informed by anthropological studies of nineteenth- and twentieth-century Buddhism in Sri Lanka, Thailand, and Burma, particularly the work of Tambiah (1976, 1982, 1984) and Carrithers (1979, 1983; see also Strenski 1983) on the role of asceticism within the larger community of Buddhists monks. As stated by Tambiah (1984:2), the focus of these anthropological studies was the relationship “between town-and-village dwelling monks and forest dwelling monks; between the vocation of scholarship and books and that of meditation; and between a (more) ascetic and reclusive mode of life and a (more) active laity-oriented life, which tended in various combinations to divide, if not bifurcate, the sangha.” While forest monks often lived in proximity to other forest monks, most of their time was spent meditating in isolated caves, huts, or other similar locations. Tambiah and Carrithers examined the relationships between the vast majority of monks who lived communally within monasteries and the fewer forest monks who lived mostly solitary lives as ascetics. While forest monks often explained their desire for isolation by pointing to the corrupting influence of monastic life, equally important was the stated desire to return to a more medi- tative, solitary form of asceticism as exemplified in the biography of the Buddha (Carrithers 1983:ch. 3; Tambiah 1984:ch. 7). Much of Tambiah’s and Carrithers’s examinations of forest monks focused on the processes that led to the “domestication” of forest monks on the one hand, and a tendency for some village monks to abandon monasteries for life in the forest on the other—the push and pull between the solitary life of the for- est and the communal life of village monasteries.

    3. By the middle of the first millennium ce, Mahayana Buddhists postulated the existence of Bodhisattvas, beings who had achieved enlightenment but delayed their ascension to nirvana in order to aid those who had yet to achieve enlightenment (see Chapter 5).

      While partially inspired by the work of Tambiah and Carrithers, the focus of this study is slightly different. Where Tambiah and Carrithers examined the tension between individualizing and communal elements of Buddhist practice through the comparison of forest monks and village monks, I am examining the contradiction between individual and com- munal tendencies within Buddhism as a whole. Phrased another way, I am examining how both the sangha and the laity overcame the individual- izing aspects of their religion while allowing for its limited expression. While not the focus of either Tambiah’s or Carrithers’s research, this focus is congruent with some of their observations. For example, Tambiah (1984:53) suggests that

      while these dualities of oppositions may seem sharply defined and mutually exclusive, it is important to realize that ideally the [monk] should combine both vocations and that in actuality one vocation does not necessarily exclude the other [I]n Thailand, as elsewhere, town-dwelling monks and urban

      monasteries have promoted meditation, and there have been forest-monk communities that have produced well-known scholars.

      Similarly, Carrithers (1983:142–143) notes that forest monks in Sri Lanka are far more faithful than village monks in their observance of the uposatha ceremony, a fortnightly gathering of monks where they com- munally recite 227 rules of the monastic code. Where monks living com- munally in monasteries need less frequent reminder of their communal identity, Carrithers’s argues, forest monks living in relative isolation do. Whether meditation in urban monasteries, or the uposatha ceremony among forest monks, the modern Buddhist communities observed by Tambiah and Carrithers were forced to balance the need for individual and communal expression of their religion. The difference between forest and village monks, therefore, is not that one is ascetic and the other commu- nal, but rather that each group held opposing views on the proper balance between individualism and communalism. I argue that the same tension existed among the Buddhist laity, and that just as different members of the sangha had different views on the proper balance, different groups of lay Buddhists also disagreed on the proper balance between the individual and communal ritual.

      In addition to, and separate from, their studies of nineteenth and twen- tieth century Buddhism, both Tambiah and Carrithers employ ancient Buddhist texts to demonstrate that the division between forest and village monks was a long-lasting tension in Buddhism (Carrithers 1983:ch. 5; Tambiah 1984:ch. 2). Tambiah and Carrithers differ, however, in just how

      ancient this division might be. Where Tambiah (1984:ch. 7) sees the dual- ity of forest and village monks as a tension originating in India at the time of the Buddha or during the Mauryan period, Carrithers (1983:140) sees this duality emerging only in the last millennium in Sri Lanka as early ascetic sects became “domesticated” by and dependent on “sedentary, agrarian societies.” Prior to the last millennium, Carrithers argues that monks lived a more purely ascetic lifestyle, living in monasteries only dur- ing the rainy season.

      By at least the first century bce, the earliest period for which Buddhist monasteries are archaeologically ubiquitous, Buddhist monks and nuns in India were living year-round in compact, well-apportioned monasteries. As discussed earlier, a few Ashokan inscriptions demonstrate that at least a few Buddhist monasteries existed in the third century bce. In contrast, with the exception of some isolated cave sites from Sri Lanka dating to the first and second centuries bce (Carrithers 1983:6; Coningham 1995, 2001), there is little archaeological evidence for early Buddhist forest monks in India. It should be noted, however, that the lack of archaeological traces of forest monks could well be the product of the ephemeral nature of their habitations (Coningham 2001). Given the prominence of solitary ascetics in the religious literature of early India, it does seem possible that ascet- ics were present, even if their habitations have not been archaeologically identified. The point here is that, contrary to Carrithers’s claim, at least some member of the sangha were living collectively in Indian monaster- ies from at least the third century bce. Further, based on Tambiah’s and Carrithers’s interpretations of early Buddhist texts, it should be expected that the early Buddhist sangha suffered from contradictions between the ascetic and communal desires of its membership.4

      The tension between the individual and the group in early Buddhism can be seen in early textual descriptions of the two paths to becoming an arhat, a “worthy one” who has become enlightened (Lopez 2001; Tambiah 1984:13–14). One path is taken by pratyekabuddhas, who achieve enlight- enment in solitude. The other is taken by shravakas, or “listeners,” who learn the Buddha’s teachings from others. The path of shravakas concords well with the lifestyles of the early Buddhist sangha. In early Buddhist

    4. Tambiah (1984) and Carrithers (1983) are not basing their interpretation of early Buddhism in India on an analogy with nineteenth- and twentieth-century Buddhism in Thailand and Sri Lanka. Rather, they identify the duality between the ascetic and communal desires of the sangha in the nineteenth and twentieth centu- ries and separately identify a similar duality in early Buddhism through examina- tions of early Buddhist texts. As for how this duality was specifically manifested in early Indian Buddhism, neither Tambiah, Carrithers, nor I would suggest that it

monasteries, abbots stood at the top of a rigid hierarchy, teaching their subordinates the intricacies of Buddhist theology. The existence of pra- tyekabuddhas, however, undercut the role of abbots, monasteries, and even the value of a cohesive community of Buddhists by offering an alternative—individual—path to enlightenment.

In this light, the spatial organization of Kondivte and Bairat can be understood. In essence, these early chaityas were divided into two distinct ritual spaces—a circumambulatory path immediately adjacent to the stupa and a small hall abutting the circular chamber (see Figures 3.5). As such, early Buddhists designed their ritual spaces to allow for the practice of both individual and communal ritual—early Buddhists designed their ritual spaces in such a way as to help ameliorate the disjuncture between individualism and communalism. Following the insights of materiality, this layout was not merely the reflection of persistent disjunctures, but rather a creative material strategy that helped early Buddhists ignore this fundamental disjuncture.

The designs of Kondivte and Bairat, however, were not completely suc- cessful. The separation between the two ritual spaces was primarily at the expense of those engaged in collective rites in the halls abutting the stupa chamber. Devotees in the hall had only an obscured view of the central stupa through the door to the stupa chamber (see Figures 3.5), and this view would have been periodically interrupted by the passage of other devotees engaged in circumambulation. As will be discussed in Chapter 4, in the first and second centuries bce, Buddhists redesigned their chaityas in ways that served to more effectively conceal the disjunctures between asceticism and communalism—though, interestingly, by privileging col- lective rites rather than circumambulation.


Given the limited Buddhist remains from the sixth through fourth cen- turies bce, archaeology cannot substantially add to understandings of the earliest period of Indian Buddhism. Archaeology has more to offer in the study of Buddhism in the third century bce. But even here, some

was directly analogous to the colonial and postcolonial contexts of nineteenth- and twentieth-century Buddhism in Thailand and Sri Lanka. For example, Tambiah’s (1984) study of modern Thai Buddhism focuses on the manner in which amulets (blessed by forest monks) are circulated among, and valued by, the urban Buddhist laity. I have no evidence that an analogous process was, or was not, occurring in early Buddhism in India.

archaeological and epigraphic remains provide clear insights, while oth- ers are more ambiguous. Ashokan inscriptions, for example, indicate that Buddhist monasteries existed in the third century bce. As to what these monasteries would have looked like, that can only be stated more tentatively through reference to the first-century bce rock-cut chaitya at Kondivte and the third-century bce sites of Bairat, Lomas Rishi, and Sudama. Based on these sites, it is likely that in the third century bce Buddhists circumambulated stupas in circular wood and thatch huts, with collective rites performed in wooden halls abutting the chamber contain- ing the stupa. This, in turn, suggests that the need to balance the ascetic and communal disjunctures in Buddhism date to at least this time.

Based on the evidence for a tree-shrine at Lumbini, it is likely that Buddhist pilgrimage was practiced in pre-Mauryan times, and was com- mon by the third century bce. Ashokan inscriptions make reference to Buddhist pilgrimage and Ashokan inscriptions and Mauryan period col- umns are found at several sites known to be pilgrimage centers in the second century bce. As to the form and layout of these early pilgrimage sites took, that is less certain. Relying on the Mahaparinibbana-sutta, the poorly dated stupa at Vaisali, and the form of second-century bce pilgrim- age sites, it is likely that they were centered on brick, clay, or earthen stu- pas containing relics of the Buddha. Relying on the recent excavations at Lumbini, earlier pilgrimage sites were likely made of wood, and may have centered on trees as well as stupas.

The practice of circumambulation and collective rites by early Buddhists suggest that the long-standing, fundamental disjuncture between the sol- itary and communal elements of Buddhism may have already existed in the third century bce. More so, considering the priority given to circumambu- lation at Kondivte, Bairat, and the ancestral chaityas on which they were modeled, the establishment of a coherent community of Buddhists may have been limited by the individualistic tendencies of early Buddhists. The creation of spaces within early monastic chaityas for collective rites do, however, suggest that early Buddhists were attempting to foster a more collective Buddhist identity.

While some of the Sri Lankan accounts of Ashoka are confirmed by archaeological and epigraphic remains, a close reading of the inscriptions and their placement at non-Buddhist sites challenges the traditional depic- tion of Ashoka as being wholeheartedly Buddhist. Rather, it appears that while a Buddhist himself, Ashoka gave to multiple religious orders and promoted a more generic, nonsectarian form of Dharma in the majority of his inscriptions. Through a combination of material acts (e.g., construc- tion of columns at Buddhist pilgrimage sites) and immaterial acts (e.g.,

organizing the third Buddhist conference), Ashoka attempted to legiti- mize his rule by routinizing the charisma of the Buddha. At least in Sri Lanka, where Ashoka is remembered as the prototypical divine king, this attempt at legitimization was wildly successful. In India, in contrast, the Mauryan Empire survived for only about 50 years after Ashoka’s death. This is not to say that Ashoka’s legitimations failed in India, only that they were not as successful as they were elsewhere.

While textual sources such as the Pali Canon provide more detailed accounts of the early history of Buddhism, we must be careful not to con- fuse detail with accuracy. Other than Ashokan inscriptions from the third century bce, the earliest textual sources describing early Buddhism were written five to ten centuries after the Buddha’s death. As for the archaeo- logical evidence, it too is problematic. The few material remains of early Buddhism are often difficult to date and interpret. It would be a mistake to claim more knowledge of early Buddhism than either archaeological or historical sources can provide. It is also a mistake, however, to say that archaeologists and historians know nothing of the early centuries of Buddhism. Archaeologists and historians have successfully identified important historical trends and social disjunctures in early Buddhism, trends and disjunctures critical for understanding the long-term devel- opment of Indian Buddhism in subsequent centuries. Rather than dwell on the limitations, it is best to note what we can and cannot know of Buddhism from the sixth through third centuries bce and proceed to studies of later centuries. As stated at the beginning of this chapter, a robust archaeological history of Indian Buddhism can only begin in the second century bce—the first century in which archaeological and epi- graphic remains become ubiquitous.




The Sangha and the Laity

c. 200 bce–200 ce


rior to the second century bce, existing material and historical evi- dence only allows for general or speculative understandings of

the practice of Indian Buddhism. Starting in the second century bce, however, the archaeological, epigraphic, and iconographic remains of Buddhism begin to become more ubiquitous. In terms of the archaeologi- cal remains, there are two general types of Buddhist institutions known from this period. The first are large stupa complexes intended for pilgrim- age by both the sangha and the laity. These include the second-century bce pilgrimage centers at Sanchi and Barhut in central India and later centers found throughout India. Beginning in the first century bce, a sec- ond type of Buddhist institution—monasteries—became common. The best-preserved early monasteries are found in the cliffs of the Western Ghats in peninsular India, with other concentrations of early monasteries found in Andhra Pradesh, Orissa, and the Northwest.

The bulk of this chapter compares the architectural, iconographic, and epigraphic elements of pilgrimage centers and monasteries to show that between the first century bce and the second century ce the ritual practices of the sangha and laity began to diverge. In their monasteries, the sangha began creating ritual spaces intended to establish themselves as the legiti- mate heirs to the Buddha’s charisma. As such, the sangha substantially reduced their ability to live in isolation, focusing instead on establish- ing their authority over the Buddhist laity. These attempts at legitima- tion also significantly affected the disjuncture between individualism

( 104 )

and communalism among the sangha. By emphasizing their authority over the laity, the sangha lost, at least to some degree, their ability to live a solitary ascetic lifestyle. Rather, life in Buddhist monasteries became increasingly communal. In contrast, the layout of pilgrimage centers in this period shows that the laity never fully acquiesced to the legitimations of the sangha—choosing instead to conduct their ritual observances in ways that emphasized their personal relationship with the Buddha. In the disjuncture between individualism and communalism, the laity contin- ued to favor individualism.

One of the difficulties in discussing Buddhism in the second century bce through the second century ce is distinguishing those practices that developed at this time from those practices that merely become visible in the archaeological record for the first time. In the previous chapters I discussed the tension between the individual and the group in early Buddhism. Given the paucity of evidence concerning Buddhism prior to the second century bce, my identification of this tension at that time was somewhat tentative. Beginning in the second century bce, however, the material evidence for this tension dramatically increases. Whether or not this tension existed prior to this, the material evidence shows that, begin- ning in the second century bce, Buddhists of all sorts sought to exploit, overcome, and ignore the persistent social disjuncture between individu- alism and communalism.

Contrary to the traditional histories that claim the members of the sangha were initially isolated ascetics who were progressively cor- rupted through their interactions with the laity, the material remains of early Buddhism indicate something else—that asceticism was a later, first-millennium ce ideal that the sangha projected onto their own his- tory. That is, while asceticism may well have been practiced in the first few centuries of Buddhism, perhaps by the third century bce and cer- tainly by the first and second centuries bce, the sangha had abandoned the wandering ascetic lifestyle in favor of sedentary life in monasteries and a high degree of interaction with the Buddhist laity. Rather than an ascetic sangha being contaminated by the communal practices of the laity, in the second century bce through the second century ce, it was the laity who preserved the long-standing emphasis on individualism over monas- tic attempts to promote a new communal Buddhist identity centered on the sangha.

In this chapter I examine the ways that the sangha attempted to routin-

ize the charisma of the Buddha and establish themselves as the leaders of the larger Buddhist community. These attempts are shown in the creation of new architectural spaces at monasteries, and through physical and

symbolic manipulations of stupas placed within these new architectural spaces. At the same time, I examine the ways the Buddhist laity resisted the sangha’s legitimations by preserving and maintaining their ritual pref- erences. The differences between lay Buddhism and monastic Buddhism are shown through the differences in the design and layout of Buddhist pilgrimage centers and monasteries. To a large degree, the differences between lay and monastic Buddhism are shown in the diverging ways that each sought to accommodate the disjuncture between individualism and communalism. Before examining the archaeological evidence, however, I first provide a general background on the history, epigraphy, and archae- ology of the period.


It is difficult to succinctly summarize the historical context of Buddhism between 200 bce and 200 ce. This difficulty is not simply the result of the 400-year sweep of the analysis. Rather, this difficulty is the product of the dynamic and fractious nature of the political, economic, and reli- gious lives of South Asians throughout the period. During this time, the fortunes of different kingdoms waxed and waned. New sources of wealth and power emerged as guilds and merchants assumed greater promi- nence. Within this dynamic backdrop, religious ascetics—Buddhists included—competed for financial support.

Following the collapse of the Mauryan Empire in 185 bce, numer- ous successor states emerged across India. None of these states ever achieved the same territorial control or hegemonic power as the Mauryas. Rather, these states cycled through periods of greater and lesser power, greater and lesser centralization. In the Gangetic Plain, the Mauryan Empire was followed by a succession of states, including the Shungas (185–73 bce) and Kushanas (first–third centuries ce). In South India, the Satavahanas controlled large areas between the sec- ond century bce and the second century ce, though the fortunes of the empire were highly variable (Sinopoli 2001). In northeastern India, the Kalinga re-established themselves as an independent state after the col- lapse of the Mauryas. As illustrated in a long inscription by the Kalinga King Kharavela (c. second century bce), later kings throughout India presented themselves in roughly the same way that Ashoka had—listing their conquests, good deeds, and professing support for all faiths within their territories (Strong 1983, 1994).










Bodh Gaya

Sudama, Lomas Rishi




Kanheri Kondivte










Thotlakonda, Bavikonda


Arabian Sea

Bay of Bengal

Figure 4.1: Archaeological sites discussed in Chapter 4

Through much of Indian history, groups from outside India heavily influenced the Northwest. In the first and second centuries bce, a series of Indo-Greek kingdoms were dominant in the Northwest. Beginning in the first century bce, the region increasingly came under the control of Central Asian states, including the Sakas (first century bce–first cen- tury ce) and later the Kushanas (first–third centuries ce). Throughout

this period, the Northwest was a crossroads for the trade of goods and ideas between China, Central Asia, India, and the Mediterranean. At the same time, ports along the coast of peninsular India were heavily engaged in trade that linked Africa, Egypt, the Persian Gulf, India, and, by the first century ce, Southeast Asia. Goods transported by ocean and overland routes included gold, silk, aromatic woods, resins, spices, wine, olive oil, beads, and many others. With the trade came the movement of ideas—religious, economic, artistic, and political.

Numerous historians have suggested that new religious orders, par- ticularly Jainism and Buddhism, allowed merchants and traders to more easily interact with foreigners and each other in ways that Brahmanical caste prohibitions had formerly prevented (Singh 2009; Thapar 2002). For this reason, Buddhism and Jainism flourished in the cities where new trade relationships were being forged. With the increase in international and local trade came new sources of power. Between 200 bce and 200 ce, associations of craftspeople, merchants, and traders began to emerge as rivals to the traditional political elite. In addition to setting prices, train- ing apprentices, and establishing production standards, guilds increas- ingly acted as moneylenders and landholders. With their rise in power, guilds began to engage in activities that were formerly restricted to royal families—they began making large donations to religious institutions.

Thapar (2002) and Ray (1986) argued that, by the first century bce, much of the support for Buddhist institutions (monasteries and pilgrim- age centers) was derived from craft guilds rather than the royalty. Thapar interpreted these gifts as evidence that Buddhist institutions were the recipients of donations resulting from competitive giving between guilds, lesser elite, and the royalty. In Weber’s terms, these competitive dona- tions were legitimations performed by the economic and political elite in an attempt to associate themselves with the divine. In contrast, Ray saw Buddhist institutions as more actively engaged in economic activi- ties, serving as nodes on long-distance trade networks (see also Lahiri 1992; but see Morrison 1995) and managing agricultural production in the peripheries of developing states (see also Heitzman 1997). Schopen (2004) has argued that monasteries even acted as moneylenders, build- ing their endowments by lending to Buddhist and non-Buddhists alike. By this view, gifts to religious orders were not solely intended as legiti- mations, but rather as investments with tangible rewards. Whatever the case, it seems that merchants and guilds adopted the same “noble” prac- tices that had been promulgated by Ashoka and that defined the proper behavior of a king. At the same time, monasteries took on some of the eco- nomic roles of guilds and landholders. As a result, between 200 bce and

200 ce the sources of financial support for Buddhist institutions greatly expanded. Where previously almost all the support came from royal fami- lies, by the first century bce the sangha could rely on the support of royal families, guilds, prominent merchants, and landholders, not to mention their own economic activities.

Between 200 bce and 200 ce, Buddhism continued to expand beyond India. In Sri Lanka, Buddhism became well established and supported by the rulers. At the same time, the sangha, following the Silk Road, brought Buddhism to Central Asia and China. With the development of sea trade between India and China in the first century ce, Buddhism also began to be established in Southeast Asia. With the geographic spread of Buddhism across Asia came a gradual fragmentation of Buddhist orders, both out- side and within India. This fragmentation is best illustrated by the fourth Buddhist council, or more accurately, councils. The “first” fourth Buddhist council is traditionally dated to the first century bce in Sri Lanka. This council is credited with producing the first written form of the Pali Canon of the Theravada tradition. The “second” fourth Buddhist council was held in Kashmir in the first century ce. Similar to the Sri Lankan fourth Buddhist council, the Kashmiri fourth council was primarily concerned with recording, collating, and systematizing Buddhist commentaries. This council appears to have been primarily the product of the Sarvastivadan school of Buddhism, though these texts were later foundational in the emergence of Mahayana Buddhism (see Chapter 5). The Sarvastividans were one of many Buddhist schools in India during this period, all with overlapping and distinct traditions and doctrinal interpretations.1


While I generally argue that Buddhist texts are just as useful as archaeo- logical and epigraphic evidence, when investigating Buddhism in the first few centuries bce, textual sources are highly problematic. As discussed at length in Chapter 1, early Buddhist texts, while referring to events that allegedly occurred during the life of the Buddha and the centuries after his death, are best understood as describing the attitudes of a small

  1. Many early Buddhist sources claim that there were eighteen distinct schools of Buddhism in India during this period, though the different sources name varying schools. For this reason, most modern scholars consider the number “eighteen” more as a convention rather than an actual count of Buddhist schools. Taken in a more general sense, these accounts should only be interpreted to mean that there were many schools of Buddhism, with differences large and small dividing them.

    number of elite monks and nuns in the early and mid-first millennia ce. That said, these later textual accounts should not be entirely discounted. While codified centuries later, they do purport to describe earlier periods, and some of the accounts compliment more robust lines of archaeological and epigraphic evidence. In other cases, particularly in regard to textual descriptions of Buddhist asceticism, there is little or no corresponding evidence found in the archaeology or epigraphy of the era. Here I review the existing textual scholarship of early Buddhism with an eye toward those accounts that have the greatest concordance with what is known from archaeological and epigraphic sources.

    Given the centrality of stupas in almost all Buddhist institutions in India between 200 bce and 200 ce, the textual accounts of mortuary ritual have particular importance for understanding early Buddhist his- tory. Buddhist textual accounts of mortuary ritual vary greatly in detail, depending upon the status of the person being memorialized. The most elaborate descriptions, found in the sutras, center on the treatment of the Buddha. There is progressively less detail concerning the burial of promi- nent monks, ordinary monks, and the laity, and those accounts that do exist are almost exclusively found in the vinayas. However, through- out the varying textual accounts is a consistent ambivalence concern- ing mortuary activities. This ambivalence is rooted in the emphasis on asceticism and corresponding dismissal of the mundane world (samsara) that pervades Buddhist literature from the beginning of the first millen- nium ce. This ambivalence is, perhaps, most clearly illustrated in textual accounts of the Buddha’s death and eventual veneration. In one account, preserved within the Mahaparinibbana-sutta (Davids and Davids [1910] 2007) of Sri Lanka, the Buddha is asked by one of his principal disciples, Ananda, what should be done with the Buddha’s body after death. The Buddha responded,

    Hinder not yourselves, Ananda, by honoring the remains of the Tathagata [Buddha]. Be zealous, I beseech you, Ananda, in your own behalf! Devote yourselves to your own good! Be earnest, be zealous, be intent on your own good! There are wise men, Ananda, among the nobles, among the Brahmins, among the heads of houses, who are firm believers in the Tathagata, and they will do due honor to the remains of the Tathagata. (Davids and Davids [1910] 2007:154)

    Subsequent passages state that the nobles should cremate the Buddha’s body and place the ashes within a stupa where those “who will take a gar- land or perfume or paint there, or will salute, or will cause their mind to

    be tranquil, that will be for their benefit and ease for a long time” (Davids and Davids [1910] 2007:156).

    Traditionally, this account has been interpreted as contrasting the proper behavior of the Buddhist sangha and laity toward the Buddha’s remains (Coomaraswamy 1927; Lamotte 1988; Oldenberg 1882; but see Schopen 1997:99–113 for an alternative interpretation). The sangha, with their greater knowledge and sophistication, should abstain from venerat- ing the Buddha’s relics, but instead should focus on their own personal achievement of nirvana through religious instruction and meditation. The laity, with their lesser understanding of Buddhism, could obtain merit through ritual directed toward the Buddha’s relics interred within stupas. It is clear that by at least the second century bce this traditional under- standing of the sangha is wrong. As will be discussed below, in the second century bce through the second century ce, archaeological and epigraphic evidence unequivocally shows that both the sangha and the laity were heavily invested in veneration of stupas, though in different ways and for different ends.

    Buddhist textual sources provide far less detail on the mortuary treatment of Buddhist monks. It appears that some of the most promi- nent monks were given similar treatment to the Buddha. Their remains were interred within stupas. Though the stupas of prominent monks were often smaller than stupas of the Buddha (mahastupas) and often omit- ted circumambulatory paths, they were still large structures, similar in most respects to mahastupas. The same cannot be said of the treat- ment of ordinary monks—those monks who did not warrant post-death veneration.

    Textual descriptions of the funerals of ordinary monks do not occur in the same accounts as those of the Buddha or prominent monks. Rather, these funerals are typically addressed in the vinayas, texts that provide rules and procedures for monastic life. One account in the Mulasarvastivada-vinaya (Schopen 1997:217) reports that after the death of a monk the sangha dumped his body in a ditch outside the monastery. When the deceased monk’s relatives found the body, they complained to the Buddha. In response, the Buddha decreed that the monk should receive a proper cremation. In another account within the Mulasarvastivada-vinaya (Schopen 1997:210–214), after the death of a monk the members of the monastery went to his cell to redistribute his belongings. In the cell they found the dead monk’s ghost, who refused to part with his belongings until he received a proper funeral. This funeral was then described as including the removal of the body to the cremation ground, the washing of the body, the recitation of Buddhist scripture over the body, and the

    final cremation of the body. All participants were then required to ritually wash themselves and their clothes before returning to the monastery.

    While this passage does provide a detailed account of a monastic funeral, it is important to note that this description is not the central point of the account. Rather, the account concludes by stating that only those who par- ticipate in a proper funeral are entitled to a share of the deceased monk’s belongings. By this account, then, the impetus for a proper funeral is not religious, per se, but rather pecuniary (Schopen 1997:204–237). As in the description of the Buddha’s funeral, there appears to be a general disin- terest in the affairs of the body. Based on these accounts, it appears that the Buddhist sangha only held funerals because the communities in which they lived demanded them, and as a means to properly identify those who had a right to inheritance. As in the account of the Buddha’s mortuary treatment, it appears that monastic interest in mortuary practices had little to do with religious concerns, but rather with mundane practical concerns.

    If accounts of monastic funerals are limited in Buddhist textual sources, discussions of lay funerals are almost nonexistent. In some Sri Lankan and Chinese vinayas (Schopen 1995:105–106), participation in lay funer- als is listed as one of the few reasons that a member of the sangha may leave a monastery during the normally restrictive rainy season retreat. Otherwise, Buddhist texts are silent about lay funerals. The Buddhist sangha participated in lay funerals, but no details are provided of the funerals themselves.

    Overall, the picture of Buddhist mortuary behavior that can be gleaned from texts is one of ambivalence. The deceased were given funerals grudg- ingly or, in the case of the Buddha, given to the laity for funeral rights and subsequent veneration. Funerals were a necessary chore, a distraction from the real focus of their actions, meditation, and learning. Funerals were more a social requirement than a religious obligation. If it were not for the additional information provided by Buddhist inscriptions and archaeological studies, this would be the end of it. However, archaeologi- cal and epigraphic sources provide a complementary view of the nature of Buddhist mortuary behavior.


    In contrast to more traditional textual histories of Buddhism, in the last few decades many Buddhist scholars have begun to focus on epig- raphy (e.g., Schopen 1997, 2004; Trainor 1997). As argued by Schopen

    (1997:30), inscriptions have at least two major advantages over the analysis of Buddhist texts—they are earlier than existing texts and they were written at the behest of ordinary Buddhists, not the monastic elite. These advantages of the inscriptional material are balanced by one major disadvantage—most of these early inscriptions were short—briefly recording donations to monasteries or pilgrimage complexes. Most often, inscriptions provided little more than the name of the donor and, per- haps, their hometown and occupation. Despite this, much can be learned of early Buddhism from donation inscriptions.

    One of the more startling revelations about early Buddhism derived from an analysis of donation inscriptions is that substantial proportions record the donations of Buddhist monks and nuns. For example, the sangha make up almost a third of the donation inscriptions carved onto the rail- ings surrounding the second-century bce pilgrimage stupas at Sanchi and Bharhut (Fogelin 2003; Luders 1912; Schopen 1997). Despite the prohi- bition against the accumulation of personal wealth attested to in some textual traditions, donation inscriptions unequivocally demonstrate that the sangha had access to wealth, and at times this wealth was significant (Schopen 1997). This disjuncture with texts is interesting on two fronts. First, it helps to explain the importance of monastic funeral practices dis- cussed earlier in regard to the inheritance of a deceased monk’s property. Simply put, participation in a monastic funeral could be very lucrative. Second, many of the donation inscriptions listing the sangha are found in the large pilgrimage stupas rather than in the monasteries. This, in turn, suggests that the sangha was not as isolated from the laity as the textual accounts would suggest and that they were actively engaged with what Schopen (1997) calls the “stupa cult” at large pilgrimage complexes.


    An examination of the layout and design of the earliest pilgrimage com- plexes informs on the practice of popular Buddhism by the sangha and par- ticularly by the laity. With the possible exception of Vaisali and Lumbini, the earliest archaeologically known pilgrimage complexes in India date to the second or third centuries bce. Again, with the exception of Vaisali and Lumbini, none of the early pilgrimage complexes is found in the Gangetic Plain, the heartland of Buddhism (see Figure 4.1). While inscriptions, Mauryan columns, and other evidence demonstrate that pilgrimage com- plexes were erected in the Gangetic Plain, later building episodes have completely obliterated the material traces dating from 200 bce to 200 ce.

    What is known of early pilgrimage, therefore, must be learned from a small number of early complexes constructed in central, northwestern, and southern India. All of these complexes are centered on massive stu- pas. The largest and most prominent of these early pilgrimage complexes are Sanchi and Bharhut (Sanchi: Cunningham [1854] 1997; Luders 1912; Marshall and Foucher 1983; Mitra 1971; Shaw 1999, 2000, 2007, 2013b;

    Shaw and Sutcliffe 2001; Bharhut: Barua 1979; Cunningham [1876] 1962;

    Hawkes 2008, 2009; Luders 1912).

    The earliest stupas at Sanchi and Bharhut were constructed in the third century bce. However, both were enlarged and modified in the second centuries bce. Bharhut is very poorly preserved. Before the first archaeo- logical excavations and research, the stupa had been extensively mined for its bricks (Barua 1979; Cunningham [1876] 1962; Hawkes 2008, 2009). While the railings and gateways have been found and reconstructed, the anda is now almost completely missing. The importance of Bharhut lies in the large number of inscriptions and reliefs found on its railings, now located in the Indian Museum on Kolkata. Three distinct stupas have been found at Sanchi, each reconstructed several times. While some excava- tions occurred in the early nineteenth century at Sanchi, Cunningham’s ([1854] 1966) excavations from the mid-nineteenth century provided the first systematic study. Later, Sir John Marshall conducted excavations at Sanchi between 1912 and 1919 (Marshall and Foucher 1983).

    More recently, Julia Shaw conducted a large-scale survey project in the area immediately surrounding Sanchi, identifying numerous smaller sites and large irrigation features (2000, 2007, 2013b; Shaw and Sutcliffe 2001). Based on these features, Shaw (2013b:103) argued that Buddhist monasteries were a major part of the process where North Indian social and economic practices spread to South India.

    Lay support of the sangha was essential to the latter’s survival, but practical services provided by the monastery, in this case, water for domestic and agri- cultural use, formed the backbone to changing social and economic conditions during the late centuries bc. These changing conditions included urbanization and agricultural ‘involution.’ The sangha’s close relationship with agricultural improvement and water management was an important instrument of lay patronage, but it was also closely related to Buddhism’s deeper preoccupation with human suffering (dukkha) and the means of its alleviation.

    Shaw also challenged the notion that Buddhist monks were purely ascetic. Rather, Shaw (2013b:103) saw the elaborate and extensive rework- ing of the agricultural landscape surrounding Sanchi as “demonstrating

    that evidence for a ‘domesticated’ and socially integrated form of Buddhist monasticism was already in place in central India by the late centuries bc.” Several other large pilgrimage stupas are known from the first century bce. Among the best known are Amaravati (Barret 1954; Burgess [1882] 1972; Shimada 2012; Ramachandra Rao 2002; Sewell [1880] 1973) in Andhra Pradesh and the Dharmarajika stupa (Marshall [1951] 1975; Sarkar 1966) near modern Islamabad. While some have claimed that earlier stupas at Amaravati may date to Mauryan times based on some fragments of what may be a Mauryan column (Shimada 2012), the known remains likely date to major renovations that occurred in the first century bce. Earlier excavations at Amaravati obliterated any evidence of these earlier stupas, or even the ability to know if earlier stupas were present at all. Today, the archaeological materials from Amaravati are widely scat- tered throughout the world, with major portions of the railing and sculp- tural elements in the British Museum and the Government Museum in


    Between 1913 and 1934, Sir John Marshall conducted large-scale excavations at the ancient city of Taxila, near modern Islamabad in the Northwest. Excavations focused on two sections of the city (Bhir Mound and Sirkap Mound), several peripheral monasteries, and the Dharmarajika stupa. Like Sanchi and Barhut, Dharmarajika was likely initially con- structed in the third century bce, but later constructions continued to modify the general architectural form. This resulted in changes to the stupa and the destruction of the railing (Marshall [1951] 1975). In its pres- ent form, the core of the complex preserves the layout from the first cen- tury bce. The courtyard surrounding the monastery, and the numerous shrines it contains, were constructed over subsequent centuries.

    Marshall’s excavations on the Sirkap mound were massive in scale, revealing approximately 11 hectares of the second-century bce through first-century ce city (Marshall [1951] 1975). Excavations revealed the square bases of several stupas in courtyards within the city, though it is not clear whether these stupas had Buddhist or Jain affiliation. While noting this problem, Mitra (1971:124) argued, “the predominance of Buddhism in this region during the period to which they belonged would favor their Buddhist affiliation.” Most of the stupas on the Sirkap mound seem to be located in the courtyards of wealthy households, though it is possible that these complexes were urban shrines. The largest of these stupas was located in a large courtyard (33.5 m x 29.2 m) in Block A. At the center of the courtyard was a square drum (10 m x 10 m). While the remains of a crystal relic casket and other decorative elements confirm that stupa was erected on the drum, it is too damaged to determine its

    size. In addition to the central stupa, several other smaller votive stupas were found in the courtyard. Finally, the entire courtyard was surrounded by upwards of twenty-five, somewhat haphazardly placed, rooms. As with the other stupas found on the Sirkap mound, this stupa may have been located in the courtyard of a wealthy household, though, given the large size of the stupa and its courtyard, it is perhaps the most likely to have served as an urban shrine.

    More recently, a new stupa has been found and excavated near Sannati in northern Karnataka. The Kanaganahalli stupa was found in 1994, with continued excavations through 2002. The site appears to date between the first century bce and the third century ce. Initial reports suggest that it is the site of a mahastupa, resembling in most respects other contemporary pilgrimage stupas. Among the most important finds is a frieze depicting a king and queen with an inscription reading “King Ashoka” (Poonacha 2007). Critically, since the stupa at Kanaganahalli dates at least two centuries after Ashoka’s death, this frieze indicates that Buddhists—in India—preserved the memory of Ashoka for at least several centuries. Prior to this discovery, Ashoka was only known from the inscriptions carved during his lifetime and in textual accounts from Sri Lanka and China.

    While there is some variation in the form of pilgrimage stupas, they all tend to follow the same general layout (Shimada and Hawkes 2009; see Figure 4.2). All of the stupas at pilgrimage complexes seem to contain relics, often attributed to the Buddha himself. The andas are generally large. The anda of Stupa 1 at Sanchi is over 12 meters high, with a diam- eter of roughly 32 meters. The more heavily damaged andas at Bharhut, Dharmarajika, Amaravati, and Kanaganahalli appear to have been com- parable in size. At all these early sites the andas were placed on raised plat- forms (drums). The role of the drums was fairly straightforward—they elevated the anda. Raising the anda achieved two purposes. First, on the practical side, raising the anda allowed more distant viewers to see it above the shoulders of closer viewers. Second, the drums also raised the perceived status of the anda through physical elevation (Moore 1996). One additional benefit of the drums is that they increased the mass of the stupa, giving it a greater physical presence. Surrounding the drum(s) and anda were circumambulatory paths defined by elaborately decorated rail- ings. Beyond the railings was a large courtyard where larger group ritu- als could be performed. Gates (toranas) were located in the railing at the cardinal points to allow for access between the two ritual spaces. The rail- ings and toranas also served as location for inscriptions—recording the donations of Buddhist monks, nuns, and laity from far-flung locations

    Figure 4.2: Typical early pilgrimage stupa

    (Basham 1967; Brown 1965; Lamotte 1988; Luders 1912). At Amaravati, ayaka platforms and pillars were erected directly behind the toranas. Finally, beginning in the second century bce, stupas were constructed of stone, brick, and stucco rather than earth.

    At the apex of the earliest archaeologically known stupas were stylized parasols (chhatras). In India, kings were often depicted under parasols, with parasols serving as an index of their status. In this sense, the place- ment of a parasol atop a stupa served to index the royal character of the Buddha’s relics interred within. Chhatras also symbolized the Bodhi tree, the location where the Buddha gained enlightenment at Bodh Gaya. In Peirce’s semiotics, signs are rarely purely symbolic. In this case, a parasol

    does have some superficial, iconic resemblance to a tree. Both have a trunk and canopy. Both can also serve to protect a person from the elements. Even so, there are sufficient differences between the two that an observer of one would not immediately think of the other without prompting or convention.

    There is also evidence that in the second century bce Buddhists viewed stupas as symbols of Buddhism more generally. In a semiotic analysis, Dehejia (1997) identified several multivalent signs in the friezes carved on the railings of Sanchi, Bharhut, and Amaravati. In many cases, the depictions of stupas were straightforwardly iconic—depicting devotees actively engaging in ritual around stupas (see Figure 4.5). Other friezes depicted the Buddha through indexical signs. Following the conventions of early Buddhism, artists never depicted the Buddha in human form. Rather, they depicted the Buddha by his “conspicuous absence” (Fowles 2008), or a by an index that served to represent him. For example, in many instances friezes depict crowds of people looking intently at an empty chair (see Figures 1.4 and 5.9). The empty chair, argues Dehejia, served as an index of the presence of the Buddha. In other cases footsteps, para- sols, or trees indexed the Buddha’s presence (Dehejia 1997:41–51). Dehejia argues that in other cases, particularly at Sanchi and Amaravati, depic- tions of some stupas were symbolic, signifying more abstract theological principles and Buddhism in general, rather than serving as icons of actual stupas. It remains unclear whether the symbolic association between stu- pas and Buddhism developed in the second century bce, or if the same association existed from the fifth through third centuries bce. Given the lack of friezes dating to the earlier period, it is impossible to come to any conclusion on this question.

    The large pilgrimage stupas of the second century bce are best under- stood as a multivalent signs, similar in most respects to the ancestral stu- pas discussed in Chapter 3. Though the medium of second century bce pilgrimage stupas was different from earthen ancestral stupas, the form of the anda (low hemisphere) and the contents of the anda (relic) remained the same. At most, the relatively greater grandeur of stone, brick, and stucco indexed a greater prominence or power of the Buddha’s relics interred within the stupas. Chhatras were added, but overall, the sign as a whole still fundamentally resembled the eight ancestral stupas through the resemblance of its distinct attributes. Beginning in the second century bce, stupas were also an overarching symbol of Buddhism. Like the ances- tral stupas on which they were based, pilgrimage stupas containing the Buddha’s corporeal relics were icons of the Buddha encased within indexes of his presence.

    Large outdoor stupa complexes are generally recognized as having served as pilgrimage centers. Several lines of evidence support this. Pilgrimage is recognized in assorted versions of the Mahaparinibbana-sutta (Schopen 1997:115–119), several of which record the Buddha as stating that pil- grimage confers merit upon the pilgrim. In other sections, the Buddha is recorded as stating that those people who die while traveling to or from a pilgrimage center will attain nirvana. In addition to these literary sources, three other lines of evidence support the claim that large out- door stupa complexes were pilgrimage locations. First, many of the out- door stupa complexes are very large. While viharas may have been located in or near pilgrimage complexes, the space available for worship was far greater than needed by the resident clergy.2 Second, no doors block access to either the circumambulatory path or the courtyards surround- ing the central stupa. As will be discussed later, the entrances to the cir- cumambulatory path were carefully designed to allow the free movement of people, while maintaining the separation between different worship areas. Third, donation inscriptions from Sanchi, Bharhut, and Amaravati show that devotees, including monks, nuns, and the laity, traveled great distances to worship at them (Basham 1967; Cunningham [1854] 1966, [1876] 1962).

    Within the courtyards at Sanchi, Dharmarajika, and Block A on the Sirkap mound (Mitra 1971) were numerous haphazardly placed “votive” stupas. These small stupas generally consisted of two parts, a small drum with a separate capping stone in the form of an anda. A small cavity between the two parts held the cremated remains of devotees who had been interred in ways that allowed perpetual worship of the relics interred within the stupa. Schopen (Schopen 1997:114–147) has referred to the placement of votive stupas adjacent to mahastupas as “burial ad sanctos.” Just as proximity to the Buddha during circumambulation was merito- rious, burial adjacent to the Buddha was similarly meritorious. In fact, burial would be one step better. Circumambulation had a definable end point, where burial adjacent to a stupa could potentially allow for the accu- mulation or merit perpetually. Buddhist inscriptions also provide insight into the role of votive stupas. In many cases, short inscriptions recording

  2. The known viharas at early Buddhist pilgrimage sites all date to the mid-first millennium ce. It is possible that these monasteries were built on the foundations of earlier monasteries, but these foundations have not been archaeologically identi- fied. When saying that pilgrimage centers were larger than necessary for the resident monastic population, I am relying on the size of later, first-millennium ce, viharas. That is, earlier viharas were certainly no larger than the later viharas, or perhaps even smaller, or not present at all.

the name and occupation of the deceased are carved into votive stupas. Like the variability of donors shown in inscriptions carved on the railings at Sanchi, Bharhut, and Amaravati, inscriptions on votive stupas named people from many occupations, including Buddhist monks and nuns. Taken together, the evidence from donation inscriptions and votive stupa inscriptions suggest that Buddhist monks and nuns were actively engaged in worship at pilgrimage complexes.

Individual and Communal Ritual in Pilgrimage Complexes

In Chapter 3, I discussed the creation of distinct ritual spaces within Buddhist chaityas that allowed the simultaneous performance of individ- ual and communal ritual. At sites like Kondivte, Bairat, and the wooden prototypes on which both were based, early Buddhists separated the prac- tice of individual and communal ritual in an attempt to ameliorate the tensions between the two. I also argued that the layout of these early cha- ityas privileged the performance of circumambulation over group rituals. While a hall was built in front of the chamber containing the stupa, from the point of view of people engaged in communal ritual in the hall, the stupa was only partially visible through a narrow door (see Figure 3.5). Even this limited view would have been periodically interrupted by the passage of people engaged in circumambulation within the chamber. For these reasons I argued that the design of early Buddhist chaityas only partially ameliorated the tension between the individual and communal desires of the incipient Buddhist community.

Beginning in at least the second century bce, Buddhists began building ritual spaces in ways that more effectively concealed the tensions between the individual and the group. At pilgrimage sites, railings were erected around the circumambulatory paths. These railings effectively separated devotees engaged in circumambulation from others simultaneously engaged in more communal forms of ritual in the surrounding courtyard. The construction of railings at pilgrimage stupas marks an important development in the way that Buddhists ameliorated the tension between the individual and the group at pilgrimage complexes. At all pilgrimage centers where railings are preserved, they are well over the height of a person. While these railings are important as decorative elements in the architectural design, and the gateways provide surfaces on which to carve religious scenes, neither function seems to fully explain the particular forms of the railings or gateways. Circumambulation is an individual rit- ual, but the courtyards at pilgrimage centers were used for group worship.

Figure 4.3: L-shaped gates

It would be distracting for the people in the pathway to become enmeshed with people milling about in the assembly area. Likewise, the sight of people silently walking around the stupa would have been distracting for those engaged in worship in the assembly area. The railings at pilgrimage stupas served to block the view of each ritual area from the other. This served to reduce the tension inherent in the competing ritual demands of the space.

This point can be best illustrated by examining at the layout of the gate- ways that link the circumambulatory path with the assembly area. At both Sanchi and Bharhut, the gates are L-shaped. By constructing them in this form, the potential for looking into or out of the circumambulatory path was reduced, while still allowing for unobstructed entrances and exits to the path (see Figure 4.3). As such, the gates served to promote the sepa- ration of individual and communal ritual within pilgrimage complexes, while fostering movement between them. As will be discussed below, in the monasteries the sangha achieved the same separation between indi- vidual and communal ritual, but in different ways. These differences greatly inform an understanding of the relative importance of individual and communal ritual for the laity and the sangha.

Arenas and Halls

Another difference between monastic and pilgrimage complexes illumi- nates the style of ritual conducted within the different ritual spaces. The

Figure 4.4: Arena format of a pilgrimage center and the hall format of a chaitya

courtyards at pilgrimage complexes completely encircle the stupa, while in monastic complexes stupas are placed at one end of a long hall (see Figure 4.4). In Chapter 2, I compared the more communal participation in ritual fostered by Quaker meetinghouses in contrast to the more directed and mediated form of ritual found in a Catholic church. While both lay- outs promoted social solidarity between participants, Quaker meeting- houses promoted a communal identity and limited the elevation of ritual leaders. Catholic churches, on the other hand, fostered the authority of priests by placing them at the end of a hall, with all the parishioners fac- ing him. The same differences can be seen in pilgrimage centers like Sanchi and Bharhut and monastic chaityas like Kondivte, Bairat, and the wooden chaityas in which they were modeled.

The large courtyards surrounding the anda at pilgrimage complexes allowed worshipers to engage and participate with other worshipers while maintaining a focus on the stupa, which rose above the railings defining the circumambulatory path. The courtyard areas at pilgrimage complexes would have allowed for interaction, celebration, and worship. Given this, Sastri’s (1966) claim that early Buddhist ritual had an important ecstatic component is also possible. This claim is reinforced by evidence from some reliefs at Bharhut and Sanchi depicting more ecstatic forms of ritual. For example, one frieze from the railing at Sanchi depicts devotees individu- ally engaged in circumambulation within the paths, while more festive

Figure 4.5: Sanchi frieze (c. first or second century bce; after Cunningham 1854)

forms of ritual—including music and dancing—were conducted in the courtyard surrounding the stupa (see Figure 4.5). Given this level of activ- ity in the assembly area, the need for the railings to define a segregated circumambulatory path becomes more critical.

Over the next millennia, the early stupas at Sanchi, Bharhut, Dharmarajika, and Amaravati continued to be used as pilgrimage sites. Other pilgrimage stupas, of roughly similar design and signification, continued to be constructed throughout India. To this day, stupas in Sri Lanka and Nepal follow the same general layout of Indian stupas from the second century bce. As such, the large pilgrimage stupas frequented by the Buddhist laity can be viewed as a particularly robust and long-lasting

assemblage of signs. However, pilgrimage stupas were not the only type of stupa constructed in the first few centuries bce. Across India, the sangha also constructed stupas in the primary worship halls (chaityas) of Buddhist monasteries. An examination of Buddhist monasteries and chaityas dem- onstrates that the ritual practices of the Buddhist sangha began diverg- ing from the ritual practices of laity beginning in the first century bce. These differences centered on the different ways that the sangha sought to balance their individual and communal desires.


The archaeological remains of Buddhist monasteries become common in the first century bce. Like pilgrimage stupas, however, little is known about Buddhist monasteries in the heartland of Buddhism in the Gangetic Plain. Based on numerous inscriptions and stray finds at later monastic complexes, it is clear that monasteries were present in the Gangetic Plain between the second century bce and the second century ce. While coins, pottery, and other archaeological remains have been found, early monas- teries in the Gangetic Plain were wiped away by later remodeling. From an archaeological perspective, we are left knowing that monasteries were present, but ignorant of their form, organization, and style. For the most part, the earliest archaeologically known monasteries are found in three clusters outside the Gangetic Plain: in the Western Ghats of peninsular India, in coastal Andhra Pradesh and Orissa, and in the Northwest. In the Western Ghats, monasteries were carved into cliff faces in much the same way as Kondivte, Lomas Rishi, and Sudama had been in previous cen- turies (see Chapter 3). In Andhra Pradesh and Orissa, monasteries were freestanding and rock-cut, with some sites using both techniques simul- taneously (e.g., Sankaram, Salihundam). In the Northwest, monasteries were freestanding. These differences have much to do with archaeological preservation and the availability of carvable cliff faces, but also suggest regional differences in how the sangha sought to engage with the laity.

The largest concentrations of Buddhist monasteries dating from 200

bce through 200 ce are in the Western Ghats and the east coast of penin- sular India. Dozens of monasteries were carved in the Western Ghats in this period. These monasteries range in size from a few isolated monas- tic cells to major complexes with multiple viharas and chaityas. Given the excellent preservation of the monasteries in the Western Ghats, they have been the focus of archaeological research for more than 150 years (Fergusson and Burgess [1880] 1988; Dehejia 1972; Nagaraju 1981). Key

Figure 4.6: Bhaja (c. first century bce; Fergusson and Burgess 1880

sites include Bhaja, Bedsa, Karla, Ajanta, Kondane, Kanheri, and Junnar. Most of the monasteries in the Western Ghats are found in mountain passes that served as trade corridors between the west coast and the inte- rior of the Peninsula. For this reason, some scholars (Lahiri 1992; Ray 1986) have suggested that the monasteries may have served as way sta- tions for long-distance trade. In contrast, Morrison (1995) has noted that the monasteries in the Western Ghats lacked storerooms and space for itinerant merchants. Rather than way stations, Morrison argues that monasteries were located to benefit from the material support of multiple communities on either side of the passes.

Figure 4.7: Thotlakonda (c. second century bce–third century ce)

On the opposite side of the peninsula, in coastal Orissa and Andhra Pradesh, was another concentration of archaeologically known Buddhist monasteries. These monasteries were also located along major trade routes—in this case, coastal and river trade routes that extended down the east coast of the peninsula. Monastic engagement with coastal trade is at least partially shown by small numbers of Roman coins and other highly valued trade goods present at numerous monasteries (e.g., Thotlakonda, Salihundam, Bavikonda). It is unclear, again, whether the monasteries served as way stations or simply as the recipients of high value goods through donation. The relatively small quantity of prestige goods might suggest the latter. Whatever the case, these earliest monasteries in Andhra Pradesh and Orissa were located on low hills along the coast (e.g., Thotlakonda, Bavikonda, Salihundam) with slightly later (c. first century bce) monasteries constructed tens of kilometers inland, usually in agriculturally productive areas near navigable rivers (e.g., Sankaram, Ramatirthan, Lalitagiri).

The last concentration of early Buddhist monasteries is located in the Northwest. This region has the fewest archaeologically known early mon- asteries, and those that are known are poorly preserved and understood. The best preserved is the site of Takht-I-Bahi (Brown 1965; Mitra 1971; see Figure 4.8). Located on a ridge top, Takht-I-Bahi was initially constructed in the first century bce or first century ce, with substantial expansion and renovations through the seventh century ce. The portions dating to the first century ce include a square vihara with fifteen monastic cells, a courtyard with numerous votive stupas erected on square drums and small shrines. Later, another vihara was added with ten additional cells. Several other contemporary monasteries have been found near the ancient city

Figure 4.8: Takht-I-Bahi Monastery, main vihara on left and stupa court on right (c. first century bce–seventh century ce; after Brown 1965)

of Taxila (e.g., Kalawan, Mohra-Moradu, Jaulian, and Pippala; Marshall [1951] 1975; Mitra 1971). Renovations at all of these monasteries make it difficult to determine their layout in the first century ce, but it appears that they follow the same general plan as Takht-I-Bahi. With the possible exception of Kalawan monastery (2 km southeast of the Dharmarajika pilgrimage complex), it seems that monasteries in the Northwest omit- ted chaityas in favor of courtyards containing small open-air stupas and votive stupas. This does not mean that chaityas were not constructed in the Northwest.

In Block D of the Sirkap mound at Taxila, Marshall ([1951] 1975: 150–155) excavated a large (39.3 m long x 15.5 m wide) apsidal chaitya (see Figure 4.9). Marshall dated the chaitya to between the second century bce and the first century ce, though most modern scholars favor the latter portion of that range (Mitra 1971). At the time the chaitya was first exca- vated by Cunningham, the central chamber no longer contained a stupa (Marshall [1951] 1975:151). It is possible that the central chamber never contained a stupa, thus resembling the layout of the rock-cut caves at

Figure 4.9: Apsidal chaitya in Block D at Taxila (c. first century bce or first century ce; after Marshall 1951)

Lomas Rishi and Sudama. That said, most scholars believe that the chaitya in Block D was affiliated with Buddhism and that the stupa in its central chamber had been destroyed in antiquity. There is also another critical difference between this chaitya and contemporary monastic chaityas dis- cussed thus far. The chaitya in Block D was located in a large courtyard within the city of Taxila. On the western end of the courtyard, flanking the entrance to the complex, were eight rooms. Marshall interpreted these as either monastic cells that opened onto the temple courtyard or shops that opened onto the street (Marshall [1951] 1975:150). It is also possible that these rooms were used for temple storage or other mundane pur- poses. Even if these rooms served as sleeping quarters for a few resident monks or nuns, the small number suggests this complex was not a mon- astery. Rather, the chaitya in Block D was most likely an urban shrine for lay worship.

With the possible exception of the temple in Block D at Taxila, Buddhist monasteries in this period were somewhat isolated from, but usually near, larger population centers. In Andhra Pradesh and Orissa, for instance, monasteries were typically located on otherwise uninhabited hilltops. In the Western Ghats, monasteries were carved into the cliffs in moun- tain passes. In the Northwest, monasteries were constructed in valleys and ridge tops, though usually within ten kilometers of nearby cities and towns. In some sense, this placement fits well with the idea that the early sangha sought to live in isolation from lay populations, while remaining close enough to benefit from the support that cities could provide. For earlier generations of textual scholars, this placement was explained by the textual proscription that the sangha must leave their monasteries each day to beg for their food from the laity. At least in the case of Andhra Pradesh, Orissa, and the Northwest, this view is wrong.

On the east coast of the peninsula and in the Northwest, food was prepared and served to the sangha in specially constructed refectories attached to the monasteries (Fogelin 2006:163; Prasad 1994; Sastry et al. 1992). While monasteries in the Western Ghats do not have clearly identifiable refectories, open areas in the viharas may well have served the same purpose. Donation inscriptions from monasteries throughout India testify to the donation of significant wealth to the sangha. When combined with monastic participation in trade and money lending, it seems unlikely that in the first century bce the members of the sangha were begging for their food on a daily basis. Given this, the placement of monasteries in peripheral locations to cities cannot be explained by the textual proscriptions concerning begging. Rather, the placement of mon- asteries is best explained as a means to balance the sangha’s desire for

isolation and asceticism with their desire to create and lead a community of lay Buddhists. Monasteries were placed close enough to cities to allow the sangha to lead ceremonies and benefit from the donations of the laity, but far enough away to foster the perception of liminality, isolation, and asceticism that served as the foundation of the sangha’s spiritual power. This same balance can be seen in the architectural layout and design of the monasteries themselves. Between the first century bce and the sec- ond century ce, Buddhist monasteries were designed to foster monastic isolation while simultaneously drawing Buddhist laity to monastically led rituals.

In India between 100 bce and 200 ce almost all Buddhist monaster- ies were divided into two or three spatially distinct areas—viharas with restricted access, public worship areas consisting of chaityas and open-air stupas, and in Andhra Pradesh, Orissa and the Northwest, refectories. Viharas were typically square, with private assembly areas at their center (see Figures 4.6). The public worship areas, in contrast, show significant architectural variability across India. In the Western Ghats, worship areas consisted almost exclusively of rock-cut chaityas. On the east coast of the peninsula, public worship occurred in both freestanding and rock-cut chaityas and at open-air stupas. In the Northwest, there seems a greater emphasis on open-air stupas at monasteries, with chaityas associated with lay, rather than monastic, ritual (discussed below). At least some of the differences in the layout of monasteries were the result of how they were built. It is simply impossible to create an open-air stupa complex when carving a monastery into a cliff face. The omission of open-air stupas at monasteries in the Western Ghats, then, is not surprising. Other archi- tectural features, however, indicate regional variation in how the early Buddhist sangha accommodated and ameliorated their desire for isolation and asceticism with their need to forge and lead a Buddhist lay commu- nity. These tensions are shown in the architectural techniques used to sep- arate circumambulation and group ritual, and the layout of the assembly areas in monastic chaityas.

Spatial Separation in Buddhist Monasteries

Beginning in the first century bce, the sangha employed varying archi- tectural strategies to separate individual and group ritual in Buddhist monasteries (Fogelin 2011). In the Western Ghats, the sangha separated the two forms of worship through a new style of chaitya. This new style was first executed at Bhaja c. 90–80 bce, though like Kondivte, it appears

Figure 4.10: Bhaja main chaitya (c. first century bce; plan after Fergusson and Burgess 1880)

to be based on earlier wooden prototypes (Brown 1965; Dehejia 1972; Fergusson and Burgess [1880] 1988). Rather than place the stupa within a small room in which circumambulation could be performed, at Bhaja the stupa was placed at the end of an apsidal hall, with the circumam- bulatory path moved to the periphery of the structure (see Figure 4.10). The circumambulatory path was separated from the group worship area by a row of columns that served to visually separate the path from the central hall. The placement of the columns assured that neither those in the hall facing the stupa, nor those engaged in circumambulation, would see each other (Fogelin 2003). Meanwhile, those engaged in group ritual in the hall gained an unobstructed view of the central stupa. In short, the design of Bhaja allowed those engaged in individual and collective group ritual to ignore each other more effectively. The success of this layout can be shown in its ubiquity in the Western Ghats over the following centu- ries, as numerous new chaityas were constructed following the same plan (e.g., Karla and Ajanta).

In Andhra Pradesh and Orissa, a more straightforward approach was taken to resolving the tension between individuals and groups. Between

Figure 4.11: Bavikonda (c. second century bce–third century ce; after Prasad 1994)

the first century bce and the second century ce, the sangha had separate buildings erected for individual and group ritual (see Figure 4.11). This pattern is evident at sites including Bavikonda (Prasad 1994), Thotlakonda (Fogelin 2006; Sastry et al. 1992) and Salihundam (Subrahmanyam 1964). For group rituals the sangha created apsidal halls, similar to those of the Western Ghats except for the omission of the columns and circumambula- tory paths. For circumambulation they employed small circular chaityas of stone and brick, similar to earlier circular huts containing stupas in most details other than the building material.

The role of chaityas in the Northwest is different from that of the peninsula. This may be the result of genuine differences in the worship practices of the sangha, or it could be the product of the limited number of archaeologically known monasteries from this period. The key differ- ence in the Northwest seems to be the omission of chaityas at Takht-I- Bahi and, perhaps, the monasteries near Taxila. As discussed earlier, in the Northwest the sangha favored courtyards containing open-air stupas to chaityas. In some sense, these courtyards resembled the large, con- temporary pilgrimage stupas discussed earlier, only smaller. However, these smaller open-air stupas lacked railings to define a circumambula- tory path. Circumambulation was possible on the square drums on which the primary stupas were erected, but the lack of railings would not have

effectively separated circumambulation and group ritual. Further com- plicating the issue is that the best-known chaitya from the Northwest in the period is not located at a monastery, but rather within the city walls of Taxila.

The architectural layout of the chaitya in Block D at Taxila most closely resembles the layouts of Kondivte, Lomas Rishi, Sudama, and the wooden prototypes on which they were based. Rather than placing the stupa at the end of an apsidal hall, in Block D the stupa was placed in a circular room with a 4.3 m wide doorway allowing access from an attached hall (Marshall [1951] 1975:151–155). Marshall believed that the entire structure was surrounded by a narrow circumambulatory path located between two walls along the exterior of the structure (see Figure 4.9). While it is pos- sible that Marshall is correct in this interpretation, I believe it is unlikely. First, as shown in his map of the temple, the path would have been very narrow—less than one meter wide. This would be much narrower than the circumambulatory paths at any other archaeologically known chaitya. Second, given the design of the structure, circumambulation could easily have been performed in the circular room in which the stupa would have been located. Given all of this, it seems unlikely that Marshal’s identifi- cation of an exterior circumambulatory path is correct. Rather, the apsi- dal chaitya in Block D is best understood as a non-monastic urban shrine, recalling, in most respects, the design and layout of prototypical wooden chaityas from earlier centuries.

Whether by creating separate ritual spaces within one structure or by

creating separate ritual spaces in multiple structures, beginning in the first century bce the sangha in peninsular India created new architectural strategies that successfully spatialized the contradiction between the indi- vidual and the group, allowing worshipers to more effectively ignore the contradictions between the two. In the context of a developing religious tradition in the first century bce, the design of the chaityas in the Western Ghats and Andhra Pradesh did not resolve the tension between the indi- vidualizing aspects of Buddhism and the need to forge a common Buddhist community. Rather, the design of these chaityas allowed those engaged in ritual to more effectively ignore the problem. The same cannot be said of the monasteries in the Northwest. Here the layout of public worship spaces at monasteries followed the organization and layout of contemporary pil- grimage stupas. However, without railings, the spaces were less able to separate those circumambulating the stupas from those engaged in group rituals in the courtyards surrounding them. These differences in the layout of public worship space in the peninsula and the Northwest also indicate differences in the relationships between the sangha and the laity.

As discussed earlier, the spatial layout of the courtyards at pilgrimage complexes fostered more festive and communal forms of worship. Due to the similarities in the layout of courtyards at the monasteries in the Northwest, it seems likely that the same types of group worship would have occurred there as well. In contrast, in the peninsula the assembly areas of monastic chaityas were in the form of halls. Rather than the more festive and communal forms of group ritual practiced in the courtyards of pilgrimage complexes, the halls of monastic chaityas would have fostered more directed or mediated forms of ritual. As discussed in Chapter 2, halls fostered corporate worship by emphasizing the object of worship and rit- ual mediators while de-emphasizing fellow audience members. In monas- tic chaityas of the peninsula, the stupa was located at one end of a hall. This placement had the effect of forcing all audience members to face the same direction, allowing ritual specialists to place themselves between the audience and the stupa. This layout betrays an emphasis on more for- malized, monastically led worship. Brown’s (1965) description of abbots leading worship from wooden thrones at the base of stupas in modern Tibet and Nepal serves to illustrate this point.

Using the proximity of the different worship spaces to the stupa as a measure of their relative importance allows for the different forms of worship in the two types of stupa complexes to be ranked. Monastic chaityas in the Western Ghats place the circumambulatory path in a sec- ondary position to the stupa. As a result, the assembly hall has the most direct view of the stupa. On the east coast of the peninsula, circumam- bulatory paths are not even provided within the apsidal chaityas, but rather moved to smaller, subsidiary circular chaityas. All of the stupas in monastic chaityas in the peninsula are sufficiently elevated to allow for direct circumambulation and small railings. There is no functional reason that the same presentational systems could not be used in penin- sular monastic chaityas as were used in contemporary pilgrimage stupas. Thus, the placement of the circumambulatory path in a secondary posi- tion in peninsular chaityas was a decision based on monastic priorities, not architectural necessities. While the courtyards surrounding stupas in northwestern monasteries would not have hindered monastically led rit- uals, the lack of clearly defined circumambulatory paths forced devotees to circumambulate on the drums of stupas in full view of the worshipers in the courtyard surrounding them. Thus, in monasteries across India, individual worship was de-emphasized at the expense of communal or monastically led worship.

Together these analyses suggest some important differences between pilgrimage and monastic stupas from the second century bce through the

second century ce. Ritual at pilgrimage complexes was either communal or individual, with an architectural emphasis on the latter. The architec- tural forms of monastic chaityas de-emphasized individual worship and emphasized monastically led or mediated worship. Whatever the perspec- tive, the primary source of power and wealth for early Buddhist monaster- ies was their privileged relationship in regard to the Buddha. Support for monasteries was derived through the belief of guild members, merchants, royalty, and the laity that the sangha was the rightful heir of the Buddha and the arbiter of Buddhist practice. This power was achieved through the liminal location of monasteries, the sangha’s ascetic practices, and through the materialized presence of the living Buddha in the form of stu- pas at their monasteries. It is not surprising, then, that the sangha sought to project a sense of power by making their stupas appear large and impos- ing. If the sangha served as intermediaries between the Buddha and the laity, the more powerfully they presented the ritual focus of Buddhism, the more donations would flow into monastic coffers.


Beginning in the first century bce, there is clear evidence that the sangha was manipulating the shape of primary stupas in the Western Ghats to make them appear taller and more massive than they really were (Fogelin 2012). In the process, the sangha significantly altered the semiotic signifi- cance of stupas and, perhaps, some of the foundation of Buddhist theol- ogy. These changes can be seen in subtle alterations to the shape of the anda. Unfortunately, this type of analysis can only be done in the mon- asteries of the Western Ghats. The reason for this is simple. Whereas the well-preserved chaityas of the Western Ghats contain almost wholly intact stupas, elsewhere in India stupas are heavily damaged. The subtle differ- ences analyzed here cannot be seen on damaged andas.

While there is some variation in the form of monastic stupas in the Western Ghats, they were usually simpler than the contemporary pilgrim- age stupas. Monastic stupas typically consisted of three parts, one or two drums, the anda, and a chhatra (see Figures 4.13). While monastic chaitya halls did have circumambulatory paths, the sangha situated them along the perimeter of the worship halls rather than immediately adjacent to the anda. In some cases, the sangha had railings carved in low relief on the upper portion of the drums, but overall, circumambulatory paths were not part of the visual assemblage of signs in monastic contexts. In contrast to the stability of the design of pilgrimage stupas, monastic stupas’ material

forms and associated significations changed substantially between the first century bce and the second century ce.

As discussed in Chapter 2, archaeologists have long recognized that people often use elevation to assert power (Moore 1996). A ruler, for example, could seat himself or herself higher than the commoners who came to petition him or her. From a practice theory perspective, this material action helped create the power differential that existed between the ruler and the ruled. From a semiotic perspective, the elevation of the ruler was an index of his or her greater power. From either perspective, the design and construction of a royal hall reinforced the ideology that the ruler was attempting to promote. More broadly, throughout history rulers often attempted to build the largest and most elaborate structures to signify and create their own power. In this light, the enormous reli- gious structures of history have been regularly interpreted by archaeolo- gists as statements of power—glorifying both the gods enshrined within them and the rulers who controlled and built these structures (Moore 1996). These large structures, however, were often expensive and labor intensive. Rulers and the builders they employed often used tricks to make structures appear larger and more lavish without additional cost or labor. Pyramids, for example, could be placed on or completely encasing a small hill. By employing topography in this way, the same labor invest- ment resulted in a larger structure. In the New World, sunken courts were often excavated directly in front of pyramids. By lowering the position from which pyramids were viewed, the pyramids appeared larger (Moore 1996). In both of these cases, the builders altered the perception of eleva- tion in a simple and direct way. The structures, from the point of view of the audience, were genuinely, measurably taller.

There are, however, other manipulations to the form of structures that can make them appear taller or larger, without actually changing their measurable height or size in any significant way—manipulations that play off the expectations of semiotic objects within the minds of inter- pretants. These subtle manipulations of material objects are particularly useful when materializing power in the interior of a structure. Interior spaces can only be made so large before the roof collapses. Where room dimensions limit the physical height and mass of a material object placed within it—such as a stupa within a chaitya hall—visual tricks that make material objects appear taller and/or more massive can be employed.

While there are many different physical techniques to subtly alter the sensual perception of a material object, here I will discuss only two. I call these “attenuation” and “implied mass” (see Figure 4.12). Attenuation makes an object appear taller by making it thinner. This can be most

Figure 4.12: Attenuation and implied mass

clearly seen in depictions of the human body. In some African art, for example, the human form is severely attenuated to make figures appear very tall. Attenuation works in this case because all interpretants have pre-existing expectations of the proper proportions of the human body through regular viewing of bodies around them. An image cannot be rel- atively narrower if there is nothing to relate the image to. Attenuation, therefore, can only be employed in depictions of signs that are ubiquitous and standardized. While the human body is among the most ubiquitous signs, many other signs can also be examined in terms of attenuation. Here, I examine the attenuation of monastic stupas. Following the logic of attenuation, monastic stupas were only attenuated in comparison to contemporary pilgrimage stupas. The only reason that attenuation was successful in monastic stupas was that most interpretants viewed stupas with pre-existing knowledge of the proportions of pilgrimage stupas. This pre-existing image of an object roughly corresponds with Peirce’s defini- tion of an object—that is, the mental object that an interpretant creates when thinking of a sign.

Implied mass relies on many of the same assumptions as attenua- tion. The main difference is that rather than making a sign appear taller, implied mass alters the shape of a material object to suggest mass that is not present. Using the form of a human body again, when depicting part of a person (e.g., a person standing behind a low wall), the interpretant imagines (semiotic object) the rest of the body. The mass of that imagined portion of the body is added to the mass of the visible portion. In monas- tic stupas, the sangha implied additional mass by lifting the midpoint of the anda above the plain of the drum (see Figure 4.12). The interpretant imagined the remainder of the anda within the drum, much as he or she might have imagined the bottom side of a ball floating in water. In effect, the anda was perceived to include both the visible mass as well as the mass of the semiotic object concealed within the drum. Like attenuation, implied mass relies on a strong familiarity with the sign being depicted. When encountering a novel or irregular shape, a shape for which there is no pre-existing semiotic object, the interpretant cannot fill in what is lacking.

Together, attenuation and implied mass can be used to understand the physical metamorphosis of early monastic stupas in the Western Ghats from the first century bce through the second century ce. Relying on eleva- tions by Fergusson and Burgess ([1880] 1988), Nagaraju (1981), and Mitra (1971), I made careful measurements of the fifteen best-preserved stupas located within the largest chaityas in the Western Ghats (see Table 4.1). Based upon Dehejia’s (1972) chronology of these sites, the sangha had these chaityas carved between 100 bce and 140 ce. Of the fifteen stupas I examined, seven are attenuated, while ten are forms with implied mass (see Table 4.1). One important point is that these two techniques are not mutually exclusive. The stupas in Ajanta 9 and Karle 8 exhibit both atten- uation and implied mass. In general, attenuated forms dominate in the earliest phase, with implied mass used to the exclusion of attenuation in later times.


Particularly in the case of attenuation, the physical manipulation of the shape of stupas was subtle. For that reason, it is not possible to look at images of different stupas and easily see the effects of attenuation. Rather, I have made careful measurements of a variety of stupas that are docu- mented in the existing archaeological literature on India. I then use a simple formula to measure the degree of attenuation that has occurred.



Stupa Date of


Attenuation Implied


Degree of Attenuation (2h/d)



Kondivite 9

100 bce

  • 1.3


Bhaja 12

80 bce

  • 1.2


Ajanta 10

75 bce

  • 1.1


Junnar-Tuljalena 3

65 bce

  • 1.3


Ajanta 9

65 bce

  1.3


Aurangabad 4

60 bce


Bedsa 7

40 bce

  • 0.9


Karla 8

60 ce

  1.1


Ganesh Pahar 6

95 ce


Nasik 18

125 ce


Sivaneri 43

130 ce


Ganesh Pahar 14

130 ce


Kanheri 2e

135 ce


Kanheri 2c

140 ce


Kanheri 4

140 ce


This formula is two times the height of the anda, divided by its diameter (2h/d: see Table 4.1). In essence, this formula is only a ratio of the hori- zontal diameter to the vertical diameter of a sphere. In this formula, a perfect hemisphere would have a value of 1. Figures greater than 1 signify that the anda is attenuated, with figures less than 1 showing the opposite. As shown in Table 4.1, most of the earliest stupas were attenuated, with the exception of Bedsa 7 and Aurangabad 4. Excluding these two for the moment, the degree of attenuation of these stupas ranges between 1.3 for the stupa in Kondivte 9 to 1.1 at Karle 8. The average is 1.22. As for Bedsa 7, this is a slightly odd stupa for its time (see Figure 4.13). Unlike other con- temporary monastic stupas, it has two drums, with the upper drum the same diameter as the anda. If it were not for the relief of a railing carved onto its surface, visually separating the anda from the drum, this would be among the most attenuated of all of the stupas I examined. It seems to me that this stupa is attenuated, but through its drums rather than its anda. In any case, based upon the other contemporary stupas, I feel it is fair to argue that typically the earliest monastic stupas were attenu- ated. This, in turn, suggests that those who designed and constructed these stupas were attempting to make these stupas appear taller than they actually were. Within the restricted space of a chaitya, the sangha used

Figure 4.13: Attenuated stupas

attenuation to give their ritual focus greater authority. In the process, the sangha altered the ancestral form that the anda was based upon, a mound of earth, in favor of a legitimation that projected greater authority via greater perceived elevation.

The ancestral earthen stupas, and the later pilgrimage stupas modeled on them, were typically fairly stout (2h/d less than 1). Earthen stupas could not have steep sides due to the problems of erosion and slumping. The attenuated form of monastic rock-cut stupas would have been excep- tionally difficult to achieve using the medium (earth) of ancestral stupas. For this reason, the attenuation of early monastic stupas in the Western Ghats can be seen as making these stupas somewhat less iconic and somewhat more symbolic of ancestral stupas. This shift in the semiotic

significance of monastic stupas was accelerated with the construction of

stupas with implied mass.

Implied Mass

Numerous scholars have noted a distinct break in the construction of stupas in the Western Ghats between 40 bce and 60 ce (Dehejia 1972; Nagaraju 1981). The Satavahanas were among the primary donors to the monasteries in the Western Ghats, and this break in construction coin- cides with a prolonged period of weakness in the empire. I am also not the first to notice that there is also a profound difference in the shape of stupas before and after this break. Nagaraju (1981), for example, distin- guishes between hemispherical and 3/5 spherical andas. The difference in my work is the meaning I ascribe to these differences. Where Nagaraju and others are principally interested in constructing a chronology, my interest is in the way the shapes of stupas support monastic interests.

The identification of stupas with implied mass does not require any special measurement or analyses. They are easily identified by andas that reveal a portion of the underside of a sphere (see Figure 4.14). As discussed earlier, stupa forms with implied mass did not make stupas appear taller than they actually were. Rather, they made stupas appear to have greater mass. This effect was achieved by showing a small portion of the underside of a sphere, allowing the interpretant’s mind to complete it. I suggest that while the form of monastic stupas shifted between the earlier and later peri- ods, the different forms served the same function—promoting monastic power and wealth by increasing the perceived authority of a monastery’s primary ritual focus. Just as with elevation, the perceived increase in the mass of the stupa would have served to reinforce the importance of it. This, in turn, would have served to enhance the source of monastic ritual authority. The development of the 3/5 anda also signified another shift in the degree of abstraction from the ancestral earthen stupas of which the anda was an icon. No earthen mound could ever be undercut in the same way that these 3/5 andas were. The dirt overhang would have collapsed. If attenuation made monastic stupas less iconic, 3/5 andas further shifted the mode of stupas from iconic signs toward symbolic signs.


The shift from iconic to symbolic stupas was also facilitated by the omis- sion of relics in later stupas in the monasteries of the Western Ghats.

Figure 4.14: Implied mass stupas

In the first century bce, the sangha had the chhatras of most stupas in the Western Ghats carved from wood (see Table 4.1). They placed these wooden chhatras in sockets at the apex of the andas. Given the fragility of this construction technique, none of the original wooden chhatras remains. Beginning around 40 bce at Bedsa (see Figure 4.13), the sangha had chhatras carved in conjunction with the anda. The chhatras of these later stupas consisted of a column of stone rising from the apex of the anda, connected to the parasol carved in low relief on the ceiling of the chaitya (see Figure 4.14). This shift in the medium of the chhatras had a profound impact on the semiotic significance of monastic stupas in the Western Ghats.

Since the andas of monastic stupas were stone rather than mounded earth, the inclusion of a relic required carving a void within and an access route into the anda. No such void is present in the rock-carved stupas of the Western Ghats. Some researchers have proposed that relics could have been placed in the bottom of the sockets used to support wooden chhatras of early monastic stupas (Dehejia 1972; Mitra 1971). Just such a pattern is found in some smaller subsidiary stupas surrounding the ninth century ce stupa at Ratnagiri in northeastern India (Mitra 1981, 1983). It is likely that the earliest stupas in the Western Ghats followed a similar practice. However, by 40 bce the sangha in the Western Ghats began omitting relics from their stupas entirely. The stone chhatras of the later periods provide no possibility for the inclusion of a relic since the sangha had them carved from the very same stone as the anda—there is no socket in which to place the relics and no other access route to the interior of the anda (see Table 4.1). This suggests a major shift in the significance of semiotic elements that constitute monastic stupas. Beginning in the mid-first century bce and more commonly in the first century ce, monastic stupas of the Western Ghats were no longer indexes for icons residing within.

Signs of Legitimation in the Western Ghats

From the first century bce through the second century ce, the Buddhist sangha altered the form of monastic stupas in the Western Ghats from a low mound, to an attenuated hemisphere, and finally to a 3/5 sphere. Each of these steps resulted in progressively greater detachment from andas’ ancestral form. Each step made the stupas less iconic and more symbolic. Each modification to the shape of monastic stupas was an attempt to pro- mote monastic authority—initially by making stupas look taller and later making stupas look bigger. The sangha also began omitting relics from many stupas, removing the most iconic element of the Buddha from their ritual focus. The omission of the relic ended the indexical significance of the anda. In the end, the sangha shifted the signification of stupas from an icon of the Buddha encased within an index of his presence, to a symbol of the Buddha with progressively less iconic similarities to the eight ancestral stupas of the Buddha. Following the logic of Peirce’s semiotics, by the sec- ond century ce, stupas were less emotionally immediate (firstness) than they were originally and were more intellectual and abstract (thirdness). In Chapter 5, I argue that this reduction in the emotional immediacy of stupas partially explains the abandonment of monastic stupas, the origin

of Mahayana Buddhism, and the creation of Buddha images in the first through sixth centuries ce.


Since the advent of textual scholarship on early Indian Buddhism in the mid-nineteenth century, the sangha has been consistently depicted as isolated ascetics, allowed to meditate on otherworldly concerns through the devotion and support of the laity. This understanding of the early sangha was derived from selected Buddhist sutras composed in the first millennium ce. When examined in light of what is known from archaeol- ogy, epigraphy, and iconography—the only sources that actually date to the period—the sangha appear far less interested in asceticism than has been typically presented. Rather, between the second century bce and the second century ce the sangha sought to lead a unified community of Buddhists. To this end, the sangha created public worship halls in their monasteries that promoted monastic authority. In the Western Ghats, the sangha even manipulated the shape of their stupas to make them appear larger and more massive than they actually were—further reinforcing the sangha’s authority over the laity. Based on Shaw’s (2000, 2007, 2013b) work at Sanchi, the sangha may also have been actively involved in distinctly non-ascetic practices like intensifiying the agricultural infrastructure of the regions in which they lived. Through these assorted legitimations, the sangha attempted to routinize the charisma of the Buddha for their own material gain and began to accumulate substantial endowments. In short, by at least the first century bce and perhaps earlier, the sangha engaged politically, economically, and spiritually with the laity. As such, from the earliest time in which evidence is available, the sangha was forced to bal- ance their individual and communal desires. For this reason the sangha created spaces within their monasteries (monastic cells and circumam- bulatory paths) to allow for the continued practice of private, meditative rituals. However, by the first century bce the styles of worship used by the sangha to balance their individual and communal roles had significantly diverged from the styles of worship engaged in by the laity.

While the sangha de-emphasized the practice of individual circumam-

bulation, the laity continued to consider it the primary form of worship. While the sangha sought to create monastically led group rituals, the laity favored festive communal rituals that hindered the elevation of ritual leaders. While the sangha increasingly worshiped symbolic signs of the

Buddha, the laity continued to worship iconic relics of the Buddha in much the same way as they always had. Contrary to standard Buddhist histo- ries, the laity preserved the ancestral forms of Buddhism and resisted the sangha’s attempts to alter Buddhist practices.

Over subsequent centuries, the sangha continued to modify and alter their ritual practices and theological beliefs, while the laity continued in their religious practices in much the same way as they always had. Beginning in the first century bce, monastic and lay Buddhism began to diverge. Over the first millennium ce the laity slowly shifted their allegiance to new religions that better fit the individual and communal desires that the laity had never abandoned (see Chapter 6). With the failure of their legitimations, the sangha began to withdraw from regu- lar contact with all but the wealthiest laity beginning in second century ce—retreating into their monasteries and surviving on their endowments and business dealings. In this context, the sangha began creating a new form of Buddhism—Mahayana Buddhism—that legitimized a newfound interest in isolation and asceticism.




The Beginnings of Mahayana Buddhism, Buddha Images, and Monastic Isolation

c. 100–600 ce


n the first half of the first millennium ce, the Buddhist sangha progres- sively withdrew from regular contact with all but the wealthiest laity as monasteries became more economically self-sufficient. This chapter explores the political, doctrinal, architectural, and symbolic changes that facilitated this inward shift. In both theological and material ways, the sangha forged a more personal or individual connection to the Buddha, conceived as the prototypical ascetic. Among the more important mate- rial developments in this period was the creation of Buddha images, often placed within the living quarters of monasteries. The creation and place- ment of Buddha images materialized this new ideology of asceticism. Textual accounts of the idealized practices of ascetic monks—the sources used in most text-based histories of Buddhism—are derived from this period and do not necessarily reflect earlier Buddhist practice. As such, these sources are not straightforward historical accounts of the Buddha or early Buddhism but rather the invention of tradition. This is not to say that these texts do not have value. While some of the information pro- vided in these texts is valuable for understanding early Buddhism, these accounts are even more valuable for understanding monastic Buddhism at

the time of their composition.

( 146 )


The time period covered in this chapter, 100 ce–600 ce, overlaps with Chapter 4 by a century. This overlap is meant to accommodate the gradual developments in Indian Buddhism in the first half of the first millennium ce. There is no clear break between the forms of Buddhism discussed in earlier chapters and the emergence of Mahayana Buddhism discussed in this chapter. The same can be said of the political developments of the period. While several new states emerged in the fourth century ce, sev- eral of the states discussed in Chapter 4 continued to wield significant power through the third century ce. In the Northwest, the Kushanas held power through the third century, while King Kharavela of the Kalinga lived in the second century, with his descendents slowly losing power over the course of the third century. The two largest empires in India between 100 and 600 ce were the Guptas (c. fourth–sixth centuries ce) of North India and the Vakatakas (c. mid third–fifth centuries ce) of the Deccan Plateau. Numerous smaller states were located in the peripheries of the Northwest, Northeast, and extreme South.

In many histories of India, the Guptas have an outsized importance that well exceeds what is actually known of them. For a variety of his- torical and political reasons, the Guptas have become central to modern Indian identity. First, the foundational texts of modern Hinduism, includ- ing the Puranas, Mahabharata, and Ramayana, are likely to have been tran- scribed and collated during this time. Thus, the “Gupta period” is often seen as the time when more ancient Vedic Brahmanism was transformed into modern Hinduism. Second, several lines of evidence, particularly inscriptions recording land donations, show that the Guptas supported nascent Hinduism. These historical developments tie directly to more modern political reasons for the importance of the Guptas in postcolonial histories. Histories of India written by British colonial officials tended to focus on the Mauryas. Colonial era histories also emphasized the Aryan or foreign origin of the Mauryas, with obvious implications for the legiti- macy of British rule in Colonial India. In contrast,

Indian historians who lived during the period of Nationalist resistance to colonial rule portrayed the Gupta period as ‘golden age.’ The glorification of the Gupta period can be viewed as a reaction of Nationalist historians to Imperialist historiography. The features that were highlighted included the political unification of a large part of the subcontinent under what was pre- sumed to be a centralized government, the production of exceptionally fine works of Sanskrit literature, significant developments in the spheres of stone

sculpture and architecture, and a presumption that all of this was based on economic prosperity and social harmony. (Singh 2008:473)

In short, nationalist historians emphasized the Guptas, in part, to pro- vide historical precedent for Indian self-rule.

By the 1960s and 1970s, Indian postcolonial histories of the Guptas took on a more Marxist orientation—viewing the period as characterized by a form of feudalism in which Indian peasants were tied to the land they worked while being forced to pay taxes and rent to higher status land own- ers (Kosambi 1965; Sharma 1980; Thapar 1966). Rather than a “golden age,” from the Marxist perspective the Gupta period was a time of political fragmentation and urban decline (Singh 2008:473). More recently, Indian historians have taken a more balanced approach to the period, seeing it more as a time in which new political, religious, economic, and artistic traditions emerged, serving as the foundation for later Indian states and empires. Thapar succinctly summarizes this line of thought by labeling the period “threshold times” (Thapar 2002:ch. 9). While there is clearly value in seeing the Gupta period as a cauldron for later developments, there is something odd about discussing the Guptas only as a step toward something else. By treating the Guptas as merely a foundation for later states and empires, the Guptas themselves are lost.

Whether understood as a golden age, feudal society, or threshold to what came later, the sources of evidence for the Gupta period consist of the same material: royal inscriptions, coins, references in Sanskrit literature, and foreign (mostly Chinese) travelers’ accounts. While listed in genealo- gies found in Gupta inscriptions, little is known of the first two Gupta kings. It appears that these kings ruled in portions of modern Bihar, but it is unclear if they were an independent state or subordinate to another. In 319 ce, Chandragupta I ascended the throne. Over the next sixteen to seventeen years he expanded the kingdom and established a marriage alli- ance with the Licchavi kingdom in northern Bihar. Chandragupta I was succeeded by Samudragupta I (c. 335/350–370 ce), though some evidence suggests that Samudragupta I usurped the throne from an intermedi- ary king. The details of Samudragupta I’s reign come mostly from a long eulogy on the Mauryan pillar at Allahabad. Here Samudragupta I had his own biography carved above that of Ashoka, detailing his many conquests and his love for poetry and music. The peak of Gupta power was achieved under the reign of Chandragupta II (c. 376–413/415). During this time the Guptas controlled most of North India with significant influence, but not territorial control, in the peninsula. Over the course of the fifth and sixth centuries, rivals progressively challenged Gupta hegemony. After

Chandragupta II, the Gupta Empire gradually lost power and slowly con- tracted to its original base in Bihar.

Like the Guptas, the origin of the Vakatakas is murky, with some schol- ars placing it in Andhra Pradesh and others in the Deccan (Shastri 1997). Wherever the Vakatakas originated, they first became a major dynasty in the Deccan beginning in the mid-third century ce. A later inscription by the king Harishena (c. 475–500 ce) names Vindhyasakti I as the founder of the dynasty. The Vakatakas expanded their empire across most of the Deccan under Pravarasena (c. 270–330 ce). Pravarasena is also recorded as forming a marriage alliance with the Guptas. After Pravarasena, the genealogy of the Vakatakas becomes muddled. Inscriptions and accounts in the Puranas list numerous kings, often serving at the same time. Overall, it appears that the Vakatakas bifurcated into at least two sepa- rate branches, with the dynasty eventually ending in the late fifth or early sixth century. In terms of Buddhist history, the fifth century Vakataka king Harishena is notable for his patronage of the Buddhist monasteries at Ajanta.

Between the second and sixth centuries ce, several different kingdoms controlled portions of the northwest. Until the mid-third century, the region was dominated by the Kushanas (discussed in Chapter 4). In the fourth and fifth centuries, the Guptas may have had nominal control in the Northwest, though regional rulers, particularly in the more distant margins, may have retained a large degree of autonomy. Beginning in the mid-fifth century ce, Huns from Central Asia began a series of cam- paigns in the Northwest. While initially repulsed by the Guptas, by the early sixth century ce, the Hun king Toramana had established an empire in Kashmir and Gandhara. While deriving from Central Asia, the Huns rapidly adopted Indian patterns of rulership, with eighth-century ce Jain accounts claiming that Toramana converted to Jainism.

Like the Mauryas before them, the Guptas, Vakatakas, and other con- temporary states materially supported a wide variety of religious sects. This included both Buddhist and Jain orders, but also the develop- ing Shaivite and Vaishnavite sects of Hinduism. Based on inscriptions, numismatics, and the Puranas, it appears that both Gupta and Vatataka kings performed Vedic sacrifices and were devotees of the nascent Hindu gods. In Chapter 6, I extensively discuss how the emergence and spread of Hinduism through Indian society helped lead to Buddhism’s gradual decline in India. For now, it is sufficient to note that Buddhism was one of many religions in simultaneous practice in India and that all of these reli- gions were highly syncretic, combining and borrowing elements from rival religious orders.

Among the more critical developments during this period was a gradual shift in the type of donations that the royalty and other elite gave to reli- gious orders. Where previously these gifts consisted of money, goods and, in a few instances, land (e.g., Nasik), between the second and sixth centu- ries ce, donations increasingly took the form of land. This was part of a larger shift in the economic organization of Indian society. For both the Guptas and the Vakatakas, control and taxation of land was the primary basis of wealth and power. Taxes were typically one sixth to one fourth of local production. While few modern scholars argue that land tenure during this period can be described as feudal, and the people working the land as serfs, there is no question that Indian kings had significant rights to levy taxes within their domain and, more important, to reassign these rights to other individuals and groups. That is, land—along with the taxes produced by the people working the land—was regularly gifted to military leaders, lesser elite, and religious orders. The full details of land tenure were com- plicated and varied throughout India with, for example, substantial differ- ences between the gifting of cultivated and uncultivated land. In numerous inscriptions on temple walls and copper plates, the details of these gifts are painstakingly detailed. The inscriptions typically demarcate the land that was given, the specific taxes owed to the king that were waived, and the conditions that must be met for the recipient to keep the land. In the end, however, the gifting of land to religious orders led to a marked shift in their finances. Where previously religious orders, the Buddhist sangha included, were heavily dependent upon regular donations to keep their monasteries solvent, with the donation of land in perpetuity, religious orders became more self-sufficient. This new economic foundation allowed for a marked change in the religious doctrines of monastic Buddhism. Without the need of financial support from the bulk of the laity, the sangha was free to refashion their religion in ways that emphasized monastic seclusion.


In the first half of the first millennium ce, a new strand of Buddhism grad- ually emerged in China and India.1 This new strand was called Mahayana

  1. Contrary to the India-centric view of Buddhism, it appears that the Mahayana tradition was initially more commonly practiced in China than it was in India (Schopen 2005:Ch. 1). That is, in China, Mahayana was a common and widespread tradition from at least the third century ce. In India, Mahayana only really began to take hold in the fifth century ce, and Mahayana practitioners were likely a minority of the sangha for several centuries after that.

    (greater vehicle) Buddhism. Mahayana Buddhists referred to earlier forms of Buddhism as Hinayana (lesser vehicle). Since Hinayana was a somewhat derogative term, here I use the term “early Buddhism” to refer to the Buddhist traditions that predated and coexisted with Mahayana Buddhism in India. Whereas early Buddhism focused on the life, and pre- vious lives, of the Buddha himself, Mahayana Buddhism added several other incarnations of the Buddha as well as numerous other Bodhisattvas. Bodhisattvas were beings who had achieved enlightenment but, unlike the historical Buddha (Siddhartha Gautama) of early Buddhism, some Bodhisattvas (e.g., Avalokiteshvara) delayed final enlightenment in order to assist all beings in their own attainment of salvation. Where early Buddhists believed the Buddha had attained nirvana and left this world behind, Mahayana Buddhists understood the Buddha and other Bodhisattvas to be continuing, compassionate presences. In this sense, Mahayana Buddhists viewed the Buddha and the Bodhisattvas as more active and immediate than early Buddhists.

    Mahayana thought and scripture emerged slowly over several centu- ries. In his account of his travels in India in the early fifth century ce, the Chinese pilgrim Faxian noted that while Mahayana and early Buddhism were more or less popular in different regions, both were practiced in most regions and, critically, Mahayana and early Buddhists often lived within the same monasteries. Mahayana Buddhism was not a single, coherent body of thought that emerged at a specific time and place. For this rea- son, it is not possible to provide a simple description of the principles of Mahayana Buddhism that differentiate it from earlier Buddhist tradi- tions, though, with time, the differences between Mahayana and earlier Buddhist traditions become more pronounced.

    Many of the foundational concepts of early Mahayana Buddhism are found in several texts, collectively known as the Perfection of Wisdom Sutras. While some argue that the Perfection of Wisdom Sutras were authored in the first few centuries bce, modern scholars place them in the first half of the first millennium ce (Lopez 2001). In either case, the Perfection of Wisdom Sutras were later additions to the Buddhist canon, and marked the beginnings of what became Mahayana Buddhism. Though they are later additions to the Buddhist canon, Mahayana texts were presented as faithful accounts of the Buddha’s sermons. This seem- ing discrepancy was explained through calls to a set of secret sermons and texts, hidden from the Buddha’s followers, only to be rediscovered when the sangha was ready for these more advanced teachings. Thus, vari- ous Perfection of Wisdom Sutras were said to be hidden in caves, placed

    in relic caskets at the bottom of the sea, or divinely revealed to their authors.

    [T]he Perfection of Wisdom texts were not systematic treatises that set forth philosophical points and doctrinal categories in a straightforward manner. Instead, they strike the modern reader as having something of the nature of revelations, bold pronouncements proclaimed with certainty rather than speculative arguments developed in a linear fashion. The perfection of wisdom that the sutras repeatedly praised, rather than presented, was the knowledge of emptiness (sunyata). To see that all phenomena are empty is to see the truth. (Lopez 2001:27)

    Among the more important of the early proponents of doctrine of empti- ness was Nagarjuna. Though his biography is poorly known, most modern scholars believe he lived in the first and second centuries ce, perhaps in South India. He is credited with numerous texts, including the Perfection of Wisdom in One Hundred Thousand Stanzas and the Treatise on the Middle Way. Both of these texts are extremely complex and cryptic, resulting in numerous Buddhist commentaries and an extensive academic study that cannot be suitably addressed here.2 Perhaps the most concise explanation of emptiness can be found in the Heart Sutra (Lopez 1988, 1996), a short work that has become the most popular and common sutra of Mahayana Buddhism (see Box 5.1). Though still complex, the Heart Sutra shows that the recognition that everything, including the self, is empty is a prerequi- site for the attainment of nirvana. As such, the development of Mahayana Buddhism marked a shift in the degree of philosophical abstraction over and above the forms of Buddhism that had preceded it.

    In the centuries after the creation of the Perfection of Wisdom Sutras, Mahayana scholars also began arguing that people have an innate Buddha-nature—an inner Buddha consciousness yet to be discovered. Through the guidance of Bodhisattvas, prominent monks, and the discov- ery of the Buddha-nature, Mahayana Buddhists saw a faster and more uni- versal path to enlightenment. This view, in at least an abstract sense, also favored more ascetic and meditative practices. That is, if people already have an innate Buddha-nature, the path to enlightenment is ultimately a path of self-discovery. This explains, in part, the emphasis on asceticism within early Mahayana texts—though, as discussed below, this roman- tic view of asceticism was not commonly practiced in the second through

  2. For a concise discussion of the Perfection of Wisdom Sutras and the concept of emptiness, see Lopez (2001:27–33; for a more detailed discussion, see Lopez 1996).

Box 5.1


(Lopez 1996:vii–viii)

Thus did I hear. At one time the Bhagavan [Lord Buddha] was abiding at Vulture Peak in Rajagrha with a great assembly of monks and a great assembly of bodhisattvas. At that time, the Bhagavan entered into a samadhi on the categories of phenomena called “perception of the pro- found.” Also at that time, the bodhisattva, the mahasattva, the noble Avalokitesvara beheld the practice of the profound perfection of wisdom and saw that those five aggregates also are empty of intrinsic existence. Then, by the power of the Buddha, the venerable Sariputra said this to the bodhisattva, the mahasattva, the noble Avalokitesvara, “How should a son of good lineage who wishes to practice the profound perfection of wisdom train?” He said that and the bodhisattva, the mahasattva, the noble Avalokitesvara said this to the venerable Sariputra, “Sariputra, a son of good lineage or a daughter of good lineage who wishes to prac- tice the profound perfection of wisdom should perceive things in this way: form is empty; emptiness is form. Emptiness is not other than form; form is not other than emptiness. In the same way, feeling, discrimina- tion, conditioning factors, and consciousnesses are empty. Therefore. Sariputra, all phenomena are empty, without characteristic, unproduced, unceased, stainless, nor stainless, undiminished, unfilled. Therefore, Sariputra, in emptiness there is no form, no feeling, no discrimination, no conditioning factors, no consciousnesses, no eye, no ear, no nose, no tongue, no body, no mind, no form, no sound, no odor, no taste, no object of touch, no phenomenon, no eye constituent up to and includ- ing no mental consciousness constituent, no ignorance, no extinction of ignorance, no aging and death up to and including no extinction of aging and death. In the same way, no suffering, origin, cessation, path, no wis- dom, no attainment, no nonattainment. Therefore, Saliputra, because bodhisattvas have no attainment, they rely on and abide in the perfec- tion of wisdom; because their minds are without obstruction, they have no fear. They pass completely beyond error and go to the fulfillment of nirvana. All the Buddhas who abide in the three times have fully awak- ened into unsurpassed, complete, perfect enlightenment in dependence on the perfection of wisdom. Therefore, the mantra of the perfection of wisdom is the mantra of great knowledge, the unsurpassed mantra, the mantra equal to the unequaled, the mantra that completely pacifies all suffering. Because it is not false, it should be known to be true. The man- tra of the perfection of wisdom is stated thus: [om] gate gate paragate para- samgate bodhi svaha. Sariputra, a bodhisattva mahasattva should train in


the profound perfection of wisdom in that way.” Then the Bhagavan rose from samadhi and said, “Well done” to the bodhisattva, the mahasattva, the noble Avalokitesvara. “Well done, well done, child of good lineage, it is like that. It is like that; the practice of the profound perfection of wisdom is just as you have taught it. Even the tathagatas admire it.” The Bhagavan having so spoken, the venerable Sariputra, the bodhisattva, the mahasattva, the noble Avalokitesvara, and all those surrounding, and the entire world, the gods, humans, demigods, and gandharvas, admire and praised the speech of the Bhagavan.

sixth centuries ce. In later centuries, with the rise of Tantric Buddhism (Chapter 7), the idea of an innate Buddha-nature became progressively more central to Buddhist thought, with a corresponding increase in the number of Buddhist ascetics.

There is substantial debate over the geographic origins of Mahayana Buddhism in India. Some argue that Mahayana Buddhism emerged in the mainstream Buddhist monasteries of the Gangetic Plain in the first through fifth centuries ce, spreading outward from there (Lamotte 1988). More recently, Schopen (2005:ch. 1) has argued that Mahayana Buddhism initially developed among small factions of the sangha living in the periph- eries of India, in the Northwest, the Northeast, and peninsular India. For example, some of the biographies of Nagarjuna claim he lived in South India, well outside the Buddhist heartland in the Gangetic Plain. The rhet- oric of these fringe monastics, Schopen (2005:14) argues, reveal them to be “a small, isolated, embattled minority group struggling for recognition” from the settled monastic communities. Using inscriptional evidence from the second through sixth centuries ce, Schopen argues that fringe monastics, like Nagarjuna, initially developed and promoted Mahayana Buddhism, with the mainstream monasteries in the Buddhist heartland only adopting Mahayana teachings beginning in the sixth century ce (Schopen 2005:ch. 1). Whether emerging in the Gangetic Plain or in the peripheries, the shift from early Buddhism to Mahayana Buddhism was a gradual process, taking several centuries. The debate centers on the degree of marginalization that early Mahayana practitioners in India faced in the first through fifth centuries ce.

A final element of many early Mahayana sutras are their frequent calls for the abandonment of monasteries and a return to the more ascetic life of the forest. In part, this call to asceticism can be viewed as a way that early Mahayana Buddhists discredited the arguments of the mainstream

by pointing to the corrupting influences of wealth, land, honors, and even wives on the mainstream sangha (Schopen 2005:15).

It is clear that by the time of the final composition of the mainstream Vinayas . . . ascetic practices were—for the compilers—all but dead letter. . . . It is, however, equally clear that some strands of early Mahayana sutra litera- ture were attempting to reinvent, revitalize, or resurrect these extreme ascetic practices.

While it is clear that early Mahayana sutras celebrated asceticism, advocating life in the forest and the renunciation of the corrupt life of the mainstream monasteries, it less clear whether early Mahayana Buddhists actually abandoned life in the monastery for the solitary lives of the ascet- ics. Phrased more simply, was the asceticism of Mahayana texts real or ideal? These questions can only be addressed through examinations of the material evidence of Buddhism, the archaeological and iconographic evi- dence of Buddhist practice in the second through sixth centuries ce.


While it is tempting to see Buddha images as indicative of the advent of Mahayana Buddhism in India, evidence from the Northwest of India chal- lenges this simple understanding (Behrendt 2007; Leidy 2008). The earliest Buddha images in India were sculpted in Gandhara and Mathura in the sec- ond and third centuries ce. From this location in the northwestern periph- ery of India, the tradition of Buddha images spread across other portions of India and beyond. The earliest Buddha images, however, did not depict Mahayana figures—they were depictions of biographical events in the life of the historical Buddha (Lamotte 1960; Leidy 2008; Schopen 2005:11–12). These early images were consistent with the theology and doctrines of early Buddhism. While these images demonstrate that the earlier taboo on depictions of the Buddha was being challenged, they do not signify the advent of Mahayana Buddhism. This is not to say that the origin of Buddha images and Mahayana Buddhism are unrelated. They both testify to a shift the practice of Buddhism between the first and fifth centuries ce. Both Buddha images and Mahayana Buddhism signify a greater emphasis on the person of the Buddha. By the fifth century ce, Mahayana elements first begin to appear in Buddha images.

Among the earliest Buddha images from Mathura are a series of larger-than-life standing Buddhas that were erected at Shravasti,

Figure 5.1: Early Buddha image from Sarnath (second century ce)

Courtesy of the Digital South Asia Library and the American Institute of Indian Studies (Accession No. 28775).

Kausambi, and Sarnath (see Figure 5.1). Inscriptions on these images record that they were commissioned by a small number of interconnected people, primarily a monk named Bala and a nun named Buddhamitra (Schopen 1997:Ch. 11). In the inscriptions, Bala and Buddhamitra are described as knowing the three Pitakas, a phrase intended to convey their great learning. Finally, these inscriptions record that the images were made in the early years of the Kushan Emperor Kanishka’s reign, about 130–135 ce. As summarized by Schopen (1997:248),

The earliest cult images at three of the most important Buddhist sites in the Ganges Basin—Kausambi, Sravasti, and Sarnath—almost certainly came from Mathura, where scholarly opinion is more and more inclined to locate the production of the first Buddha images. The production, transportation, and installation of all these images—again, the first at these sites—was effected by at least two monastics who knew one another in one or more capacities. And both of these individuals were, in their contemporary idiom, very learned. All of the evidence suggests that these learned monastics were, in Basham’s

(1981:30) words, ‘propagandists for a new cult,’ and that this propaganda was effected in a systematic manner.

Stylistically, the earliest Buddha images from Gandhara (c. second– fourth centuries ce) are distinct from those in Mathura. First, Gandharan images have more Greek influence, particularly in the way the artists carved the folds in the Buddha’s robes. Second, in Gandharan images there is a greater emphasis on depictions of the Buddha’s life and previous lives. Thus, various early Gandharan Buddha images depict the Buddha’s birth, his gaunt form after practicing austerities, his first sermon, and his death. In this sense, early Gandharan images have more narrative elements than their Mathuran counterparts.

The earliest images with Mahayana elements may date as early as the second or third century ce, with Mahayana images becoming com- mon only in the fifth and sixth centuries ce (Bareau 1985; Leidy 2008; Schopen 2005:11–12). In Mahayana Buddhism, the pantheon of Buddhas and Bodhisattvas was large and complex. Artistic representations of Bodhisattvas were differentiated by the use of standardized signs of their identity. Avalokiteshvara, for example, was typically depicted as a king holding a lotus bud. Maitreya, on the other hand, was often depicted as wearing elaborate jewelry and holding a water jar in his left hand. These standardized depictions of the Buddha and Bodhisattvas varied geo- graphically and temporally, allowing researchers to roughly date and provenance their carving.

Traditional academic scholars viewed Buddha images as a degradation of early Buddhism through the gradual incorporation of lay Buddhist rit- ual practices (Coomaraswamy 1927; Lamotte 1988; Tambiah 1976). That is, between 500 bce and 500 ce, Buddhist monks and nuns progressively adopted the “vulgar” practices of the laity in order to gain their material support—initially through the worship of relics within stupas and sub- sequently the worship of Buddha images. As argued in Chapters 3 and 4, based on the archaeological evidence, it appears that monks and nuns participated in stupa ritual with the laity from at least the third century bce—the earliest period for which there is any direct archaeological or inscriptional evidence (Fogelin 2003, 2006; Schopen 1997; Strong 2004; Trainor 1997). Similarly, Schopen (1997:ch. 238) has argued that in the second through fifth centuries ce the sangha initially and disproportion- ately promoted Buddha images, with the laity only adopting the practice later. Of the eighteen inscriptions on early Buddha images at Sarnath from the Kushan and Gupta periods that list the occupation of the donor, fifteen list the donor as a member of the sangha while only three list the donor as a layperson (Schopen 1997:240). At Ajanta, in the fifth century

ce, of the thirty-six donation inscriptions on Buddha images, thirty-three name members of the sangha and only three name laypeople (Schopen 1997:241). Schopen identifies similar percentages of image donation by the sangha at other sites in western and northwestern India (Schopen 1997:241–244).

Taken together, it appears that the Buddhist sangha was “vulgar” right from the start and, in the case of Buddha images, led the way. To these revisions of Buddhist history I add another. Traditional historians see early monasteries as isolated retreats and later monasteries as corrupted through their regular contact with the laity. In contrast, I argue that Buddhist monasteries were actively engaged the laity from the start, only becoming isolated retreats in about the fifth century ce (Fogelin 2008c). This new isolation is shown by the abandonment of chaitya halls and the stupas they contained, the only spaces at Buddhist monasteries open to the laity. The question remains, however, why Mahayana Buddhism and the image cult emerged between the first and fifth centuries ce—and what, if anything, monastic architecture had to do with it.


The number of well-preserved, archaeologically known Buddhist sites greatly increases between 100 and 600 ce. This is particularly true in the Gangetic Plain, where numerous stupas and viharas were either constructed or extensively refurbished during this time. Numerous new monasteries were also founded in the peninsula, most notably at Nagarjunakonda, an important southern center of Buddhism located in central Andhra Pradesh along the Krishna River. Many of the monasteries constructed or refurbished in this period were located near, though set apart from, major pilgrimage centers. The large pilgrimage stupas from this period and their association with monasteries will be discussed in Chapter 6. Here I focus on the development of Buddhist monastic architecture and the implica- tions of these developments on Buddhist monasticism.

Overall, Buddhist monasteries in the first millennium ce followed a fairly homogenous layout and ground plan throughout India—a large square enclosure, with monastic cells arrayed around the periphery (see Figures 5.2). Monastic cells opened onto a central courtyard, which often contained a central shrine. Finally, Buddha images were commonly placed within a monastic cell opposite the main entrance to the vihara. Within this common plan existed some significant variation, attributable to

Figure 5.2: Square viharas from around India (not to scale). Clockwise from top left, Nagarjunakonda (c. third–fifth centuries ce; after Sarkar 1966), Ajanta Cave 17 (c. fifth–sixth centuries ce; after Fergusson and Burgess 1880), Monastery 45 at Sanchi (c. 10th century ce; after Marshall and Foucher 1983), and Takht-I-Bahi (c. first century bce–seventh century ce; after Brown 1965)

regional traditions or the local topography of the building site (e.g., cliff faces or hilltops).

In the Northwest, Buddhist monasteries were divided into multiple courts, with some used as viharas and others containing large stupas or chaityas. Important sites include Takht-I-Bahi (discussed in Chapter 4), Manikwala, and Jamalgarhi (Mitra 1971). Near Taxila, monastic remains dating from this period are found at Kalawan, Mohra-Moradu, Pippala,

and Jaulian (Mitra 1971; Marshall [1951] 1975). In the Gangetic Plain, monasteries—or at least portions of monasteries—dating from 100 to 600 ce are found at almost every important Buddhist pilgrimage center. The monasteries at Sarnath and Nalanda are particularly well studied and










Nalanda, Rajgir

Bodh Gaya











Arabian Sea

Bay of Bengal

Figure 5.3: Archaeological sites discussed in Chapter 5

excavated (Mitra 1971). Most monasteries in the Gangetic Plain follow the typical square layout and were often two or more stories high.

Along the east coast of India, from Orissa down to Tamil Nadu, are the remains of numerous Buddhist monasteries (see Figure 5.3). While most of the Buddhist remains in Orissa date to the sixth century or later, some of the later monuments at Lalitagiri, for example, may overlay earlier constructions dating to the fourth century ce or, possibly, even earlier (Chauley 2000). Farther south, several monasteries along the coast of Andhra Pradesh, discussed in Chapter 4, continued to be used (e.g., Salihundam and Sankaram). Other Buddhist monasteries were founded further inland, particularly along the Krishna and Godavari Rivers. Among the most important of these were several monasteries located at the capitol of the Ikshvaku Dynasty, in what is now called Nagarjunakonda (Sarkar 1966). While named for the Mahayana scholar Nagarjuna, it is not possible to know if Nagarjuna was ever a resident.

What is known is that Nagarjunakonda was a major early center of Mahayana Buddhism in South India in the third and fourth centuries ce. During that time, roughly thirty Buddhist sites were constructed, includ- ing several monasteries of sects that are critical in the early development of Mahayana teachings. Extensive excavations at Nagarjunakonda pre- ceded the construction of the Nagarjuna Sagar Dam in the 1960s (Sarkar 1966; R. Subrahmanyam et al. 1975). Prior to the inundation of the area, several of the most prominent Buddhist remains were moved to a hilltop—now an island in the reservoir. For the most part, the monaster- ies at Nagarjunakonda follow the same format as monasteries through- out India, though in some cases apsidal chaityas were incorporated into the viharas (see Figure 5.2). Excavations at Nagarjunakonda also revealed numerous Buddha images, making the site one of the early centers of Buddha imagery in India.

On the west coast of the Peninsula, several rock-cut Buddhist mon- asteries were founded or expanded beginning in the late fifth and early sixth centuries ce. This followed a general hiatus in construction after the second century ce, corresponding to a period of weak political power in the Deccan. Among the rock-cut monasteries founded or expanded in this period are Ajanta, Aurangabad, Ellora, and Bagh (Dehejia 1972; Nagaraju 1981). At Ajanta, twenty-one newly carved viharas and two cha- ityas expanded on the nucleus of five earlier viharas and chaityas. While all of the new constructions were produced under the auspices of the Vakatakas, there is significant scholarly debate as to when new construc- tions began at Ajanta, and how long the new construction program lasted. Until recently, most scholars believed that the earliest new constructions

Figure 5.4: Cave 19 at Ajanta (c. fifth or sixth century ce)

Courtesy of the Digital South Asia Library and the American Institute of Indian Studies (Accession No. 61591).

at Ajanta began somewhere in the mid-fifth century, and lasted until the mid-sixth century. More recently, Spink (2006) has argued that all of caves were carved in roughly twenty years between 460 and 480 bce.

The new constructions at Ajanta began with the construction of an elaborate chaitya, Cave 19 (see Figure 5.4). In terms of the layout, Cave 19 is almost identical to the earlier chaityas at Ajanta—it is an apsidal hall, with a stupa at one end and a circumambulatory path along the periphery, demarcated by columns. The key difference between Cave 19 and earlier rock-cut chaityas in the Western Ghats is the far greater ornamenta- tion, and the carving of a Buddha image on the front of the drum and anda of the rock-cut stupa. Like the first and second century andas dis- cussed in Chapter 4, the anda in Cave 19 was raised above the plane of the drum, implying mass that is not present. A short time after the carving of Cave 19, a second new chaitya was carved at Ajanta (Cave 26). While more richly ornamented than Cave 19, it follows the same general lay- out and also combines a 3/5 anda and Buddha image on its central stupa.

Critically, these two rock-cut chaityas are among the last constructed in the Western Ghats. Thus, the slightly later sixth century ce monasteries at Aurangabad, Ellora, and Bagh omitted chaityas entirely.

The viharas at Ajanta follow the general format of Buddhist monaster- ies throughout India—square courtyards surrounded by monastic cells with a Buddha image installed in a large cell opposite the entrance. In contrast to Ajanta, at Bagh a stupa was placed in the perfumed chamber of a vihara (Cave 4) rather than a Buddha image (Mitra 1971). The placement of the stupa within the perfumed chamber at Bagh helps demonstrate the ritual equivalence of images and stupas for the sangha in the sixth century ce. The viharas at both Bagh and Ajanta are also heavily decorated with beautiful and elaborate frescoes. This tradition continued in the viharas at Aurangabad and Ellora, though in these later viharas the frescoes are not nearly as well preserved. Critically, however, there was a slight shift in the layout of some of the viharas at Aurangabad and Ellora. At Ajanta and Bagh, the cell containing the Buddha image or stupa was carved into the wall opposite the main entrance to the vihara. The same format was

Figure 5.5: Circumambulation in Aurangabad Cave 7 (c. sixth century ce; after Fergusson and Burgess 1880)

used in some of the viharas at Aurangabad and Ellora. However, in some of the viharas at Aurangabad and Ellora, paths were carved behind the cell containing the Buddha image, allowing the entire cell containing the Buddha to be circumambulated (see Figure 5.5). When viewed in light of the omission of chaityas at Aurangabad and Ellora—and the circumam- bulatory paths that chaityas contained—it appears that the sangha in the Western Ghats was attempting to preserve the ability to circumambulate the Buddha, but had shifted their ritual interests from stupas to Buddha images.

Perhaps the most surprising element of Buddhist monastic architecture in India between 100 and 600 ce is its homogeneity. Despite the known proliferation of Buddhist sects, despite the advent of Mahayana Buddhism, and despite the widespread fragmentation of political and economic pow- ers in India, Buddhist monasteries across India look pretty much the same—square courtyards with cells arrayed around the periphery. The only significant developments in Buddhist monasticism over the period were the gradual abandonment of stupas and chaityas at purely monas- tic centers (e.g., Aurangabad, Ellora), the creation of distinct and separate monastic spaces at pilgrimage centers (e.g., Sarnath) and the progressive inclusion of Buddha images within the viharas across India. Though these architectural and artistic developments were seemingly slight, they had profound effects on the way in which the sangha interacted with both the Buddha and the laity.


In the traditional histories of Indian monastic Buddhism, the period between 100 and 600 ce is viewed as a period when the Buddhist sangha progressively became more enmeshed with the laity, allowing lay prac- tices such as ritual directed at stupas and Buddha images to percolate through the sangha. In contrast, the architecture of Buddhist monasteries from this period suggests almost the complete opposite, that the sangha progressively withdrew into their monasteries in this period, severing their ties to the non-elite laity. While apparent in the design and layout of monasteries in Andhra Pradesh and the Gangetic Plain, the progres- sive withdrawal of the sangha from regular interaction with the laity is most clearly shown in the developments in monastic architecture in the Western Ghats.

Until the fifth century ce, rock-cut monasteries in the Western Ghats consisted of one or more large chaityas surrounded by several viharas. The

Figure 5.6: Bhaja (c. first century bce: top) and Aurangabad (c. sixth century ce)

Photo of Bhaja courtesy of the Digital South Asia Library and the American Institute of Indian Studies (Accession No. 80006).

viharas served as monastic dormitories, and the chaityas as both a public and monastic worship space. The placement of chaityas within Buddhist monasteries associated the primary symbol of the Buddha, the stupa, with the sangha—spatially asserting monastic control over the ritu- als performed in front of the stupa. In contrast to the often haphazard placement and low profile of viharas, chaityas had large, elaborately carved

façades that would have been visible at great distances (see Figure 5.6). Using Moore’s (1996) criteria of size, permanence, visibility, and central- ity, prior to the sixth century ce, Buddhist monasteries is the Western Ghats were legitimations that proclaimed the importance of the sangha to the Buddhist laity.

This pattern began to change in the Western Ghats at Ajanta with the carving of Buddha images on stupas within chaityas 19 and 26 in the mid- to late fifth century ce. After this, rock-cut chaityas of the ear- lier form were no longer constructed in western India. Instead, Buddha images were placed within viharas, in monastic cells referred to as per- fumed chambers. The symbolism here was fairly straightforward. The image of the Buddha was installed within a monastic cell, symbolically asserting that the Buddha was a resident of the monastery. In terms of legitimizing monastic authority, this change in the location of the pri- mary ritual focus had profound implications. While the images were still spatially associated with the sangha, perhaps even more so than before, the visibility and accessibility of the images to the general public were dramatically reduced. With the abandonment of chaityas, the façades of the viharas were not elaborated to any great degree (see Figure 5.6). The visibility of the monasteries in the Western Ghats, therefore, was sub- stantially less than it had been in the previous period. Unlike the stupas and chaityas of the earlier period, Buddha images were now seques- tered within the viharas, with no outside indicator of their presence. It would seem that by this later period the Buddhist sangha was no longer attempting to assert authority over the laity, or at least did so to a much lesser degree.

In Andhra Pradesh after the second century bce, monastic architecture also facilitated the withdrawal of the sangha from regular contact with the laity. Prior to the third century ce, monasteries in Andhra Pradesh (e.g., Thotlakonda and Bavikonda) had more informal layouts, with gaps between the individual viharas that collectively defined the courtyard in which the sangha lived and received religious instruction (see Figure 5.7). Likewise, elaborate public worship spaces were located adjacent to the viharas. While separated from the viharas, this separation was often only partial, with views into the courtyard through railings or other more per- meable barriers. In contrast, the viharas at Nagarjunakonda in the third and fourth centuries were designed to be far more impermeable—completely enclosed square courtyards with a single gate (see Figure 5.7). Just as with the sangha in the Western Ghats in the sixth century ce, in the third and fourth centuries ce the sangha in Andhra Pradesh erected physical bar- riers between themselves and the broader world. While fostering a more

Figure 5.7: Nagarjunakonda (c. third–fifth centuries ce; after Sarkar 1966) and Thotlakonda (c. second century bce–third century ce) monasteries in Andhra Pradesh

symbolic separation between the sangha and the laity, the real physical materiality of these walls should not be minimized. The high walls and single gate allowed the sangha at Nagarjunakonda to significantly increase their control of who could—and who could not—enter their monasteries or even witness their rituals from afar.

In the Gangetic Plain, it is not possible to document a shift from more engaged to more secluded monasteries for the simple reason that the earli- est archaeologically known monasteries date only to this period. What can be said, however, is that the earliest archaeologically known monasteries in the Gangetic Plain, dating from the third or fourth centuries ce, all fol- low the same general layout as those found at Nagarjunakonda—square enclosures with a single entrance. Thus, from the earliest period for which the layout of Buddhist monasteries is known in the Gangetic Plain, the monasteries were designed to facilitate monastic isolation rather than engagement with the laity. There is, however, one argument against this line of thinking; Buddhist monasteries in the Gangetic Plain were typically found at the major pilgrimage sites frequented by the laity. This could be interpreted as demonstrating that the sangha was more actively engaged with the laity, leading worship and otherwise serving as the caretakers of major pilgrimage sites.

There is clearly some element of truth to the idea that if the sangha

was not interested in engaging with the laity, they would not have erected their monasteries at pilgrimage sites where the laity congregated. But there are problems with this interpretation as well. First, this inter- pretation is based, to some degree, on the traditional understanding that

the sangha was initially uninterested in stupa ritual and sought a life of seclusion. As discussed in previous chapters, this just isn’t the case. From the earliest periods in which we have direct archaeological and epigraphic evidence, the sangha was actively involved in the stupa cult. It is just as likely that the sangha established monasteries at the major pilgrimage sites in the Gangetic Plain for the simple reason that they wanted to wor- ship at these sites, that they desired to be close to major memorials of the founder of their religion. In this light, the location of monasteries at the key pilgrimage sites could have occurred despite the presence of lay worshipers rather than in service to the lay worshipers. Unable to exclude the rabble from the major pilgrimage sites, the sangha did the next best thing—they secluded themselves as best they could in private courtyards behind high walls. Similarly, the installation of Buddha images by Bala and Buddhamitra at Sarnath, Shravasti, and Kausambi could have been for the simple reason that they, and their followers, engaged in a form of devotion that employed Buddha images. That is, if the image cult first began among the sangha, the installation of images at key Buddhist sites was not intended for the benefit of lay pilgrims, but rather for the benefit of the sangha themselves.

My point here is not to replace the image of entirely beneficent sangha

with an image of entirely venal sangha. Most likely, the behavior of the sangha was somewhere between these two extremes. My point is that in many of the traditional histories the beneficence of the sangha is assumed, serving as both the start and endpoint of the analysis. That is, because of their beneficence, the sangha established monasteries at pilgrimage sites to facilitate the worship of the laity—with the evidence of this benefi- cence being that the sangha established monasteries at pilgrimage sites to facilitate the worship of the laity. This same beneficence also explains how well-meaning members of the sangha gradually incorporated the ritual practices of the laity. In contrast, when looking at the layout of monaster- ies across India in this period, it seems the sangha was progressively sepa- rating themselves from the laity after a long period of active engagement. This new seclusion can only be explained by a more balanced view of the beneficence of the sangha—that the sangha was neither wholly beneficent nor wholly venal.

As discussed in previous chapters, from the start the Buddhist sangha balanced their desires for isolation with the need to engage the laity for economic support. Thus, in monasteries from the first century bce through the second century ce, the sangha created both private and public spaces, viharas and chaityas. Within the chaityas Western Ghats, the sangha even manipulated the form and proportions of their stupas as a form of

legitimation. By the third century ce in the Gangetic Plain and Andhra Pradesh, and the fifth century in the Western Ghats, the sangha mostly abandoned their outreach to the laity. The reason for this newfound iso- lation was actually fairly simple—for the first time the sangha could afford to be isolated. Where previous generations required regular donations of money, food, and cloth to maintain themselves, beginning in the third century ce, the sangha was increasingly receiving gifts of land and the labor of entire villages. With the shift to the more agrarian, land-based system of giving by the Guptas and Vakatakas, Buddhist monasteries became major landholders. The sangha no longer needed the pious dona- tions of the Buddhist laity to support themselves, rather the sangha could simply collect taxes from the people working their lands, whatever reli- gious sect those people favored. This, in turn, created the economic con- ditions that allowed the sangha to finally practice the seclusion they had long desired. The previous need to balance seclusion and engagement was no longer necessary. And so the sangha separated themselves from regular contact with the Buddhist laity, creating new conceptions of Buddhism in the process.


Between the second and sixth centuries ce, monastic seclusion was rein- forced by the sangha shifting the ritual focus from stupas within apsidal chaityas to Buddha images within viharas. At the same time, some portions of the sangha began practicing Mahayana Buddhism. These changes can best be explained through the combined insights of materiality and semi- otics (Fogelin 2012). From a material practice perspective, the abandon- ment of stupas within chaityas and the carving of images within viharas materialized monastic isolation. The placement of Buddha images within viharas also materialized a new relationship between the sangha and the foci of their rituals. Buddha images were a return to the iconic, emotion- ally immediate, worship of the Buddha lost after centuries of manipula- tion to the form of monastic stupas. Where previously monastic stupas had been constructed in ways that the sangha felt would assert their authority over the laity and speed the flow of donations, Buddha images were now intended to be meaningful only for the sangha themselves (Fogelin 2008c). Without the need to assert authority over the laity, the sangha was free to refashion the foci of the their rituals in ways that concorded with their new, secluded lifestyle. In this light, the creation of Buddha images and the development of Mahayana Buddhism were part of the construction of

a new monastic identity. Materiality, however, is somewhat mute on what the new images signified to those who erected them. In contrast, a semi- otic perspective can explain why monastic stupas in apsidal chaityas were no longer meaningful to the Buddhist sangha, and why Buddha images and Mahayana Buddhism were more satisfying in the context of developing monastic isolation in the beginning of the first millennium ce.

In the second through sixth centuries ce, the sangha lived in a world created by their monastic ancestors. The primary focus of their ritual actions—stupas—had, through physical manipulation of their predeces- sors, become a symbol of Buddhism. Monastic stupas lacked the emotional immediacy of the pilgrimage stupas and relics that continued to receive the devotion of the laity. Successive generations of the sangha, material- izing their power and authority in the stupas of apsidal chaityas, had cre- ated symbols that emphasized thirdness and promoted an increasingly intellectual and abstract relationship between the sangha and Buddhism’s founder. The signs of monastic Buddhism had become, by semiotic defini- tion, conventional.

Between the second and sixth centuries ce, the sangha progressively cre- ated a more satisfying link between themselves and the Buddha through the creation of Buddha images and the adoption of Mahayana Buddhism. In semiotic terms, Buddha images were icons of the Buddha and symbols of Buddhism. Unlike the purely symbolic monastic stupas they replaced, Buddha images were multivalent signs. While Buddha images continued to symbolically signify Buddhism in an abstract sense (thirdness), their iconicity simultaneously created a sense of firstness and emotional imme- diacy for the members of the sangha who viewed them. In the late fifth century ce, Buddha images helped signify that the Buddha was immedi- ate and active in the emotional lives of the sangha. As such, the theology of Mahayana Buddhism and the construction of Buddha images can be understood as a revitalization movement—defined here as a conscious effort to construct a more satisfying culture (Wallace 1956:215). Freed from dependence on the laity, the sangha refashioned their monasteries to both signify and create this new material and spiritual reality.

Nothing in this analysis should be taken to suggest that Mahayana Buddhism and Buddha images necessarily arose directly from the actions of earlier monks. Rather, the relationship between monastic seclusions, Buddha images, Mahayana Buddhism, and the manipulations of stupas by previous generations of monks are best understood using Weber’s con- cept of “elective affinity” (see Chapter 2). That is, many other potentiali- ties were possible. Just as earlier monks had agency when attenuating stupas, later monks had agency when creating Mahayana Buddhism and

Buddha images. For example, in the second through sixth centuries ce the sangha could have chosen to abandon their monasteries and rejoin lay Buddhists in their devotions at the pilgrimage centers. In fact, at Sanchi, Sarnath, and numerous other pilgrimage centers in the Gangetic Plain, it appears that the sangha did just that—establishing or renovating mon- asteries near iconic stupas marking the key points in the Buddha’s life. Alternatively, the sangha could have, but did not, hollowed out their stupas and placed relics within them. Many other actions were possible as well. By saying that physical manipulations of stupas created the preconditions that help explain the origin of Mahayana Buddhism and Buddha images, I am not arguing that these manipulations caused, in any direct way, Buddha images. The Buddhist sangha in the second through sixth centu- ries ce had their own reasons for sculpting Buddha images and refash- ioning Buddhist theology in the particular ways they did. The important point here is that we cannot let semiotic categories over-determine our analyses.

Following the perspectives of practice theory and materiality, people are not automatons enacting structural rules. People are creative actors engaging with and altering the material world in which they find them- selves for specific purposes. But it would be mistake to ignore the weight of signs inherited from the past. Signs have real meaning and real impact on the people who use and experience them. Rather than relying on a single perspective, archaeologists should combine the insights of materiality and semiotics in their research. Here I have employed materiality and semiot- ics to examine the metamorphosis of Buddhist ritual foci from 500 bce through 500 ce, from ancestral earthen stupas to Buddha images carved in the viharas of the Western Ghats (see Figure 5.8). Individually, each step in the metamorphosis seems explicable from the perspective of material- ity alone. Early Buddhists enlarged earthen stupas and altered the medium to stone, brick, and stucco in order to assert the power and importance of the relics interred within. Later, the sangha constructed attenuated and implied mass stupas to establish their authority over the laity. Later still, the sangha abandoned their stupas in apsidal chaityas in favor of images within viharas in order to foster monastic seclusion. While the motivations of the actors at each stage are interesting and important, by themselves they do not fully account for the long-term metamorphosis of Buddhist ritual foci. Attenuated stupas are inexplicable without the knowledge that they were icons of ancestral stupas. The emotional immediacy of Buddha images is only interesting in relation to the intellectual detachment of the stupas that immediately preceded them. Without semiotic theory, the changes in significance due to the specific manipulations of stupas and the

No Archaeologically Known Examples

Ancestral Stupa

c. 500 BCE

Icon (relic) of the Buddha Index (anda) of the relic

Pilgrimage Stupa

c. 3rd century BCE–Present

Icon (relic) of the Buddha Icon of ancestral stupas Index (anda) of the relic Symbol of Buddhism

Attenuated Stupa

c. 1st century BCE

Icon (relic) of the Buddha Less iconic of ancestral stupas Index (mound) of the relic Symbol of Buddhism

Implied Mass Stupa

c. 1st–5th centuries CE

Symbol of ancestral stupas Symbol of Buddhism

Buddha Image

c. 2nd century CE–Present

Icon of Buddha Symbol of Buddhism

Figure 5.8: The semiotic transformation of Buddhist monastic ritual foci

creation of images by ancient Buddhists would be difficult to ascertain. People are actors and interpretants, objects are of the world and of the mind, and signs have real impacts on people’s actions. A full accounting of archaeological pasts must employ multiple theories that together can account for the dialectic between signs inherited from the past and the actions of agents that alter those signs for the future. Archaeologists must also place these accounts within the specific histories of the people they study—in the case of the Buddhist sangha between the second and sixth centuries ce, within a context of the invention of a new, ascetic ideal.


Between the second and sixth centuries ce, several new traditions emerged within Buddhism, including Mahayana Buddhism, Buddha images, and an emphasis on monastic seclusion. These new traditions coincide with the period in which the bulk of early Buddhist textual sources, whether Hinayana or Mahayana, were recorded and collated. Mahayana texts were initially the product of geographically and politically marginal members of the sangha in the peripheries of India. In reaction to the polemic of early champions of Mahayana like Nagarjuna, the more traditional monks in the Buddhist heartland began recording and collating their own texts. As such, the tradition of Buddhist scholasticism—the study, commentary, and debate over texts—also began in this period. All of these new tradi- tions fed into a complex new conception of Buddhist asceticism—a new- found celebration of the ascetic tradition, however rarely practiced, that informed the sangha’s actions and identities thereafter.

In Chapter 1, I discussed the concept of the invention of tradition or memory. As developed by Hobsbawm and Ranger (1983; see also Connerton 1989), the invention of tradition refers to a process whereby real or imagined histories are promoted to legitimate practices in the present—to provide modern practices the aura of historical precedent. Thus, as the British mon- archy lost political power at the end of the nineteenth century, they created new rituals and traditions that were intentionally meant to seem timeless and unchanging. Similarly, in the late eighteenth and early nineteenth cen- turies, the new government of the United States constructed their capitol in a neoclassical style to recall both European capitols and the architecture of classical Greece and Rome. In both cases, traditions were invented as legiti- mations of political power in the present. What I did not discuss in Chapter 1 was how, precisely, traditions are invented and instilled in a population.

Anticipating the insights of Humphrey and Laidlaw (1994) and Bell (1992, 1997) on ritual practice (see Chapter 2), Hobsbawm argued that the invention of tradition was accomplished through a process of ritualization.

Inventing traditions, it is assumed here, is essentially a process of formaliza- tion and ritualization, characterized by reference to the past, if only by impos- ing repetition. (Hobsbawm 1983:4)

Hobsbawm’s use of the terms “ritualization,” “formalism,” and “repeti- tion” bear striking similarity to many of the characteristics of ritual and ritual-like behaviors noted by Catherine Bell (1997): formalism, tradition- alism, invariance, rule-governance, sacral symbolism, and performance. Formalism and invariance reinforce the invention of tradition through the repetition of restricted actions. Traditionalism refers to the use archaic or anachronistic elements or, in essence, older traditions. Finally, performance provides the necessary public venue in which an inven- tion of tradition can be communicated to, instilled in, and created for its intended audience. Critically, this audience need not be large; traditions can be practiced by small secret societies just as much as they are by entire nations or world religions. Like Hobsbawm and Ranger, Bell also argued that ritual is more a process than an event. Following this insight and the similarities between Bell’s characteristics of ritual and Hobsbawm’s pre- sentation of the invention of tradition, memory can also be understood as a process. That is, ordinary or invented traditions gain greater mean- ing and significance as they become ritualized and formalized. Just as in the study of ritual, these processes often leave clear material traces for archaeological study. The archaeological evidence of Buddhism between the second and sixth centuries ce provides numerous traces that suggest the invention of a new tradition of Buddhist asceticism, though this new tradition was more ideal than practice.

Ascetic Ideal

The emergence of a new tradition of asceticism is shown in diverse strands of textual, iconographic, and archaeological evidence from the second through sixth centuries ce. The rhetoric of Buddhist texts from this period, particularly the early Mahayana texts, show all of the hallmarks of the invention of tradition. Rather than arguing for a new tradition or a new set of Buddhist practices, the authors of early Mahayana sutras pre- sented their work as the original teachings of the Buddha himself. The

texts themselves were said to have been hidden for centuries—in caves, on the ocean floor, or wherever—only to be discovered by Mahayana schol- ars. In this sense, early Mahayana texts relied on “general links to a vague mythological antiquity” to buttress their claims to historical authenticity (Van Dyke and Alcock 2003:3). Similarly, the repeated calls to return to the life of forest found throughout early Mahayana texts rely on invocations of a past, real, or imagined, in which the Buddha and the early sangha lived a more truly ascetic lifestyle.

As discussed earlier, the advent of Buddha images can be understood, in a semiotic sense, as a return to the iconic, emotionally immediate wor- ship of the Buddha by the sangha. Critically, between the second and sixth centuries, the tradition of Buddha images spread from Gandhara and Mathura throughout all of India. If anything, Buddha images appear ear- lier in the mainstream monasteries and pilgrimage sites in the Gangetic Plain than they do in the peripheries where Mahayana Buddhists may have originated. The earliest images at Sarnath, Kausambi, and Shravasti date to the second century ce, while the images at Nagarjunakonda date to the third or fourth century ce and the images at Ajanta date to the fifth

Figure 5.9: Relief at Bharhut (c. first or second century bce) and Mathura style Buddha image (c. second century ce)

Courtesy of the Digital South Asia Library and the American Institute of Indian Studies (Accession Nos. 34333, 44506).

century ce. Thus, if anything, the adoption of Buddha images was initiated by mainstream Buddhist monasteries prior to the adoption of images by Mahayana Buddhists. However, like the rhetoric of the Mahayana texts, the form of the Buddha images favored by mainstream Buddhist monas- teries emphasized the ascetic nature of the Buddha. Prior to the advent of Buddha images, Buddhist iconography commonly depicted crowds of people surrounding an empty throne, footprints, or other markers of the Buddha (see Figure 5.9). In contrast, with the advent of Buddha images in the second century ce, the Buddha was commonly depicted as a solitary, or near solitary, figure. As such, the incorporation of Buddha images in mainstream Buddhist monasteries served as an iconographic invention of tradition, similar in most ways to the textual invention of tradition by Mahayana Buddhists. Thus, whether fringe or mainstream, whether adherents to early Buddhism or Mahayana Buddhism, between the second and sixth centuries ce, the sangha sought a return to a long-standing tradi- tion of asceticism, legitimized through the presentation, either textually or iconographically, of the Buddha as the prototypical ascetic—whether or not that tradition of asceticism ever really existed.

Scholastic Reality

As discussed in Chapters 3 and 4, there is no archaeological, epigraphic, or iconographic evidence for early Buddhist solitary asceticism. On the con- trary, from the earliest period in which there is direct evidence—between the third and first century bce—the sangha lived in well-established monasteries and had regular contact with the laity. While it is possible that a small number of ascetics might have lived in the forests and left no material traces, there is no archaeological, epigraphic, or iconographic evidence of solitary asceticism dating from those early periods. In fact, the only evidence of early Buddhist asceticism comes from early Buddhist texts—texts that were authored, recorded, and transcribed in the begin- ning of the first millennium ce. If nothing else, the complete lack of mate- rial evidence for early Buddhist asceticism should give modern scholars pause. However, even if modern scholars, relying on these later texts, per- sist in believing that early Buddhism was an ascetic tradition, the rhetoric of the early Mahayana texts demonstrate that the ascetic tradition had been abandoned by the beginning of the first millennium ce. The sangha could not return to the forest unless earlier generations had already left it. Whether a tradition of Buddhist asceticism existed prior to the first mil- lennium ce or not, the rhetoric and repeated calls to return to the forest

of early Buddhist texts signify an invention of tradition. That is, the calls to asceticism were a response to the situation in which the sangha found themselves in the second through sixth centuries ce, which took the form of reference to remembered or quasi-mythological history. While most clearly apparent in the early Mahayana texts, the same tradition of asceti- cism developed among mainstream members of the sangha, as shown in the creation of images of a solitary, ascetic Buddha at pilgrimage sites and monasteries throughout India. There is, however, one more wrinkle in this newfound interest in asceticism—there is no more archaeological evidence of forest monks in the second through sixth centuries ce than there is before the second century ce. The tradition of asceticism was not invented in the second through sixth centuries ce, but rather, at best, a mostly unrealized ascetic ideal was invented in the second through sixth centuries ce.

In the early fifth century ce, the Chinese pilgrim Faxian traveled to India along the Silk Road from China (Rongxi 2002). Faxian entered India in the Northwest, traveled through the Gangetic Plain, and finally left India for Sri Lanka by boat from Tamluk in the Gangetic Delta. Upon returning to China with numerous sacred texts, he wrote a detailed account of his travels. Despite the celebration of asceticism in early Buddhist texts, in Faxian’s account of his travels in India he never men- tions solitary forest monks or any other solitary ascetics, except in his- torical contexts. For example, while Faxian recorded his visit to several isolated caves near Rajgir (Rongxi 2002), his account only states that the caves were used by the Buddha, Ananda, and various other long-passed ascetics. Far from being locations of continued asceticism, by the time of Faxian’s visit, the caves near Rajgir were the focus of pilgrimage to honor the long-dead ascetics who were said to have lived there in the past. In another case, Faxian credited the founding of a monastery to a devout monk who had cleared and restored one of the eight original stupas that contained the cremated remains of the Buddha (Rongxi 2002). In some sense, this account fits neatly within the idea of an ascetic monk aban- doning life in a monastery for a purer life in the forest. Critically, how- ever, Faxian did not meet the founding ascetic. Rather, Faxian visited the established monastery said to be founded by the long-dead ascetic. In con- trast to the celebration of solitary ascetic monks in traditional Buddhist histories, every living member of the sangha mentioned by Faxian lived in an established monastery.

In his account, Faxian described numerous Buddhist monasteries.

Though he consistently overstated the physical size and resident popu- lation of the monasteries he encountered, the large, square monasteries

described by Faxian concord almost perfectly with the mostly homog- enous layout of Buddhist monasteries known archaeologically. Through the omission of public worship spaces within their monasteries and the imposing exterior walls that surrounded them, the sangha isolated them- selves from the laity. Critically, this isolation was not individual, but com- munal. This was just as true for the more mainstream monasteries of the Gangetic Plain as it was at Ajanta and other monasteries in South India. All of this suggests that between the second and sixth centuries ce, many of the Buddhist sangha—both Mahayana Buddhists and mainstream Buddhists—struck a new balance between their contradictory individual and communal desires. With the wealth derived from their new landhold- ings, the sangha could withdraw from regular contact with the laity, but rather than adopting solitary asceticism, the sangha began practicing com- munal asceticism in the form of scholasticism. The sangha devoted them- selves to the study and creation of Buddhist texts. This, in part, explains why Buddhist textual sources become common only in the first half of the first millennium ce. What is interesting, however, is the discord between what the sangha’s texts advocated and how the sangha lived their lives.

While practicing collective asceticism, the sangha invented, debated,

and elaborated on the new ideal of solitary asceticism. Their texts cel- ebrated pratyekabuddhas (solitary Buddhas), while the sangha lived the ordered life of shravakas (disciples). By balancing the practice of collec- tive asceticism with romantic accounts of the solitary ascetics of old, the sangha fashioned a new way to ameliorate the contradictions between the individual and communal desires of their membership. First, the new ascetic ideal legitimized the sangha’s isolation from the laity, at least in the view of the sangha itself. Second, reading or hearing about legend- ary pratyekabuddhas satisfied some of the ascetic desires of the sangha. Finally, this new compromise between individualism and communalism was legitimized by the placement of Buddha images within the mon- asteries, within perfumed chambers modeled on the cells in which the sangha themselves lived. By placing the Buddha, the prototypical solitary ascetic, within their monastery, the sangha asserted that the Buddha himself approved of this new compromise. Thus, in the second through sixth centuries ce, through a process of ritualization, formalization, rep- etition, and calls to remembered or invented traditions, the sangha cre- ated a new form of collective asceticism while creating a romanticized ideal of solitary asceticism.

According to the standard histories of Buddhism, between the second and sixth centuries ce, the previously ascetic sangha was slowly domesti- cated through regular contact with the laity. In contrast, by looking at the

archaeological, iconographic, epigraphic, and textual evidence together, something very much like the opposite seems to occur. That is, a long domesticated sangha withdrew into their monasteries, severing contact with all but the wealthiest laity, as they invented a new ascetic ideal, an ideal only partially realized in their collective rituals. The sangha con- tinued to live domesticated lives within monasteries—as they had for centuries—just more isolated from the laity than had previously been possible.

But with the new ascetic ideal, something new began to happen. Beginning in the fifth or sixth century ce—possibly for the first time in the history of Indian Buddhism—some small portion of the sangha began to take the solitary ascetic ideal seriously, and they began abandoning monasteries for life in the forest. Thus, the history of the Buddhist asceti- cism is not a process of ascetics slowly being domesticated, but rather a long-domesticated sangha in which some small percentage went feral.




Lay Buddhism and Religious Syncretism in the First Millennium ce


n contrast to the marked historical developments in monastic Buddhism in the first millennium ce, lay Buddhism was more conservative. The laity continued their devotions in pretty much the same way, at pilgrim- age sites that remained more or less the same. Some new pilgrimage sites were built, some were abandoned, and Buddha images began to be erected at pilgrimage sites but, for the most part, the laity continued to circumam- bulate large pilgrimage stupas and to practice more festive communal ritu- als in the courtyards surrounding them. This is not to say that there were no changes in popular Buddhism in the first millennium ce. There were. Those changes that did occur, however, had more to do with the number of lay Buddhists worshiping at Buddhist sites, not the manner of their worship. As the sangha progressively withdrew from regular contact with the laity in the first millennium ce, the laity returned the favor, shifting their devotions to rival religious orders. This shift was facilitated by the incorporation of Buddhist architectural, ritual, and doctrinal elements by rival religious sects. As the practice of monastic Buddhism changed, the laity increasingly found the rites and rituals of nascent Hinduism and other sects more satisfying, if only because the ritual practices of these new sects seemed so familiar. As a result, numerous Buddhist pilgrimage sites across India were abandoned, with many sites taken over by Hindus, Jains, and others. By the end of the first millennium ce the laity had abandoned Buddhism throughout most of India. The most notable excep- tion to this pattern of abandonment was in the Northeast, Orissa, and the Gangetic Plain, where lay Buddhists continued to frequent pilgrimage

( 180 )

centers with at least some support of the sangha. Since the process of decline of Buddhism in the Gangetic Plain, Orissa, and the Northeast is markedly different from the rest of India, for the most part, I will reserve discussions of those regions until Chapter 7.


During the first millennium ce, many—but not all—of the large Buddhist pilgrimage sites first constructed in the late first millennium bce con- tinued to be a focus of lay devotions (e.g., Sanchi, Sarnath, Bodh Gaya, Amaravati; see Figure 6.1). While often rebuilt, expanded, or otherwise modified, most Buddhist pilgrimage sites preserved the same basic lay- out and organization throughout the period of their use—usually a large courtyard with a central stupa and circumambulatory path. While several new pilgrimage centers were constructed and some older ones were aban- doned, for the most part, the newer centers followed the same general layout as the earlier pilgrimage sites. The failure of some pilgrimage cen- ters and the creation of new centers was not the product of their design, but rather the waxing and waning popularity of lay Buddhism in different regions in India. None of this is intended to suggest that pilgrimage sites were completely unchanging or that the ritual and architectural develop- ment in monastic Buddhism did not extend to pilgrimage sites. They did, but only to a limited degree. Lay Buddhism did change over the first mil- lennium ce, just less dramatically than monastic Buddhism did.

Buddhism originated in the Gangetic Plain, with major pilgrimage cen- ters established at places associated with the life of the Buddha—though, with the exception of Lumbini (Coningham et al. 2013), the archaeologi- cal form of these early pilgrimage centers is unknown due to later con- structions and modifications. Once established in the heartland, popular Buddhism rapidly spread southward, with the creation of major Buddhist pilgrimage sites at Bharhut, Sanchi in the second century bce, Amaravati in the first century bce, and the Kanaganahalli Stupa in Karnataka between the first century bce and the third century ce. The expansion of popular Buddhism into the Northwest is shown by the construction of the Dharmarajika monastery at Taxila between the third and first centuries bce, as well by the construction of numerous other stupas in the early to mid-first millennium ce. In Orissa and the Northeast, the spread is more complicated.

Shingardar Takht-I-Bahi








Bharhut Sanchi


Bodh Gaya






Ratnagiri, Udayagiri, Lalitagiri



Arabian Sea


Bay of Bengal

Figure 6.1: Archaeological sites discussed in Chapter 6

While there is clear evidence and mention of Buddhism in Ashokan inscriptions in Orissa, and some potentially Mauryan period finds have recently been reported (e.g. Lalitagiri [Chauley 2000]), it appears that a major fluorescence of monastic Buddhism occurred in the mid-first mil- lennium ce. This is shown by the founding or enlargement of several monasteries, including Lalitagiri, Udayagiri (Nigam 2000), and Ratnagiri

(Mitra 1981, 1983). All contained several viharas following the typical square layout, numerous Buddha images, and large open-air stupas that were likely intended for pilgrimage. The Northeast follows the same pat- tern, with clear evidence of Buddhism from the earliest periods, but sig- nificant expansion and enlargement of Buddhist monastic and pilgrimage centers in the mid- to late first millennium ce, including Somapura (Mitra 1971) and Vikramasila (Dutt 1988; Prasad 1987). These late Buddhist centers will be discussed at length in Chapter 7.

Ritual in pilgrimage centers typically consisted of festive communal ritual in the courtyards and circumambulation in the paths directly cir- cling the central anda. In some cases, particularly at the pilgrimage sites in the Gangetic Plain that memorialized specific moments in the life of the Buddha, the central foci of the courtyards were not stupas. At Bodh Gaya, for example, devotees circumambulated the Bodhi tree and the Gupta period temple that marked the location of the Buddha’s enlightenment. At Lumbini, the foci was a tree, and later a stone, that marked the place of the Buddha’s birth (Coningham et al. 2013). Whatever the focus, the patterns of ritual at pilgrimage sites balanced the individual and communal desires of lay devotees. In comparison to contemporary monastic centers, ritual at pilgrimage centers tended to favor individual ritual in the circumam- bulatory paths, and more egalitarian communal ritual in the courtyards surrounding them. Critically, in contrast to the ever-changing design of Buddhist monasteries, the stability in the design and layout of pilgrimage centers throughout the first millennium ce belies stability in the balance between the individual and communal desires of Buddhist pilgrims in the first millennium ce. None of this should be taken to imply that there were not attempts by the sangha to alter pilgrimage centers to foster their own power, only that monastic legitimations at pilgrimage centers seem to have mostly failed.

In Chapter 4, I discussed the design tricks that the sangha used to make their stupas appear taller and more massive than they actually were—initially by attenuating the proportions of andas and later by implying greater mass by raising the midpoint of andas above the drum on which they rested. While not common, in the early and mid-first mil- lennium ce, the same visual tricks were applied to the design of the andas at some major pilgrimage sites. Perhaps the clearest example of this is at Sarnath, the location of the Buddha’s first sermon. As discussed in earlier chapters, Sarnath was a Buddhist pilgrimage center from at least Mauryan times, as shown by a column with Ashoka’s Schism Edict. While Ashoka is credited with numerous edifices at Sarnath, the structures available for archaeological study date mostly to Gupta times. Sarnath was also an

early and important location for Buddha images. These images were placed throughout the new constructions and along a processional path used by pilgrims. The main stupa Sarnath—the Dharmarajika stupa (not to be con- fused with a stupa of the same name at Taxila)—was extensively mined for its bricks in the late eighteenth century. While it appears similar in most respects to other pilgrimage stupas, little more can be said of it.

The largest surviving stupa at Sarnath is the Dhamek stupa (43.5 m tall and 23.3 m diameter; Figure 6.2). In its present form, the Dhamek stupa dates to Gupta times, though the existing edifice may enclose earlier con- structions (Mitra 1971). The Dhamek stupa is a striking departure from the typical form of large, open-air stupas in India. Rather than a low hemi- sphere, the anda of the Dhamek stupa is more like a giant column. In a sense, the Dhamek stupa at Sarnath was an attenuated, open-air stupa. That is, verticality was emphasized to such a degree that the anda lost all iconic resemblance to the earthen mound it was intended to signify. The manipulations of the anda proportions were so extreme that the Dhamek

Figure 6.2: Dhamek stupa at Sarnath (c. fifth or sixth century ce)

Courtesy of the Digital South Asia Library and the American Institute of Indian Studies (Accession No. 25261).

stupa barely resembles a stupa at all. Just like the attenuated stupas dis- cussed in Chapter 4, it is possible that the form of the Dhamek stupa was intended to help assert authority over those who viewed it. The same can be said of the several similar open-air stupas found in Northwest India, including the Shingardar stupa in the Swat Valley.

By the standards of Moore (1996), the Dhamek stupa and other attenu- ated stupas were very large, highly visible, associated with a location of special religious importance, and made of very durable materials. By any standard, they can be understood as legitimations. What seems less likely, however, is that they were successful. Given how few attenuated open-air stupas were constructed in India, that even when constructed they were of secondary importance to more traditional stupas at the same site, and that stupas with traditional proportions continued to be erected across India, it is likely that attenuated open-air stupas mostly failed as legitimations.

In contrast to the relative failure of attenuated open-air stupas among the laity, the incorporation of Buddha images at pilgrimage sites may have been more successful. In part this success was due to images’ unobtrusive- ness. Across India, Buddha images were added to processional paths lead- ing to pilgrimage stupas (e.g., Sarnath), placed in subsidiary chapels that surrounded pilgrimage stupas (e.g., Dharmarajika stupa at Taxila, Takht-I- Bahi), carved into the gates, railings, and Ayaka platforms of pilgrimage stupas (e.g., Amaravati). In some cases, like the Mahabodhi temple at Bodh Gaya and Temple 31 at Sanchi, small temples centered on a Buddha image were constructed, but these temples served as an addition to the repertoire of ritual spaces available at pilgrimage sites, not as replace- ments for the traditional ritual foci of these sites. Lavish decorations at pilgrimage sites were nothing new, with elaborate carvings in stone and stucco common from the earliest archaeologically known pilgrimage sites in the second century bce (e.g., Sanchi and Barhut). With the end of the taboo on depicting the Buddha in the early to mid-first millennium ce, Buddha images were added at pilgrimage complexes, but not in a way that would have forced the laity to alter their ritual practices. Those who found Buddha images ritually satisfying could include them in their devotions. Those who did not could simply pass them by and perform circumambula- tion and group rituals just as they always had.

As discussed in Chapter 5, the tradition of Buddha images originated

within the sangha at a time when the sangha was increasingly constructing fortress-like viharas at the largest Buddhist pilgrimage sites in India. By their design, these new monasteries were not intended to facilitate greater interaction with the laity, but rather to allow the sangha to engage in their own rituals at the most holy sites of Buddhism while minimizing the

distraction of lay worshipers. Behind the high walls of these new viharas, Buddha images had taken the place of stupas as the central ritual foci. It is possible, then, that the unobtrusive addition of Buddha images at pilgrim- age sites was not intended for the laity at all. Rather, the addition of Buddha images might have served that group that is known to have originated and championed the image cult in Buddhism—the sangha itself. While some laity might have followed the sangha’s lead, given the stability in architec- tural layout of pilgrimage stupas in the first millennium ce, it does not appear that the bulk of the laity ever fully embraced the image cult to the same degree as the sangha did. Whereas the sangha rebuilt and refashioned their world to concord with their new ritual interests—abandoning stupas and chaityas while creating Buddha images and the perfumed chambers that housed them—the laity did nothing of the sort. At the sites favored by the laity, stupas were not torn down and replaced by massive images of the Buddha. Rather, Buddhist pilgrimage sites remained pretty much the same, with only the unobtrusive addition of some Buddha images, mostly for the benefit of the newly resident sangha.

Overall, based on the architecture and layout of Buddhist pilgrimage sites, the Buddhist laity settled on a ritual format that satisfied their individual and communal desires by at least the second century bce and stuck with it for more than a millennium. This layout consisted of a large central stupa encircled by a railing that separated the circumambulatory path from a larger communal worship space. The layout of the main stupa at Sanchi reached its final form, more or less, in the fourth or fifth cen- tury ce, and remained that way until the site was abandoned sometime after the twelfth century ce. Even those pilgrimage sites that were first built in the first millennium continued to follow the same general layout as those that came earlier. Thus, new pilgrimage centers in Karnataka, Orissa, and elsewhere resemble, in almost all details, the earliest pilgrim- age complexes at Sanchi and Bharhut. At least for those laity who con- tinued frequenting Buddhist pilgrimage sites in the first millennium ce, it appears little changed. It is less clear, however, what happened to all those lay Buddhists who stopped going to Buddhist pilgrimage sites. If lay Buddhists were abandoning Buddhism, where were they going?


Just like all other periods of Indian history, religious practice during the first millennium ce was highly heterodox. Along with Buddhism, major religions included Hinduism, Jainism, Zoroastrianism and, beginning

in the seventh century ce, Islam. While often discussed as rival, distinct religious traditions, popular religious practice was highly syncretic, with rival religious traditions borrowing from, and blending into, one another. Jainism has been discussed in several earlier chapters. Islam will be dis- cussed in greater detail in Chapter 7. Here I will focus on the relationship between Buddhism and Hinduism.

The archaeological history of Hinduism is even more complex and prob- lematic than Buddhism. The first difficulty is the term “Hindu.” The term “Hindu” was first coined in the nineteenth century to label a bewilder- ing variety of religious beliefs, sacred texts, and ritual practices, includ- ing the more philosophical understanding of the religion by Brahmans and the daily, lived practices of ordinary Hindus. Just as in Buddhism, the interests and religious practices of the Hindu elite did not wholly align with the quotidian interests of the Hindu laity. In the first millennium ce, nobody referred to themselves as Hindu. Rather, people were devo- tees of several different gods and god-like beings including Shiva, Vishnu, Brahma, Ganesha, Durga, Krishna, and numerous others. Among those who practiced a more philosophical form of Hinduism, all of the gods were arrayed on a complex pantheon, with most of the gods viewed as avatars of either Shiva of Vishnu. The bulk of the population, however, cared little for such complex theological abstractions and simply worshiped and made offerings to their local deities.

Modern-day Hinduism differs from other major world religions in many important respects, in that it has no founder, no fixed canon which embodies its major beliefs and practices, and no organized priesthood. It is also marked by a great variety in beliefs, practices, sects, and traditions. Some scholars argue that Hinduism is not so much a religion as a set of socio-cultural prac- tices; others argue that it is inextricably linked to the existence of caste, and still others hold that we should talk of Hindu religions in the plural rather than the singular. The relative newness of the word, the problems of defini- tion, and the existence of much internal diversity, are not sufficient reasons to avoid the use of the term Hinduism. (Singh 2008:433)

Like Singh, I still see value in using the term “Hindu,” though only as a shorthand to describe the widely divergent yet overlapping religious beliefs that were developing in the first millennium ce. The term “Hindu” is a convenient label when speaking generally. The term “Hindu” is inap- propriate, however, when discussing specific archaeological sites, reli- gious beliefs, or ritual practices. When discussing specific instances, I use the appropriate terms for the specific sect of Hinduism being discussed.

The historical origins of Hinduism are difficult to pin down. For some scholars, the origin is found in the Rig Veda, the composition of which is conventionally ascribed to the mid-second millennium bce. While there is no doubt that the Vedas are a critical foundation and antecedent for modern Hinduism, they also describe a religion very different from what anybody would consider modern Hinduism. First, the primary gods in the Vedas are Indra and Agni, not Shiva and Vishnu. Second, the Vedas emphasize animal sacrifice, including the sacrifice of cattle. While sacri- fice is still practiced in modern Hinduism, animal sacrifice is rare, and cattle sacrifice specifically taboo. For these, and numerous other reasons, I do not necessarily label practitioners of Vedic rituals as Hindus. I reserve the term “Hindu” for those ancient Indians who worshiped gods that are recognized as part of the Hindu pantheon today. That said, particularly in the beginning of the first millennium ce, there was significant overlap between those people who engaged in Vedic rituals and those people who worshiped Hindu gods. Several Gupta rulers, for example, actively sup- ported nascent Hinduism while also performing the Vedic horse sacrifice (Singh 2008).

As discussed in Chapter 1, though describing events that occurred in the first millennium bce, it is likely that the earliest existing versions of the Ramayana and Mahabharata date only to the early to mid-first millen- nium ce. The Puranas, Thapar (2000:133) has convincingly argued, were likely composed at roughly the same time, and are best seen as illustrating the worldview and religious understandings of the time of their composi- tion. In contrast, the earliest archaeological and epigraphic evidence for Hinduism dates to the first few centuries bce. Among the earliest evidence for the worship of Vishnu is found in a second-century bce inscription carved on a pillar in Madhya Pradesh, near Sanchi. The inscription records the donation of the pillar to a nearby Vishnu temple by an Indo-Greek devotee. The earliest unequivocal evidence for the worship of Shiva also dates to the second century bce.1 This evidence consists of representa- tions of Shiva as a lingam (phallus) found on a relief from Mathura and lingam from Andhra Pradesh (see Figure 6.3).

Over the course of the first millennium ce, Hinduism and the practice of Hindu rituals expanded throughout India. This spread was facilitated by

1. Many scholars see evidence for Shiva on Indus Valley seals dating from the second millennium bce (Chakrabarti 2001, Kenoyer 1998; Marshall 1931; Possehl 2002). In a previous article I have questioned the value of this identification (Fogelin 2007b). It is not that I deny that the seated-figure on Indus seals appears similar to modern depictions of Shiva. Rather, I am not sure how noting the similarities between the ancient and modern depictions help to understand or explain either one.

Figure 6.3: Shiva lingam, Gudimallan Village, Andhra Pradesh (c. first or second century bce)

Hindus’ ability to incorporate regional and local gods within their broader religious framework. Rather than replace local gods with the worship of Shiva or Vishnu, local gods were simply recast as an avatar or incarnation of Shiva or Vishnu. By incorporating local religious traditions in this way, converts to Hinduism were hardly converted at all. For the most part, the ritual lives of the newly Hindu were little different from what they prac- ticed before—they continued worshiping their own gods, in their own ways and often even in their own temples. This easy syncretism allowed nascent Shaivite and Vaishnavite sects to attract significant numbers of converts in the first millennium ce. Given the Buddhist sangha’s general neglect of lay rituals and lay concerns, many of the converts to Hinduism were former Buddhists. Just as they had done with other sects, early Hindus facilitated conversions by incorporating Buddhist ritual practices, temple designs, and even the Buddha himself as an avatar of Vishnu.

While there are significant theological and complex philosophical differences between Buddhism and Hinduism, these abstract, esoteric debates would have had little impact on the ritual practices of the laity.

Rather than debates on the origin of the universe or the sources of sorrow, the laity always took a more practical approach to religion—they sought the favor of the gods to bless a marriage, move people through the after- life, or protect against misfortune. Ordinary Indians did not overly con- cern themselves with the arcane differences between different Buddhist sects, or even the larger differences between Buddhism, Hinduism, or Jainism. As long as religious figures could lead rituals in a format more or less familiar to the people experiencing them, sectarian differences didn’t overly matter. This fluidity and heterodoxy has characterized religious practice in India for millennia, and is critical for understanding the his- tory of lay Buddhism in the first millennium ce. This heterodoxy allowed the laity to shift their religious allegiances from Buddhism to Hinduism without ever significantly changing their ritual practices.


Perhaps the best place to examine the broad similarities between Buddhism, Hinduism, and Jainism in the first millennium ce is the site of Ellora in Maharashtra. In the latter part of the first millennium ce, with the support of the Rashtrakuta Dynasty, Ellora became a major religious center, with thirty-four rock-cut temples, monasteries, and shrines of the three religions erected nearly side by side on a long cliff face. On the southern end of the cliff face, twelve Buddhist caves were constructed between the seventh and eighth centuries. Several of these took the form of multistoried viharas. It is likely that at least some of the Buddhist caves were the earliest constructions at Ellora. In the middle of the cliff were seventeen Shaivite caves, including Kailashnatha, a mas- sive eighth-century temple that represents the pinnacle of the Indian rock-cut temple tradition. To the north, somewhat separated from the Shaivite and Buddhist caves, Jains carved five caves in the ninth and tenth centuries. Overall, the rock-cut caves at Ellora are the culmination of almost a millennium of development from the earliest rock-cut caves in the Barabar hills in Bihar (Chapter 3) through the Buddhist monaster- ies of the Western Ghats (Chapter 4) and Ajanta (Chapter 5). By examin- ing the long-term development of rock-cut architecture in the Western Ghats, it is possible to see the incorporation of Buddhist ritual practices into early Hinduism. As the Buddhist sangha gradually stopped building chaityas and withdrawing into their viharas, Shaivites began taking key elements of Buddhist architectural design and incorporating them into their own public worship spaces.

Early rock-cut Buddhist monasteries in the Western Ghats typically consisted of two separate spaces—apsidal chaityas for worship and square viharas that served as monastic living quarters. Over the course of the first millennium ce, the sangha gradually merged chaityas and viharas, creating single spaces that served as both living quarters and worship spaces (see Figure 6.4). This change coincided with the adoption of Buddha images by the sangha, initially carved on the front of stupas within apsidal halls (e.g., Ajanta Cave 26) and, by the mid-first millennium, placed inside spe- cial chambers within viharas themselves. With the placement of Buddha images within viharas, viharas began to take on some of the attributes of apsidal chaityas—particularly in the way that people used and moved through space.

In the earliest apsidal chaityas, devotees entered the worship spaces and circumambulated by walking clockwise in the circumambulatory path. This served to keep the central stupa on the right of the devotee. Initially, the placement of Buddha images within viharas eliminated the potential to circumambulate them. At Ajanta 17, for example, the image was placed in a small room along the back wall, preventing circumambulation. This layout required a shift in the way people moved through the space. In order to keep the image on the right, devotees were forced to walk coun- terclockwise through the space. By the sixth century ce at Aurangabad, the sangha had redesigned their viharas to allow for both image worship and circumambulation (see Figure 6.4). At Aurangabad 7, the sangha had

Figure 6.4: Left, Buddhist vihara (Aurangabad Cave 7: c. sixth century ce) and right, Hindu Temple (Ellora Cave 14: c. seventh century ce; after Fergusson and Burgess 1880)

a path carved behind the perfumed chamber, allowing for circumambula- tion. By the seventh and eighth centuries, the Buddhist sangha at Ellora resumed placing the perfumed chamber along the back wall of the viharas. Critically, however, at Ellora the Shaivite caves were carved following the same layout as Aurangabad 7. That is, the central image chamber was placed to allow both viewing of the lingam and circumambulation of the room containing the lingam (see Figure 6.4). In this way, the Shaivite temples at Ellora were modeled on earlier Buddhist viharas in the Western Ghats.

There is, however, one key difference between the Buddhist and Shaivite temples at Ellora. Unlike Buddhist viharas, the Shaivite caves at Ellora omitted monastic cells. The Shaivite caves, like the long-since-abandoned apsidal chaityas of Buddhism, were worship spaces, not living quarters for Brahman priests. Where the Buddhist sangha increasingly shunned the laity in the first millennium ce, Shaivites welcomed the laity—even to the degree of fashioning their ritual spaces to be familiar to Buddhist converts. In Shaivite temples, the Brahmans even refashioned their pri- mary focus of ritual—the lingam—to more closely resemble a stupa (see Figure 6.5).

In previous chapters I have discussed the slow change in the form of monastic stupas—from iconic stupas that signified the burial mound of the Buddha, to more symbolic attenuated stupas and 3/5 anda stupas. I argued that each of these changes represented a shift in the degree of abstraction from the original earthen stupa in which the Buddha’s corpo- real relics were interred. I also argued that the sangha’s manipulation of the form of the stupa was a legitimation, an attempt to foster their own power over those who came to worship at their chaityas. Here I argue that a similar process occurred in Shaivite temples with manipulations to the form of linga, where earlier iconic linga were gradually replaced by more symbolic, abstract linga—that is, between the first century bce and the eighth century ce, Shiva linga gradually shifted from more literal depic- tions of a phallus to stouter, more abstract depictions of a phallus (see Figure 6.5). Just as in Buddhism, these shifts corresponded with a shift from a complex, multifaceted sign that combined iconic, indexical, and symbolic elements to a more esoteric, emotionally distant sign with less iconic immediacy.

The first- or second-century bce Shiva lingam from Gudimallam Village, Andhra Pradesh (see Figures 6.3) was an icon of Shiva’s phallus with an iconic representation of Shiva carved on the front (Singh 2008). The phal- lus also served as an index of Shiva’s spiritual power and likely also served as a more general symbol of the Shaivite sect. By the beginning of the fifth century, the lingam in Cave 4 at Udayagiri (see Figure 6.5) was stouter

100 BCE

0 CE

Ajanta Cave 10

100 CE

200 CE Gudimallam linga

Kanheri Cave 4

300 CE

400 CE

Ajanta Cave 26

500 CE

Udayagiri Cave 4

600 CE

700 CE

Aurangabad Cave 7

Ellora Cave 16

800 CE

Figure 6.5: The comparative development of Buddhist and Hindu ritual foci in the early to mid-first millennium ce

(less iconic of a phallus), and only Shiva’s face was depicted (Willis 2009). By the eighth century ce, the lingam in the Kailashnatha temple had lost its iconic significance, with the symbolic elements dominant. Perhaps more interesting, the progressive abstraction of Shiva linga resulted in a shape with more or less the same proportions as an attenuated Buddhist stupa from centuries earlier—from the time when the Buddhist sangha was still actively attempting to assert their authority over the laity. The functional equivalence of Shiva linga and early stupas is most clearly dem- onstrated at Karla, one of the earliest rock-cut Buddhist monasteries in the Western Ghats. Sometime after Buddhists abandoned Karla in the seventh century ce, the chaitya was converted to a Shaivite temple and the stupa was painted to resemble a lingam (Singh 2004:291).

Overall, the metamorphosis of Shiva linga in the first millennium ce follows the same general pattern as the metamorphosis stupas. That is, both were initially iconic and slowly became more symbolic over time. The difference is that in the early to mid-first millennium ce, the Buddhist sangha returned to the more iconic worship of the Buddha in the form of images, while Shaivites did not. Rather, they continued to construct more symbolic representations of the phallus, representations that also just happened to strongly resemble the iconic representations of a stupa favored by the formerly Buddhist laity. As such, the lingam at Ellora and other late first millennium ce Shaivite temples were intentionally multi- valent, meaning different things to Shaivite and Buddhist audiences. The Shaivite temples at Ellora brought several elements of Buddhist design together to produce a ritual space that Buddhist converts would feel com- fortable worshiping in. Former Buddhists would enter and move through these new Hindu temples in much the same way they had entered and moved through earlier Buddhist chaityas. Likewise, the ritual focus of the structures, the lingam, would be familiar to Buddhist converts, as it closely resembled a stupa. Critically, these appropriations of Buddhist temple design were not limited to the Ellora, but were common across India.

Contemporary Hindu temples in India shared the same basic layout as those found at Ellora. They generally consisted of a processional path (prada- kshina) leading to the central chamber where devotees could view the image of the god who resided there. Early Hindu temples also typically included a columned hall (mandapa), strongly resembling the courtyard found at the center of Buddhist viharas. In Shaivite temples, linga were abstract and symbolic—multivalent signs that were both symbolic of the phallus and iconic of a stupa. As such, throughout India early Hindu temples were designed to help draw the Buddhist laity to Hinduism. By minimizing the differences between Hindu and Buddhist worship—by making the Buddhist

laity comfortable—early Hindu sects successfully recruited the Buddhist laity just at the point when the Buddhist sangha was turning their backs on them. This process of conversion even extended to Buddhist sites them- selves. As will be discussed further at the end of this chapter, as Buddhist sites across India failed and were abandoned in the first millennium ce, they were often re-sanctified as Hindu temples.

For the most part, the story of Hindu/Buddhist syncretism concerns the appropriation of Buddhist elements by Hindus. This imbalance makes sense given that the Buddhist sangha withdrew from regular contact with the laity over the course of the first millennium ce. If the Buddhist sangha was turning their back on their own lay followers, they were not adjusting their ritual and religious practices to attract new followers. That said, there is evidence for the appropriation of Hindu ritual and archi- tectural elements into Buddhism. These appropriations take two forms. The first are appropriations of Hindu elements in later Buddhist pilgrim- age sites. While many Buddhist pilgrimage sites were abandoned or ceded to Hindu sects in the first millennium ce, those that remained affiliated with Buddhism took on more Hindu elements. The second locus for appro- priation was the sangha itself. In terms of the Buddhist sangha, the most obvious appropriation might be Buddha images. There is no question that early Hindu sects were depicting their gods in friezes and sculptures sev- eral centuries before the sangha began creating Buddha images. Likewise, in the early centers for the production of Buddha images in Gandhara and Mathura, Hindu images predate Buddha images, and Buddha images share iconographic elements with early depictions of Hindu gods. Thus, the Buddhist sangha may have been inspired to create Buddha images after seeing the images created by early Hindus. Beyond images, there are also numerous examples of Buddhists appropriating Hindu religious con- cepts into monastic Buddhism in the late first millennium ce.

Perhaps the clearest appropriation of Hindu elements into Buddhist pil- grimage architecture occurred at Bodh Gaya. While stray finds and sculp- ture found at Bodh Gaya indicate that the site was in use from at least Mauryan times, Cunningham’s ([1892] 1998) excavations revealed that the main temple rested on foundations first constructed in Gupta times. The focus of Bodh Gaya is the Bodhi tree, said to be a descendant of the tree under which the Buddha gained enlightenment. The main temple at Bodh Gaya (see Figure 6.6) is situated directly east of the Bodhi tree, with an image of the Buddha seated on a throne placed where the Buddha sat when he attained enlightenment. Mitra (1971:61) argues that this image dates to the Gupta period, replacing an earlier image of an empty throne that concorded with the earlier taboo on iconic depictions of the Buddha.

Figure 6.6: Mahabodhi Temple at Bodh Gaya

Courtesy of Wikimedia Commons ( temple.jpg); converted from color to black and white.

While the superstructure of the temple at Bodh Gaya was regularly recon- structed, the general layout of the worship spaces conforms to the general plan of a Gupta period Hindu temple. Likewise, the Guptas were promi- nent supporters of nascent Hinduism.

The idea that early Hindu sects took in converts from Buddhism is not a new idea. Neither is the idea that Hinduism is a highly syncretic religion that incorporated Buddhist theology, ritual, and ritual architecture into its day-to-day practices. Even the claim that Hindu temple architecture is inspired by Buddhist architecture is widely noted. What is new in this analysis is the significance attributed to the architectural changes—that is, the emphasis on the sensual experience of moving through a temple and engaging with the central ritual foci. While the ground plans of

Buddhist and Hindu temples are similar, people do not experience ground plans. People experience architecture in three dimensions as they move through a space. It is the feeling of movement, the process of revelation, and the rituals performed in Hindu temples that would have been familiar to the Buddhist laity. All that said, we must be careful using the term “con- version.” As I have argued repeatedly, it is likely that the laity never wholly identified themselves as Buddhist, Hindu, Jain, or whatever. Rather, they relied on holy people to assist in marking life events and traveled to par- ticularly sacred sites, with little regard to the specific sectarian differ- ences of the people they interacted with and the sites they visited. In the long run, Hinduism successfully gained devotees as Buddhism lost them during the first millennium ce, but it is likely that from the lay devotees’ perspective, very little, if anything changed.


While, in an abstract sense, the layout and form of Buddhist pilgrim- age sites were relatively stable in the first millennium ce, this does not mean that every major pilgrimage center was stable and successful. Over the course of the first millennium ce, several established Buddhist pil- grimage sites were abandoned or neglected. Perhaps the most surpris- ing of these is Lumbini, the birthplace of the Buddha. During Faxian’s travels through North India in the early fifth century ce, he noted that Lumbini was neglected, with only a few monks living among the ruins. By the time Xuanzang visited Lumbini in the mid-seventh cen- tury, Lumbini had been all but abandoned by the Buddhist sangha. By the late first millennium ce, Chinese pilgrims increasingly noted similar patterns of neglect and abandonment at other Buddhist centers across India. While some sites, like Lumbini, were abandoned in the heartland of Buddhism in the Gangetic Plain, in general more Buddhist sites were abandoned in the South, West, and Northwest. It is important, however, not to overstate this pattern. Even in the regions where Buddhist insti- tutions were abandoned in greater numbers, some Buddhist pilgrimage centers and monasteries survived. As will be discussed in greater detail in Chapter 7, those Buddhist institutions that persisted were often those that were least associated with, or dependant on, continued state sup- port. Following this chapter’s focus on lay Buddhism, here I will only discuss the abandonment and neglect of pilgrimage centers frequented by the laity.

Before discussing the abandonment of specific Buddhist sites, it is important to note two problems in the archaeological evidence. First, the date of the last constructions at archaeological sites does not necessarily mark the abandonment of those sites. Temples and viharas may have been used for several centuries after they were constructed. Without careful excavation, attention to the smaller artifacts, and carbon-14 dating, it is often difficult to determine how long a site was used after the last con- structions. Few Buddhist sites in India have been excavated to this degree of precision. The second difficulty is the tendency for archaeologists to focus on beginnings rather than endings. Archaeologists, generally, want to find the earliest of whatever they are studying. They spend inordinate amounts of time determining when a site was first founded, with less time considering a site’s abandonment. Given these two difficulties, it is usually more difficult to study the abandonment of a site than its foundation, and the range of dates of abandonment are typically wider and more indeter- minate. With those hesitations noted, it is possible to identify some gen- eral trends in the abandonment of Buddhist sites in India. Interestingly, there is a marked difference in the patterns of neglect and abandonment between large and small Buddhist pilgrimage centers. Whereas many of the largest and most prominent pilgrimage sites in South, West, and Northwest survived through the beginning of the second millennium ce, smaller pilgrimage centers were mostly abandoned in the mid- to late first millennium ce.

Beginning with the large pilgrimage centers, the last building at Sanchi (temple 46) was constructed in the tenth or eleventh century ce, with abandonment likely dating to the twelfth or thirteenth century (Marshall and Foucher 1983; Mitra 1971). The last inscription at Bharhut—now only preserved in a single photograph—dates to the tenth century, with some Buddha images dating to the eleventh century ce (Mitra 1971:96). At Amaravati, in Andhra Pradesh, a donation inscription dates to 1234 ce, with another inscription from Sri Lanka recording the restoration of a temple at Amaravati in 1344 ce (Mitra 1971:201). Not all the large pil- grimage stupas survived so long. Based on preliminary published reports (Poonacha 2007), Kanaganahalli stupa in Karnataka was abandoned by the third or fourth century ce. In the Northwest, the last modifications of the Dharmarajika stupa at Taxila date to the fourth or fifth century ce. By the time Xuanzang passed though Taxila in the beginning of the seventh century ce, both the city and stupa had been long abandoned. This should not, however, be taken to mean that Buddhist pilgrimage stupas were abandoned in the Northwest earlier than they were in other regions of India. While the Dharmarajika stupa fell into disuse, several

other large pilgrimage complexes were erected in the Northwest (e.g., Shingardar).

In contrast to the large pilgrimage centers, smaller pilgrimage cen- ters seem to have been abandoned earlier. In part, this pattern may be the result of the difficulty distinguishing monastic sites with space devoted to lay worship from pilgrimage sites with associated monaster- ies. Outside the heartland of Buddhism in the Gangetic Plain, many pil- grimage centers were initially centered at monastic sites. As discussed in earlier chapters, early Buddhist monasteries in the South, West, and Northwest India typically included public worship spaces where the sangha sought to exert influence over the laity. Over the course of the first millennium, with the withdrawal of the sangha from day-to-day inter- action with the laity, it would make sense that the sangha would aban- don those monasteries that were most frequented by lay pilgrims. Thus, in Andhra Pradesh, early Buddhist centers at Thotlakonda, Bavikonda, Sankaram, Salihundam, and Nagarjunakonda were abandoned between the third and fifth centuries ce. In the Western Ghats, Bhaja, Bedsa, Ajanta, other early rock-cut monasteries were abandoned by roughly the seventh century ce. In the Northwest, Takht-I-Bahi and the monasteries near Taxila (e.g., Kalawan, Jaulian) were abandoned beginning around the fifth or sixth century ce. The common element of almost all the smaller monastic/pilgrimage centers that failed was that they combined public worship spaces and private viharas reserved for the sangha. In con- trast, those centers that were more clearly identifiable as pilgrimage cen- ters (e.g., Amaravati and Sanchi) survived. Critically, however, as smaller monastic/pilgrimage centers failed in the South, West, and Northwest, they were not replaced with other lay-oriented Buddhist pilgrimage sites. Rather, the small pilgrimage/monastic sites were either abandoned or taken over by rival religious orders.

There is, perhaps, no clearer archaeological evidence of syncretism as the reoccupation of religious sites. The physical and symbolic appro- priation of Buddhist sites by Hindus and Jains served as legitimations in their attempts to draw the laity to their own orders. It is not simply that the reoccupation of Buddhist sites reflected the increasing power of Hindu and Jain sects, but rather that the reoccupation of Buddhist sites actively created this power and facilitated the conversion of the laity. Just as with the adoption of Buddhist architectural and ritual practices in new temples, the reoccupation of Buddhist sites fostered an easy syncretism requiring few, if any, changes in the laity’s ritual practices. Beginning in the mid-first millennium ce, former Buddhist sites across India were reoccupied by Hindu, Jain, and other religious orders. For example, at

Sankaram in Andhra Pradesh, images of Ganesha were carved into the walls of the Buddhist shrines (Rea 1907–1908). In the Western Ghats, the early rock-cut chaitya at Karla was converted into a Shaivite temple some- time after the seventh century ce, with the primary stupa repainted to resemble a lingam (Singh 2004:291). By early to mid-second millennium ce, some of the large Buddhist pilgrimage centers in the Gangetic Plain were reoccupied by rival religious orders, with even Bodh Gaya converted to a lesser Shaivite temple sometime after the fifteenth century.


Despite the ever-changing political and religious landscape of India in the first millennium ce, the patterns of worship of the laity were oddly stable. For the most part, the laity continued to worship in much the same way as they always had. The laity continued to frequent religious sites that pro- vided them with processional paths that brought them into proximity of their gods. In Buddhism, this proximity was created through the iconic relics entombed within stupas. In early Hindu sects, this was accomplished through iconic images of the gods placed in the heart of Hindu temples. Rather than rival religious sects attempting to alter lay religious practice to new and different forms, it appears that rival religious sects altered their own practices, either conforming to the patterns of lay practice or shifting away from them. The Buddhist sangha, relying on their ever-increasing endowments of money and land, recreated Buddhism in a form distant from lay practice. Over the course of the first millennium, the sangha progressively abandoned the laity, segregating themselves within viharas centered on Buddha images. Early Hindus, in contrast, adopted the ritual and religious foci of early Buddhism, attracting the devotions of the laity. Whether Shaivite or Vaishnavite, early Hindus refashioned their temples to mimic the feel of Buddhist temples. In the case of Shaivites, this even included refashioning their ritual foci—linga—to more closely approxi- mate a stupa. In the case of Vaishnavites, Buddhists were incorporated by subsuming the Buddha as one of several avatars of Vishnu. In either case, Hindus gained the devotions of lay Buddhists by refashioning themselves rather than refashioning the laity.

By the end of the first millennium ce, the monastically based sangha had separated themselves from day-to-day interactions with the laity, and smaller Buddhist monastic/pilgrimage sites were being abandoned or taken over by Hindus, Jains, or other sects. In the early centuries of the second millennium ce, even Buddhist monasteries began to fail, and by

the mid-second millennium ce, only vestiges of Buddhism remained in the practices of rival religious orders. In the late first and early second millennium ce, the Buddhist sangha in India failed to adapt to the chang- ing political, economic, and religious context. When faced with crisis, the monastically based sangha could no longer fall back on the support of the laity since the laity had long since switched their allegiances to rival reli- gious orders. The degree to which vestiges of Buddhism remained in India was the result of a small number of Buddhist ascetics who had left the monasteries in the late first millennium ce and championed a new, highly syncretic, form of Buddhism—Tantric Buddhism.




The Consolidation and Collapse of Monastic Buddhism

c. 600–1400 ce


n about 631 ce, the Chinese Pilgrim Xuanzang entered Northwest India through the Khyber Pass. For the next decade he traveled throughout India, gaining instruction at several monasteries and collecting Buddhist texts (Rongxi 1997). The world that Xuanzang encountered was nota- bly different from the world described by Faxian, two centuries earlier. Throughout his travels he found many smaller Buddhist monasteries and pilgrimage sites abandoned, while larger monasteries had become politi- cally powerful institutions with massive landholdings that provisioned thousands of cloistered monks, nuns, and novitiates. These Buddhist monasteries had become luxurious fortresses. Within the walls of their monasteries, the sangha practiced a scholastic form of Buddhism, divorced from the day-to-day concerns of the Buddhist laity. In the texts studied, however, were frequent examples of monks who abandoned the life of the monastery for a life on the forest. As in the previous period, to a large degree, these romantic visions of life in the forest were intended to allay the ascetic desires of monks living collectively in monasteries. But in some cases, it appears that some Buddhist monks actually did it—actually aban- doned the large monasteries for an ascetic life in the peripheries of soci- ety. In the process, these newly ascetic monks may have created Tantric


( 202 )

With its more worldly focus, Tantric Buddhism allowed for more engagement with both the laity and the ascetics of rival religious orders. None of this is to suggest that Tantric Buddhism existed wholly outside the monasteries of the late first millennium ce. Just as previous forms of Buddhism had both monastic and lay versions, Tantric Buddhism came in both high and low forms. Tantric Buddhism was not simply a renunciation of monastic life, but rather another venue in which the tensions between the collective and ascetic desires of the sangha and the laity were played out. While the balance between the individual and the group was different from contemporary forms of scholastic Buddhism, a balance was struck nonetheless. More so, as Buddhist monasteries collapsed and scholastic Buddhism was abandoned in India in the beginning of the second millen- nium ce, the practitioners of Tantric Buddhism preserved Buddhist tradi- tions in India, in practice if not in name.

This chapter is divided into two parts. In the first I discuss the cen- ters of scholastic Buddhism and the origins of Tantric Buddhism in the seventh–tenth centuries. In the second part I examine the decline and col- lapse of Buddhism in most of India in the beginning of the second millen- nium ce. As discussed in Chapter 6, the decline of popular Buddhism in India had begun in the first millennium, as Hinduism and other religions began to win the allegiance of the laity. While the lack of lay support- ers weakened the sangha in much of India, the final collapse of monas- tic Buddhism began in 1192 with the invasion of North India by Central Asian Turks.


Between the seventh and twelfth centuries, India was divided into numer- ous kingdoms, both small and large. The most important of these were the Gurjara-Pratiharas in the Northwest, the Palas in the Northeast, and the Rashtrakutas in the Deccan Plateau. Between the eighth and tenth centuries these three empires were in regular conflict with each other, primarily to control the central Gangetic Plain and the city of Kanauj. In the broadest sense, the political and economic patterns of the seventh through twelfth centuries ce follow those of the period that preceded them. Different polities gained or lost prominence, but land continued to be the primary source of wealth, with aspiring rulers taking land from their rivals and giving land to their followers, including Buddhist, Jain, and Hindu religious institutions. At the same time, new land was cleared and was converted to agricultural use along the peripheries. This was

accomplished by either gifting land to military leaders or religious estab- lishments, or by incorporating hill tribes into the state.

One of the key developments of this period was the increased pro- duction of texts, both from within religious orders and, critically, from the political elite. While early texts like the Arthashastra testify to a long-standing tradition of political and dynastic texts in India, the quan- tity of these texts greatly increased after the sixth century ce. These new texts included elaborate dynastic histories, hagiographies of prominent kings, and plays. To be clear, none of these texts was wholly secular; they made frequent mention of the gods and often extolled the quasi-divine actions of the kings. The primary benefit of these texts is that they pro- vide an alternative view of the religious orders—a view that can be con- trasted with sangha’s view of themselves.

Between the seventh and twelfth centuries ce, Buddhist, Jain, and Hindu religious institutions were an integral part of the larger politi- cal, economic, and religious world, organizing agricultural production in the peripheries, participating in long-distance trade, and receiving the financial support of kings and other elite. Support of religious institu- tions was a critical element in the competition between the Pratiharas, Rashtrakutas, and Palas. The Rashtrakutas, for example, made substan- tial donations of land to the Buddhist, Hindu, and Jain institutions at Ellora (discussed in Chapter 6). The Pratiharas favored Jain and Hindu sects, but also supported Buddhists in their territories. In contrast to the Pratiharas, the Pala rulers practiced and favored Buddhism. In part, this was due to the Pala’s trade ties with dominantly Buddhist states in the Himalayas and Southeast Asia. Pala land grants were critical to the cre- ation and maintenance of several very large monasteries in the Gangetic Plain and the Northeast, including Nalanda, Somapura, Odantapuri, Vikramashila, Jaggadala. Whereas earlier monasteries housed hundreds, these monasteries had resident populations in the thousands and drew Buddhist scholars and novitiates from across Asia. The scale of land controlled by the monasteries was similarly large. Yijing, a late seventh century Chinese pilgrim to India, stated that the over 3,000 monks who lived at Nalanda were provisioned by more than 200 villages that “were offered to them as alms by the kings and monarchs of successive dynas- ties” (Rongxi 2000:63).

Between the seventh and tenth centuries, Buddhist monasticism grad- ually faded in northwestern, western, and southern India, as rulers and lay followers gradually shifted their attentions to Hinduism, Jainism, and other religious orders. Buddhism persisted in Northeast India, Orissa, and the Gangetic Plain, primarily due to the support of the Palas and the

proximity to heavily Buddhist populations in the Himalayas and Southeast Asia. Centered in Bengal, the Palas were the last major state in India to strongly support Buddhism. The first Pala king was Gopala I (750–780 ce), who ruled over much of Bengal. Under King Devepala (810–850) the Palas greatly expanded their empire, pushing up the Gangetic Plain and south- ward into Orissa and northern Andhra Pradesh. The fortunes of the empire waxed and waned in the ninth and tenth centuries, often in direct rela- tion to the strength of the Pratiharas and Rashtrakutas. Beginning in the eleventh century, the Palas began to decline in power, in part due to the threats from the Chalukya Empire in the south. In 1095 ce, a Pala tribu- tary named Hemanta Sen rebelled against the Palas and established a new independent state. His successor, Vijay Sen (1095–1159 ce), expanded the Sena Empire across the most of Bengal at the expense of the Palas. Unlike the Buddhist Palas, the Senas were Hindu and primarily supported Hindu institutions. The last Pala king reigned over a nominal territory until 1174 ce. The Senas themselves lost most of their territories in Eastern Bengal at the start of the thirteenth century, as Muhammad of Ghori extended and consolidated Turkic power throughout North India.


In the seventh through twelfth centuries ce, Buddhism continued to change and develop. Many of these changes were conditioned on the pro- gressively more distant relationship between the sangha and the laity. Within the monasteries, Buddhism became an increasingly scholastic endeavor as the sangha focused on the mastery of important texts. In opposition to the scholastic turn of monastic Buddhism, some members of the sangha began leaving monasteries for the life of wandering ascetics. For seemingly the first time in Indian Buddhism, the tensions between forest monks and the monastically based sangha, central to Tambiah’s (1976, 1984) studies of Thai Buddhism and Carrither’s (1979, 1983) stud- ies of Sri Lankan Buddhism, come into play. At the same time, a new form of Buddhism—Tantric Buddhism, or Vajrayana Buddhism—developed alongside Mahayana Buddhism. Tantric Buddhism was not the sole prov- ince of either monastic or ascetic Buddhism—rather, Tantric Buddhism had both ascetic and scholastic variants, with substantial borrowing between them.

The origins of Tantric Buddhism are even murkier than those of Mahayana Buddhism. Just as early Mahayana scholars had done, the

earliest practitioners of Tantric Buddhism claimed that their teachings were originally taught by the Buddha and were hidden until such time as the world was ready for them. Further, Tantric Buddhism was far more eclectic and heterodox than Mahayana Buddhism, subsuming a wider range of ritual practices and borrowing freely from other religious tra- ditions. For these and other reasons, “it remains exceedingly difficult to identify what it is that sets the Tantric Buddhism apart” (Lopez 2001:214). That hesitation noted, among the more important characteristics of Tantric Buddhism was its use of supernormal or magical acts to greatly speed the attainment of enlightenment. Where earlier forms of Buddhism required long periods of meditation and potentially several lives to attain enlightenment, Tantric Buddhism held out the promise of almost imme- diate salvation. Tantric Buddhism also often had a more worldly focus, with the attainment of supernormal powers often seemingly more impor- tant than enlightenment.

Most scholars place the origins of Tantric Buddhism in India in the mid-first millennium ce, with more widespread practice beginning as early as the seventh and eighth centuries ce. By the ninth and tenth cen- turies, Tantric scholars were present at Nalanda and dominated in the monasteries in the Northeast. From there, Tantric Buddhism spread to the Himalayas, where it is still practiced today. Tantric practices were recorded in a new class of writings—tantras—from which Tantric Buddhism gets its name. In contrast to sutras produced in Mahayana Buddhism, tantras have a more revelatory and less philosophical bent. As reported in the tan- tras, Tantric ritual took many forms, ranging from the intentional trans- gression of social taboos to complex visualizations of Buddhist principles. Much of ritual described in the tantras centers on the intentional break- ing of social taboos. Tantric literature makes frequent mention of sex with low caste women, eating meat, and frequenting burial grounds. In Tantric Buddhism, these socially transgressive behaviors were seen to be a source, or demonstration, of supernormal powers. In part, these transgressions demonstrate the illusory nature of the world. If the world is truly an illu- sion, then sex with a low caste woman in a cremation ground would also be an illusion—and therefore not transgressive at all. At the same time, in an anthropological sense, transgressive rituals also brought Tantric practitioners into contact with the liminal. Following Mary Douglas’s arguments concerning pollution discussed in Chapter 2, contact with pol- lution was both dangerous and powerful. Those who mastered liminal forces gained substantial power to affect the world. Thus, transgressive Tantric rituals could allow practitioners to levitate, turn enemies to stone,

or attain enlightenment.

It is not clear if transgressive Tantric rituals were ever commonly prac- ticed in India. They might have been, but in terms of the historical study of Buddhism, tantras reveal more about the monastic tradition of Buddhism than they do the ascetic tradition. The reason for this is simple. In the lat- ter first millennium ce, the vast bulk of the sangha in India lived within monasteries. Tantras were preserved in monastic libraries, read by monas- tically based scholars, and taught to novitiates as part of their education. Whether or not the rituals and activities described in the tantras ever occurred, the accounts of transgressive Tantric rituals must have reso- nated with the monastically based sangha—even if most never engaged in them. For the bulk of the sangha, tantras might have been intended more as allegories than as manuals for proper ritual behavior. In some sense, tantras might have been studied by the sangha for the same reason that Jack Kerouac’s On The Road is read by American high school students. Both the tantras and On the Road provide powerful lessons about life, but lord help any students who actually lived their lives according to either of them.

Just as modern academics can read On the Road and write articles, dis- sertations, and books about it, so Buddhist scholastics studied tantras. In scholastic Buddhism, Tantric sex could be recast as the “union of female Wisdom with male Skill in Means” (White 2000:17). In some Tantric texts, sex was limited to those with the greatest spiritual prowess; in others, sex was part of the initiation of novitiates. Further, sexual union need not literally occur, but could be meditated upon or actively visualized. Within the scholastic tradition of Tantric Buddhism, sex and other transgressive rituals were often treated as a metaphor to ponder, rather than an act to perform. In fact, among the monastic elite, Tantric Buddhism increasingly centered on the practice of meditative visualizations of a complex cosmo- gram called a mandala.

Mandalas were exceedingly complex and varied symbols. In the most

general sense, they were a cosmogram, a map of the universe. In some cases, mandalas depicted the sacred Mount Mehru in the center, with lesser peaks, continents, and oceans surrounding it. In other cases, the center was a palace. Typically, a major deity or Buddha inhabited the cen- ter of the mandala, with lesser deities or Bodhisattvas arrayed around him or her. There is no single mandala from which all others derive, but rather a huge number of different types of mandalas used by different sects, in different ways, throughout much of South, Southeast, and East Asia. In Indian Tantric Buddhism, mandalas often served as a meditation device—with a devotee visualizing a mandala with himself or herself at the center.

Overall, Tantric Buddhism was neither monastic nor ascetic, neither worldly focused nor otherworldly focused—it was all of this and more. However, due to the vagaries of historical and archaeological preserva- tion, far more is known of monastic Tantric Buddhism than ascetic Tantric Buddhism. Tantras were preserved by the monastically based sangha in monasteries that have been archaeologically identified. As discussed in earlier chapters, at present, no archaeological traces of ascetic Buddhism have ever been found in India. For this reason, the ascetic tradition can only be known through the interlocutors of monastic textual accounts. As discussed in Chapter 5, it appears that asceticism began to be valorized in Mahayana texts beginning in the third or fourth century ce. Based on readings of Tantric texts, this process accelerated in the mid- to late first millennium ce.

While there are numerous examples of the valorization of the ascetic life in Buddhist tantras, the story of Naropa will suffice as an example (Lopez 2001:214–215). Naropa was a renowned scholar and abbot of Nalanda in the tenth century ce. One day, as the account reports, while outside the monastery he encountered an old woman who “laughed at him mockingly, claiming that his knowledge of the dharma was merely intel- lectual, that he had no true understanding of the path” (Lopez 2001:214). She told him to apprentice himself to her brother, Tilopa, who turned out to be a beggar in a distant village. During the long apprenticeship, Tilopa demanded that Naropa perform several transgressive acts, including vio- lently attacking the members of a wedding procession and, later, marry- ing a woman himself. Through these acts, and the supernormal abilities of Tilopa, Naropa gained a more complete understanding of Buddhism than possible through his scholastic training.1

The story of Naropa illustrates numerous elements of Tantric Buddhism. First, the superiority of Tantric Buddhism over Mahayana Buddhism is demonstrated by Naropa’s achievements only after abandoning the scho- lastic Mahayana Buddhism of Nalanda. The second element of Tantra is shown in the old woman’s criticism of scholastic Buddhism as being ster- ile intellectualism, with the more mystical teachings of Tantric Buddhism more efficacious. Third, the revelatory rituals of Tantra often required transgressions of social or monastic norms and produced supernormal abilities for those who mastered Tantric rituals. Finally, the asceticism championed in Tantra is more worldly focused. Ascetics, as described in the tantras, were not living in the forests or meditating in caves. Tantric

  1. Specifically, Naropa came to understand the six yogas, a foundational concept in Tantric Buddhism. See Lopez (2001) for a more complete discussion of the Six Yogas of Naropa.

    ascetics lived at the peripheries of cities and villages, interacting with the laity on a daily basis, if for no other reason than to beg for food. In this way, Tantric ascetics had more contact with the laity than those members of the sangha who lived within the monastery walls.

    It must always be remembered that the existing accounts of Tantric asceticism were preserved and studied by scholastic Buddhists living within monasteries. The descriptions of Tantric asceticism cannot be straightforwardly read as simple descriptions of the lifestyles of Buddhist ascetics. In fact, these accounts, by themselves, cannot even be used to establish whether Buddhist ascetics actually existed in India, and if they did, how common they were. Despite this, it does seem likely that Buddhist ascetics were present in India in the mid- to late first century ce. In part, this is shown in the religious literatures of Jainism and Hinduism, which also lionize asceticism. Ascetics are also described in the kingly literature, though often in more disparaging terms. For example, the Mattavilasa Prahasana, an early seventh century Sanskrit play attributed to Pallava King Mahendravarman I (571–630 ce), is a comedic account of a drunken Shaivite ascetic accusing a Buddhist ascetic of stealing his begging bowl, only later to find out it was stolen by a dog (Lockwood and Bhat 1981). However negative the portrayal of asceticism in Mattavilasa Prahasana, the humor in the play would make no sense if ascetics were not a regular presence in Indian towns and villages in the latter half of the first millen- nium ce. The best evidence for the existence of Buddhist ascetics, how- ever, does not come from the secular literature from the first millennium ce, but rather from the similarities between Buddhist, Hindu, and Jain tantras themselves.

    From the start, Tantra was a highly heterodox and syncretic religious movement, encompassing Buddhism, Jainism, Hinduism, and other major and minor religious sects. Despite the divisions between rival reli- gious orders, Tantric ideas and practices were freely adopted and shared. As argued by White (2000:23),

    Within India, we may take the example of an early tenth-century Jain Tantra entitled the Jvalini Kalpa. This text . . . is in nearly every respect identical to Hindu and Buddhist Tantric sources of the same period. Nothing but the names of the deities invoked, visualized, or manipulated in these practices is specifically Jain; all the features, however, are specific to tenth century Indian Tantra, whether Hindu, Buddhist, or Jain.

    The obvious question these similarities raise is, how did they come to be? How were Tantric ideas so freely shared by rival religious orders? There are several potential answers to this question. First, throughout the

    religious literature of the period great scholars are noted as being great debaters. These debates, often conducted before kings, might have pro- vided a venue for the sharing of Tantric ideas among rival sects. Second, at locations like Ellora, where Buddhist, Hindu, and Jain religious institu- tions existed side by side, texts and religious ideas may have been shared. Finally, these ideas may have been shared among ascetics of different reli- gious orders living outside religious institutions. That is, Tantric ascetics had less interest in religious affiliations than those living within religious orders, and could thus freely learn and borrow texts and ideas from each other. Over time, these ideas and texts would have percolated up to the monasteries. This last possibility would, in part, explain the more worldly focus of Tantric texts. It is not necessary to choose one of these possibili- ties as correct. Rather, it is likely that all occurred in the mid- to late first millennium ce.

    Between the seventh and twelfth centuries in India, two strands of Buddhism become more clearly identifiable. The first was a scholastic ver- sion of Buddhism that emphasized the mastery of texts, housed in the large monasteries in the Gangetic Plain and Northeast India. The sangha living within these monasteries were largely divorced from day-to-day interaction with the laity, except as landlords over increasingly large monastic properties. The second strand was more ascetic, with practi- tioners abandoning the monasteries for a more revelatory and worldly focused form of Buddhism. Practitioners of this form of Buddhism lived on the margins of villages with ascetics of numerous faiths, depending on the donations of the laity for whom they served as ritual specialists. For individual members of the sangha, these two strands of Buddhism allowed for two different ways to balance their individual and communal desires. Some members of the sangha preferred the communal life of the monas- tery, while others preferred the ascetic life of the forest. Overall, how- ever, Buddhist ascetics were far less numerous than scholastic Buddhists, and over time ascetics became less identifiably Buddhist as their ritual practices took on more Tantric elements. All that said, the scholastic and ascetic strands of Buddhism were never wholly distinct, with ideas and people constantly moving between the two in much the same manner as Tambiah (1976, 1984) and Carrithers (1979, 1983) describe in Thailand and Sri Lanka.

    It must be remembered that all of these theological and ritual changes were occurring as the laity in much of India was increasingly abandoning Buddhism for other faiths. For the most part, the development of Tantric Buddhism had little impact in India beyond the Gangetic Plain, Orissa, and the Northeast. In some sense, the ascetic turn in Buddhism can be

    seen as an attempt to reconnect with the laity in the last remaining parts of India where the laity still had some interest in Buddhism. In the thir- teenth century, however, this attempt failed, as monasteries were aban- doned and ascetic Buddhism slowly lost its distinct identity within the increasingly heterodox world of Indian Tantra.


    As Buddhist pilgrimage centers and viharas were slowly abandoned or were converted to Hindu and Jain centers in much of India in the late first mil- lennium ce, the Palas endowed or enlarged several large viharas in the Gangetic Plain and the Northeast (see Figure 7.1). In the Gangetic Plain, where monastic Buddhism had long flourished, Pala land grants allowed a few monasteries to greatly expand. In the eastern portions of their ter- ritory, in Bengal and modern-day Bangladesh, the Palas endowed several new monasteries. Architecturally, the viharas in both regions continued to follow the ubiquitous layout of a large courtyard surrounded by monastic cells. In the Gangetic Plain, expansion consisted of the creation of multiple viharas abutting each other. In the east, new monastic centers were designed with massive viharas with large central courtyards containing monumen- tally scaled temples. In either case, the design of these new viharas allowed for a substantial increase in the monastic population while maintaining, or even possibly extending, the sangha’s isolation from the laity.

    Late Buddhist Monasteries in the Gangetic Plain

    In the Gangetic Plain, the Palas and other royal dynasties increased their support of Buddhist monasteries in the latter portions of the first millen- nium ce and the beginning of the second millennium ce. For example, at Sarnath one inscription records the restoration of the main stupa in 1026 ce, while another records the donation of a new vihara by Queen Kumaradevi in c. 1114–1154 (Mitra 1971:68). The main focus of support in the Gangetic Plain, however, was Nalanda. Perhaps founded as early as the Gupta Period, Nalanda continued as the most important Buddhist monastery through at least the tenth century ce. Due to its international prominence, there are more historical accounts of Nalanda than any other contemporary monastery. Prominent among these sources are the accounts by Xuanzang in the beginning of the seventh century and by Yijing several decades later. Whatever the source, all describe Nalanda as





    Bodh Gaya


    Somapura Nalanda, Odantapuri


    Ratnagiri, Udayagiri, Lalitagiri


    Arabian Sea

    Bay of Bengal

    Figure 7.1: Archaeological sites discussed in Chapter 7

    a large and dynamic center for the training of monks from across Asia, what might be called a university today. The dynamism of Nalanda was in part due to its intellectual eclecticism. While most associated with Mahayana Buddhism, Nalanda housed monks espousing Hinayana and, later, Tantric Buddhism as well. According to Yijing, who spent ten years at Nalanda (c. 675–685 ce), the “secular” education at Nalanda even drew students who had no intention of entering the sangha (Rongxi 2000).

    Excavations at Nalanda have revealed eleven multistory viharas and several large temples (Ghosh 1959; Mitra 1971; Stewart 1989). The main temple was rebuilt and enlarged several times, with the latest construc- tions over 15 meters high. Both the main temple and other temples con- sisted of shrines containing a Buddha image surrounded by subsidiary images, smaller shrines, and votive stupas. All of the viharas at Nalanda follow the standard square layout. All but two lie next to each other in a line running north/south (see Figure 7.2). This layout, similar to the layout of monasteries at Sarnath, suggests that Nalanda grew by gradual accretion. That is, where initially the site had a single vihara, each succes- sive vihara abutted an earlier vihara as the monastic population grew over time. All of the monasteries were rebuilt or refurbished several times dur- ing the occupation of the site, though each subsequent vihara preserved the ground plan and layout of its predecessor. For example, excavations at Monastery 1 revealed at least nine different reconstructions of the origi- nal monastery (Mitra 1971:87).

    As in earlier periods at other sites, each vihara at Nalanda housed a prominent Buddha image in a special chamber opposite the entrance. Unlike earlier periods in India, where the role of Buddha images for the sangha must be inferred from their design and placement, Yijing’s (Rongxi 2000:135) account of his time in India provides a detailed description of image worship at Nalanda.

    Although the Great Teacher has entered nirvana, his image still exists, and we should venerate it as if he were in the world. Incense and flowers should often be offered to the image; by doing so our mind may be purified. To bathe it regularly is good enough to clear away the evil influence of our deeds caused by idleness.

    Figure 7.2: Nalanda (c. fourth–early thirteenth centuries ce; after Mitra 1971)

    Yijing continues by describing an elaborate ceremony held in the court- yards of viharas where images were cleaned with perfumed water and scented pastes (Rongxi 2000:135–137). Yijing specifically notes that one monk would go from vihara to vihara each evening, leading a ceremony worshiping and praising the Buddha in “shrine halls” in “every courtyard” (Rongxi 2000:140).

    One may sit alone, facing the shrine, and praise the Buddha in his mind, or go to a temple and kneel on the ground together with others to chant in a high pitch. One may then put one’s hands on the ground and touch it with one’s head three times. This is the traditional ceremony of paying homage practiced in India.

    Critically, archaeological excavations at Nalanda have shown that all of the shrines at the site center on Buddha images. Thus, when Yijing describes monks facing shrine halls or going to a temple, he is describing image worship.

    Based on the archaeological and textual information, it is clear that the monks and novitiates at Nalanda were heavily invested in Buddha images—and worshiped the Buddha while facing images both within their viharas and in the four temples located immediately to the west of their viharas (see Figure 7.2). What is less clear is the degree to which the laity were allowed to also worship in the temples at Nalanda. Was Nalanda like Sarnath, where the vihara walls served to isolate the sangha from the laity who frequented the pilgrimage site? Or were the temples at Nalanda strictly reserved for residents of Nalanda itself? There is no clear answer to this question. There is, however, one major difference between Sarnath and the other large pilgrimage sites in the Gangetic Plain and Nalanda. At Nalanda, a massive wall surrounded all of the viharas, temples, and bathing pools. More so, Xuanzang states that the guards at the gate of Nalanda were trained to quiz all who sought to enter the monastery and “only those who [were] well-versed in both ancient and contemporary learning [could] gain admittance” (Rongxi 1997:283). Xuanzang goes on to describe numerous sacred sites outside the wall of Nalanda that were frequented by the laity (Rongxi 1996:285).

    It is not possible to definitively state whether all the laity, or perhaps only the non-elite laity, were excluded from Nalanda. Perhaps Xuanzang’s account was meant only to refer to those who came to study at Nalanda, but not those who merely sought to worship there. Whatever the case, the enclosure walls surrounding Nalanda suggests that, unlike Sarnath and other pilgrimage centers in the Gangetic Plain, the resident sangha

    determined who could, or could not, worship there. This control over access was even more pronounced in the Northeast.

    Late Buddhist Monasteries in the Northeast and Orissa

    Buddhism was established in the Northeast and Orissa from at least Mauryan times, but in contrast to other regions of India, there was a sig- nificant increase in the number of monasteries beginning in the second half of the first millennium. Faxian spent two years studying in a mon- astery in Tamluk, a major seaport in the Gangetic Delta near the Bay of Bengal. Both Xuanzang and Yijing similarly passed through Tamluk in their travels. All reported that the region had numerous monasteries with resident populations in the hundreds. The character of Buddhism in the Northeast was heavily influenced by the region’s extensive religious, polit- ical, and economic ties with Southeast Asia, the Himalayas, and Sri Lanka. This influence can be clearly seen in the design of several monasteries con- structed in the Northeast during the reign of the Pala King Dharmapala (c. 770–810).

    Unlike the monasteries in the central Gangetic Plain that grew by accretion, Pala period monasteries in the Northeast appear to have been designed and built in one coherent construction phase. Rather than mul- tiple viharas and temples built in succession, the new monasteries in the Northeast consisted of single massive viharas with large temples erected in their courtyards. Viharas following this layout are found at Vikramasila and Mainamati, but are best preserved and studied at Somapura in mod- ern Bangladesh (see Figure 7.3). Somapura consisted of an almost eleven hectare square courtyard surrounded by about 177 cells (Mitra 1971). Within the center of the courtyard was a large, terraced temple with a cruciform layout. Though poorly preserved, the ground plan and lower tiers of the temple at Somapura strongly resembled contemporary designs from the ninth to thirteenth century city of Pagan in modern Myanmar (Burma). Assuming the superstructure of the temple at Somapura was similar to those found in Pagan, the focus of the temple would likely have been a massive, highly attenuated stupa. Surrounding the stupa would have been terraced galleries lavishly decorated with Buddha images and narratives of the Buddha’s life and previous lives. It is possible, however, that the central temple might also have housed an image. Given the lack of preservation, it is impossible to decide between these two possibilities. Excavations by Patna University and the Archaeological Survey of India at the site of Vikramasila in far eastern Bihar have revealed a roughly ten

    Figure 7.3: Somapura (c. eighth–thirteenth centuries ce; after Mitra 1971)

    hectare vihara with roughly 200 monastic cells arrayed around a large central courtyard (Dutt 1988; Prasad 1987). Though published reports on Vikramasila are limited, it appears much like Somapura, with a large terraced cruciform platform located in the center of the courtyard. The platform was crowned by a large stupa with four stucco Buddha images placed in small mandapas at the cardinal points. Just as in Southeast Asian stupas, the two terraces served as circumambulatory paths.

    Both Somapura and Vikramashila are known historically through Tibetan sources, mostly Taranatha’s early seventeenth-century History of Buddhism in India (Chimpa and Chattopadhaya 2010; Dutt 1988). By these accounts, Vikramasila had surpassed Nalanda as the premier uni- versity in the ninth through eleventh centuries ce, with roughly 100

    instructors and about 1,000 students. Taranatha noted Somapura, Odantapuri, and Jaggadala as the other major Buddhist centers. Unlike Nalanda, where Mahayana Buddhism dominated, Vikramasila and Somapura were noted as centers for Tantric Buddhism, and both had close ties with Tibetan Buddhists. This may partially explain Taranatha’s preference for Vikramasila over Nalanda, but the possibility that he was right should not be discounted. In either case, the new Buddhist centers founded in the Northeast during Pala times had significantly more foreign influence—both in terms of the theology and architecture—than those monasteries in the central Gangetic Plain. Unlike the Northwest, where Islam had replaced Buddhism as a trade religion, Buddhism continued to bind the Palas with distant ports in the Bay of Bengal, Southeast Asia, and the Himalayas. These ties also provided the sangha with ready routes to leave India when their royal benefactors were deposed, their lands taken, and their monasteries began to fail.

    Architectural Analysis

    Like the wall encircling Nalanda, the layout of Somapura, Vikramasila, and other similar monasteries in the Northeast allowed the sangha to con- trol access to the shrines, temples, and stupas within. Though, just as with Nalanda, it is not possible to determine the degree to which the sangha in the Northeast welcomed or excluded different portions of the laity. There is simply no historical or archaeological information available to defini- tively address the question. That does not mean, however, that nothing can be known. However achieved, and whoever was allowed in or out, the ability of the sangha to control access to the major monasteries of the late first millennium and early second millennium ce represented a signifi- cant change in their power. Whereas the earliest monasteries were built adjacent to pilgrimage sites (e.g., Bodh Gaya, Sarnath, and Sanchi), or con- structed public worship spaces adjacent to their monasteries (e.g., Karla, Bhaja, and Thotlakonda), for the first time, large open-air temples were built within monasteries or, in the case of the Northeast, even within the sangha’s living quarters. For the first time, the gates of the monasteries served as the gates to worship itself.

    As discussed in Chapter 2, from the perspective of materiality, the design of a building or site is not simply a reflection of power, but rather part of the creation of power itself. Whether the sangha used the gates to admit or deny admission, the gates instantiated the sangha’s power over outsiders. This power is even more explicable when considering what

    the sangha was controlling access to—temples and, in the Northeast, the first new large stupas constructed in India in centuries. While the large open-air stupas at Sanchi, Sarnath, and other pilgrimage sites remained in use throughout the first millennium ce, the sangha had long since shifted their devotions to images placed within small cells in their viha- ras. At Nalanda, the viharas continued to have small shrines, but massive image shrines were also constructed outside the viharas, but within the monastery walls. At Vikramasila, and perhaps Somapura, the central temples were stupas rather than images. There are two ways to interpret the reintroduction of stupas to the architectural vocabulary of Northeast India. First, the stupas might have been constructed to win back the laity, and their donations, to the sangha. Second, the sangha themselves may have wanted to worship at stupas. It is also possible, of course, that the construction of the large stupa at Vikramasila was intended to do both. Whatever the case, the stupa constructed at Vikramashila, and perhaps Somapura, followed a radically different form than earlier monastic or pil- grimage stupas in India.

    Where earlier stupas in India had been icons of the Buddha’s burial

    mound indexing his relics and prominent locations in the Buddha’s life, the stupa at Vikramasila was all that plus a mandala, a cosmogram of the Buddhist universe. The anda, thus, was an icon of Mount Mehru. The four stucco Buddha images facing in the four directions were symbols of the all-seeing Buddha, the four lobes of the cruciform terraces were icons of the four continents. In this sense, the stupa at Vikramasila was a mate- rialization of Mahayana, and particularly Tantric Buddhism, meditative practices.2 This, in turn, might suggest that Mahayana and Tantric mem- bers of the sangha were at least one of the intended audiences of the stupa at Vikramasila, and perhaps Somapura (assuming that the cruciform platform at Somapura supported a stupa). Whatever the case, the stupa at Vikramasila was a staggeringly multivalent sign, with, in a semiotic sense, significant emotional and intellectual impact.


    The decline of Buddhism in India was not a singular event, with a singular cause; it was a centuries-long process that unfolded in a patchwork. The

  2. Stupas in the form of mandalas were elaborated and perfected at contemporary Buddhist stupas at Borobudur in Java (c. ninth century ce) and later Angkor Wat in Cambodia (c. twelfth century ce).

    seeds of Buddhism’s decline began in the mid-first millennium ce, when the sangha began withdrawing into their monasteries and divorcing them- selves from day-to-day interactions with the laity. Into this spiritual void stepped Hindu and Jain sects, who revamped their ritual practices and religious architecture to more closely resemble traditional Buddhist prac- tices. In the South and West of India, Hindu and Jain sects increasingly earned the support of the political and economic elite. In the Western Ghats, the last major Buddhist temples were constructed at Ellora in the seventh and eighth centuries ce. Across South India, the sangha aban- doned Buddhist sites, many of which were later reoccupied by Hindus and Jains. While some small Buddhist centers still persisted in South and West India in the eleventh and twelfth centuries, for the most part, both monastic and lay Buddhism had been eclipsed and replaced by Hinduism and Jainism by the end of the first millennium ce.

    The history of Buddhism in Northwest India follows closely the pat- terns of the South and West. At Takht-I-Bahi, the last major constructions occurred in sixth and seventh centuries ce. By the late seventh and early eighth centuries, the Gurjara-Pratiharas actively supported Hindu sects. At the same time, the central Asian states to the north were increasingly adopting Islam, which was rapidly replacing Buddhism as the trade reli- gion along the Silk Roads. In the eleventh century ce, Northwest India was subject to numerous attacks by the Ghaznavids, an Islamic state based in Afghanistan. In 1001 ce, Mahmud of Ghazni took the city of Peshawar in modern Pakistan and captured the Shahi King Jayapala. Rather than hold the territory, Mahmud of Ghazni looted the city, ransomed the king, and returned to Afghanistan, using the wealth to finance his campaigns against rival states in Central Asia. In the eleventh and twelfth centuries, Mahmud and later Ghaznavid rulers continued raiding the Northwest, though after Mahmud, the raids became less regular and less destructive. During these raids, religious institutions, including Buddhist monasteries and pilgrimage sites, were regularly sacked and looted. Considering the persistent threat of Ghaznavid raids and the lessening support of the local laity and regional rulers, it is not surprising that many Buddhist monas- teries and pilgrimage sites in the Northwest were abandoned in the late first millennium and early second millennium ce.

    Unlike other portions of India, Buddhism and Buddhist monasticism persisted in central and northeastern India through the first few centuries of the second millennium ce. As discussed above, massive monasteries like Nalanda, Vikramashila, and Somapura continued to serve as centers of Buddhist scholasticism. Likewise, Sarnath, Bodh Gaya, and other major pilgrimage sites in the Gangetic Plain continued to draw lay devotees and

    the viharas at the pilgrimage sites continued to be occupied by the sangha. In the early centuries of the second millennium, Hindu and Jain institu- tions in Central and Northeast India were increasingly drawing lay and royal support, but Buddhist institutions still retained their prominence and vitality. This changed with a new invasion from Central Asia into Northwest India by Muhammad Ghori of the Ghurid Dynasty in the late twelfth century ce. Unlike the earlier Ghaznavid raids, Muhammad Ghori sought to establish Ghurid suzerainty in North India.

    The Ghurid campaigns in the Northwest began with an unsuccessful attack by Muhammad Ghori on Gujarat in 1178 ce. By 1181 ce, Ghori had established a fort in Peshawar, from which he conquered Lahore in 1186 ce. In 1192, Ghori defeated Prithviraj Chauhan, who ruled over Delhi and Rajasthan. By 1193–1194 ce, Delhi was itself captured. From Delhi, Ghurid forces rapidly spread across the Gangetic Plain, reaching Bengal by 1202–1203 ce. Muhammad Ghori was assassinated in 1206 ce, and Ghurid control over North India shifted to Qutb-ud-din Aibak, the first ruler of the Delhi Sultanate.

    Temple Desecration and the Collapse of Monastic Buddhism in India

    From the earliest raids by the Ghaznavids, through the later conquest by the Ghurids, the military campaigns in India included the desecration of Hindu, Buddhist, and Jain temples. In Islamic histories written sev- eral centuries later, the invading Ghaznavids and Ghurids are sometimes described as leading a holy war in order to bring Islam to India. In these histories, Muslim kings are celebrated for smashing false idols, destroy- ing temples, and converting Hindus to Islam. In the historiography of India, the thirteenth through fifteenth centuries are often depicted as a period when Islam was forcibly imposed on the native Hindu population. For British colonial historians, this depiction of Islamic despots served to illustrate the beneficence of British rule. Some postcolonial national- ist historians have used the presumed historical oppression of Hindus by Muslims to argue for a more Hindu, rather than secular, India. Buddhism has only a small place within these larger narratives of despotism, destruc- tion, and desecration. In the British colonial and Hindu nationalist histo- ries, the Muslim invaders are credited with destroying Buddhism—with this act serving as an example of Muslim depravity.

    Recently, Richard Eaton has re-examined the history of temple desecra- tion in India (Eaton 2000). He begins by noting that the accounts of temple

    desecration are typically found in eulogies of past kings written several centuries after their deaths. When looking at accounts and inscriptions dating from the period of conquest itself, Eaton finds that temple desecra- tion was much less common than British colonial and Hindu nationalist historians claim.3 Rather than thousands of temples desecrated, relying only on historical accounts and inscriptions more-or-less contemporary with the events they describe, Eaton found only 80 examples of temple desecration between the thirteenth and eighteenth centuries in India. More so, temple desecration was normally a political act intended to reduce the authority of a rival ruler rather than a religiously motivated action.

    When such authority was vested in a ruler whose own legitimacy was asso- ciated with a royal temple—typically one that housed an image of a ruling dynasty’s state-deity, or rastra-devatd (usually Visnu or Siva)—that temple was normally looted, redefined, or destroyed, any of which would have had the effect of detaching a defeated raja from the most prominent manifestation of his former legitimacy. Temples that were not so identified, or temples formerly so identified but abandoned by their royal patrons and thereby rendered politi- cally irrelevant, were normally left unharmed. (Eaton 2000:293)

    Rather than the wanton destruction of Hindu, Buddhist, and Jain tem- ples, Eaton sees a more surgical desecration of royal temples as part of a strategy of conquest. Further, the strategic use of temple desecration was not new to India. In the seventh century, a Pallava king looted a Ganesha image from a Chalukyan Royal Temple. In the eighth century, Bengalis destroyed a Vishnu image that served as the state deity in Kashmir. In the ninth century, a Pandyan king raided Sri Lanka, returning with a gold Buddha image. Using these, and other, examples, Eaton (2000:296) dem- onstrates that temple desecration was a common indigenous practice in India well before the arrival of the Ghaznavids and the Ghurids.4

    In short, it is clear that temples had been the natural sites for the contestation of kingly authority well before the coming of Muslim Turks to India. Not sur- prisingly, Turkish invaders, when attempting to plant their own rule in early medieval India, followed and continued established patterns.

  3. Just as I have argued with later accounts of Early Buddhist asceticism, the later Islamic accounts of large-scale temple desecration are inventions of tradition—histories created at a specific time, for a specific reason. More so, just as in the academic study of Buddhism, these inventions of tradition reveal more about the time they were composed than the time to which they refer.

  4. Eaton (2000:295) provides several additional examples of pre-Muslim temple desecration.

As argued in earlier chapters, the royal support and construction of temples had always been legitimations, attempts to legitimize ruling authority. Following Moore’s criteria, the endowment of massive temples served to associate rulers with the divine. The inverse of this process was temple desecration. If endowing a temple was a legitimation, the destruc- tion of a temple was a delegitimation, an attempt to reduce the legitimacy of a rival king. Royal temples were not simply reflections of kingly power, but rather the construction and use of royal temples created and preserved kingly power. Likewise, the destruction of a royal temple was not simply a reflection of his defeat, but an action that helped ensure his defeat.

Eaton’s analysis of temple desecration is an important corrective to British colonial and postcolonial histories that portray Muslim states in India as focused on the imposition of Islamic hegemony by force. If noth- ing else, Eaton’s analysis explains how Hindus and Jains, and many of their temples, survived in India to the present day. What Eaton’s analy- sis does not do, however, is explain why Buddhism did not. While royal temples were destroyed, Hindu and Jain institutions focused on the laity were left unmolested. The same could have occurred with Buddhist insti- tutions focused on the laity, had they existed. However, by the thirteenth century ce, Buddhist monasteries in the Gangetic Plain and northeast- ern India were prominently supported by local and regional kings, and their relations with the non-elite laity consisted of little more than serv- ing as landlords. As argued by Eaton, “Detached from a Buddhist laity, these establishments had by this time become dependent on the patron- age of local royal authorities, with whom they were identified” (Eaton 2000:297). Accordingly, Nalanda, Vikramasila, and Odantapuri were sacked in c. 1202 ce, during the initial Ghurid conquest of Bihar.

While Eaton sees the desecration of Buddhist monasteries and pilgrim- age sites as primarily due to their close affiliation with royal families, it is also likely that the monasteries were viewed simply as major landown- ers with no significant lay constituency. Those monasteries spared during the initial wave of conquest in the beginning of the thirteenth century still lost control over the monastic lands as Indo-Muslim states expanded outward from the Gangetic Plain in the thirteenth through fifteenth cen- turies. Without the support of lay Buddhists to fall back on, these monas- teries were utterly dependent on the revenue generated by their monastic lands. Thus, even those monasteries that were not destroyed in the ini- tial wave, like Somapura in modern Bangladesh and Lalitagiri, Udayagiri, and Ratnagiri in Orissa, were abandoned by the sangha once they were unable to sustain themselves. Whether by violence or loss of revenue, Buddhist monasticism barely survived the thirteenth century in India,

with only a few vestigial monasteries in South India surviving until the fifteenth century.

The Decline of Lay Buddhism in India

The decline and abandonment of Buddhist pilgrimage centers was a slower process than what occurred to monastic centers like Nalanda. Across the Gangetic Plain, the viharas at major pilgrimage sites were desecrated and looted early in the Indo-Muslim period due to their extensive landhold- ings and close ties to local rulers. The temples and stupas at pilgrimage sites, however, did not receive the same treatment. Inscriptions by for- eign pilgrims at Bodh Gaya demonstrate that the Mahabodhi temple was in at least some use through the fourteenth century ce (Mitra 1971:65). According to the Taranatha’s seventeenth-century History of Buddhism in India, in the mid-fifteenth century, a Bengali queen restored the Mahabodhi temple, which had been completely abandoned (Chimpa and Chattopadhaya 2010; Mitra 1971:65). Sometime after this, the ownership of Bodh Gaya passed to local landowner, and the Mahabodhi temple was converted to a lesser Shaivite center. Less is known about the afterlives of the other prominent Buddhist pilgrimage sites, though it appears that many followed the same pattern of decline. Unlike the viharas, the stu- pas and temples of Buddhist pilgrimage sites were not desecrated by the Indo-Muslim armies for the same reason that non-royal Hindu and Jain temples were left unmolested—they were not material legitimations of rival royal families. As such, they were left mostly unharmed and small trickles of worshipers likely continued to frequent them.


In the centuries after the collapse of monastic Buddhism in India, Buddhist sites were either abandoned and left empty or were reoccupied by other religious orders. It is hard to say what happened to the bulk of the sangha in India. The decade between Mohammed of Ghori’s first push into the Gangetic Plain in 1192 ce and the desecration of Buddhist mon- asteries in Bihar in c. 1202 ce is quick by historical standards, but would have given the sangha ample time to escape. While some holdouts may have died in Indo-Muslim attacks on the great monasteries, for the most part, the sangha knew the attacks were coming and fled well before the Indo-Muslim armies reached their gates. Without viharas and libraries,

the tradition of scholastic Buddhism, as well as some of its Indian practi- tioners, moved to the Himalayas, China, and Southeast Asia. Other mem- bers of the sangha may have reverted to secular life. Still others may have adopted the life of wandering ascetics.

Without property or royal association, the Central Asian Turks and later Indo-Muslim states would not have targeted Tantric Buddhist ascet- ics. Buddhist ascetics could continue living in the margins of villages, or even within the ruins of Buddhist viharas and pilgrimage centers. After the destruction of monastic Buddhism, the long-term tension in the sangha between the individual and the group, between monasticism and asceticism, was resolved once and for all. There was no longer a defin- able Buddhist group to which ascetics could belong, even if they wanted to. Without monasteries of their own, Buddhist ascetics lived in a world populated by Hindu and Jain ascetics, surrounded by Hindu and Jain temples and Islamic mosques. Any desire for group affiliation, however small, could be met only by affiliation with the surviving, non-Buddhist religious sects. In this context, Buddhist ascetics were rapidly absorbed into the larger religious milieu of medieval India. By the fifteenth century ce, the traces of Buddhism survived only in the practices of Hindus, Jains, Muslims, and others, Buddhist texts and inscriptions, and the ruins of Buddhist monasteries and pilgrimage centers.






hen I first began studying the archaeology of Indian Buddhism in graduate school almost twenty years ago, my peers in Buddhist studies were excited by what I might add to their research. They saw my work as potentially fleshing out some of the details missing in their tex- tual sources. As I began to gain my footing and present what I was learn- ing, their excitement rapidly waned. It wasn’t that my work contradicted theirs; it’s that my work didn’t fit into their framework at all. The questions and debates that oriented their studies were not the same as the questions and debates that oriented mine. Some of the more intractable problems in Buddhist studies seemed simple from my archaeological perspective, and some of the greatest difficulties I faced seemed obvious when examined from a historical perspective. By the time I completed my PhD, it seemed that I had little to offer textual scholars and, I should admit, I felt they had little to offer me. It has taken me almost twenty years to correct my mistake—to learn how to combine the strengths of textual and archaeo-

logical sources to study the archaeological history of Indian Buddhism.

What I have learned in the last twenty years is, at some level, decep- tively simple. Archaeological evidence provides information on what people did, while historical sources provide information on what people thought (Fogelin 2013). The conceptual hurdle in archaeology is figur- ing out why people did the things they did. In contrast, the conceptual hurdle in history is figuring out how people’s thoughts translated into action. To be clear, archaeologists have developed numerous methodolo- gies to infer past beliefs, and historians know full well that they must carefully interrogate multiple sources to discern specific historical events. Neither archaeologists nor historians are ignorant of the limitations of

( 225 )

their evidence, but they both tend to be more critical of other people’s sources than their own.

I do not deny that the archaeological history presented in this volume is biased toward the archaeological sources that I favor. My bias is shown in the subjects that I choose to emphasize. In a general sense, Indian Buddhism can be studied from the perspective of history, archaeology, or a combination of the two. As an archaeologist, I have focused only on the subjects that can be addressed by archaeology alone, or in combination with history. Subjects that are only revealed in historical sources have generally been ignored. For example, throughout this archaeological his- tory, the order of Buddhist nuns has been almost completely absent from my discussions. While there is extensive textual and epigraphic evidence for Buddhist nuns, at present, no archaeological evidence of Buddhist nuns has been discovered in India.1 While I have no doubt that nunner- ies existed, and that the order of nuns mattered greatly in the history of Buddhism, I have simply been unable to crack this problem. With luck, another archaeologist soon will.

The inability of archaeology to contribute to all issues that can be cov- ered in textual studies, however, should not be taken to mean that archae- ology has nothing to offer textual studies at all. Among the central claims of this archaeological history is that the earliest Buddhist texts—the texts used to create the standard history of Buddhism—were inventions of traditions rather than straightforward historical accounts. The depic- tion of early Buddhist asceticism in the earliest Buddhist texts had more to do with the concerns of the early to mid-first millennium ce than with the actual practices of the sangha in the mid- to late first millennium bce. Where early Buddhist textual sources project the ascetic tradition onto the distant past, archaeological evidence demonstrates that the sangha was fully domesticated from at least the third century bce. This does not mean, however, that the earliest Buddhist texts were wrong, or that they can be completely disregarded. Rather, by challenging the historical verac- ity of these textual sources, it allows these texts to illuminate a period of profound change in Buddhism in the early to mid-first millennium ce. This period marked a major transformation in the way the sangha balanced their contradictory communal and ascetic desires through the creation of

  1. In the last forty years, archaeologists have made significant progress in the study of gender, developing robust methods and identifying material signatures of gender and gender relations in archaeological contexts. While I have not been able to apply these insights to the archaeological record of Indian Buddhism, it is my sincere hope that another archaeologist will succeed where I have not.

Mahayana Buddhism, Buddha images, and the ascetic ideal. This revision to the standard history of Buddhism then demands a reinterpretation of what came before and after.


With a few possible exceptions, neither archaeological nor historical sources provide insight into the origins of Buddhism. Of the potential archaeological evidence for the practice of early Buddhism, Coningham and Acharya’s recent identification of a wooden shrine centered on tree at Lumbini is the most convincing (Coningham et al. 2013). While this research may suggest that the Buddha’s death occurred in the sixth cen- tury bce, the most important aspect of the research is that it confirms the long-held assumption that the earliest Buddhist temples were made of wood, and that later stone temples were modeled on wooden prototypes. This research also suggests that in the first few centuries of Buddhism, the primary ritual foci may have been trees rather than stupas. By the last few centuries bce, however, Buddhist ritual progressively centered on stupas, both at pilgrimage sites and at monasteries.

By at least the second or third century bce, the form and layout of Buddhist pilgrimage sites had become standardized, with a large central stupa surrounded by circumambulatory paths, railings, courtyards, and subsidiary stupas and shrines. This layout allowed for individual ritual to be performed in the pradakshina adjacent to stupas, while allowing the simultaneous practice of communal ritual in the courtyards surrounding them. Over the next two millennia, the Buddhist laity in India continued to frequent many of the earliest large pilgrimage complexes, and later pil- grimage centers continued to be constructed in pretty much the same way. This long-term stability in the design and use of pilgrimage centers dem- onstrates a fundamental conservatism in the practice of lay Buddhism. By at least the second or third century bce, the Buddhist laity found a suc- cessful balance between their individual and communal desires and stuck with it for almost two millennia.

In contrast to the laity, the sangha seemed to continually alter and transform their ritual practices and theology. This religious restlessness is apparent from the very start of the archaeological sequence of monas- tic Buddhism. Initially, the sangha altered and manipulated their stupas in order to make them appear taller and more massive. These legitima- tions asserted the sangha’s authority over the laity who came to worship at their monasteries and sped the flow of donations. In the process, the

sangha began to alter the basic form and meaning of their primary reli- gious symbol—alienating their attenuated and implied mass stupas from the ancestral earthen stupas on which they were based. These shifts did not, however, significantly alter the methods used to balance their indi- vidual and communal desires. From the first century bce through the first few centuries ce, the sangha created multiple spaces that could alterna- tively satisfy their individual and communal desires. Monasteries in this period consisted of both viharas and more public worship spaces, the latter consisting of chaityas, open-air stupas, or both. In the viharas, communal events could occur in the square courtyards while individual meditation could be practiced in the monastic cells. In the chaityas, circumambulation could be practiced along the periphery of the structure while communal ritual could be practiced in the center. In those monasteries with larger, open-air stupas, the separation of individual and communal ritual was achieved in the same way as pilgrimage centers. Overall, between the first century bce and the first few centuries ce, the sangha satisfied their indi- vidual and communal desires in pretty much the same way as the laity, by creating separate spaces in which both could be practiced simultaneously. This all began to change in the early to mid-first millennium ce.

By the beginning of the first millennium ce, Buddhist monasteries had become wealthy, self-sufficient institutions that, for the first time, could seclude the sangha from day-to-day interactions with the laity. Rather than interacting with the laity as religious interlocutors, the sangha supported themselves as landlords. Where previously the sangha had to balance their desires for seclusion with the need to support and feed themselves, now the sangha could withdraw into their viharas and lead a life of seclusion. It is in this context that many of the most pro- found changes in Buddhism occurred, including the development of Mahayana, the creation of Buddha images, and the invention of the ascetic ideal. Each of these developments originated from, and served the interests of, the newly reclusive sangha. Where the Buddha had previously been portrayed as being worshiped by the laity, new images increasingly depicted him as a solitary figure; meanwhile, the sangha progressively abandoned the public worship spaces at their monasteries (e.g., chaityas and open-air stupas), placing images within monastic cells inside their viharas. At the same time, Mahayana scripture emphasized the existence of an innate Buddha nature and the value of solitude and meditation. Together, these developments were legitimations of the new ascetic ideal. Critically, however, while the sangha valorized asceticism, few, if any, practiced it.

Unlike other parts of Buddhist Asia, there is no archaeological evi- dence for forest monks in India. This absence is striking when compared to the ample evidence of forest monks in Sri Lanka and Thailand. While it is possible that the archaeological evidence of Buddhist asceticism has not yet been found in India, at some point negative evidence becomes significant—at some point it must be asked whether asceticism was ever a significant part of Buddhist practice in India, and if so, when it began? Further, even if some members of the sangha did abandon the monaster- ies for the life of the forest, why were accounts of forest monks preserved and valorized in the scholastic literature of the monastically based sangha in the first millennium ce? Whether or not Buddhist asceticism was actu- ally practiced, the accounts of Buddhist asceticism had meaning and rel- evance to the domesticated sangha—it relieved the individual desires of an increasingly isolated and communal sangha. While a monk might live in the tight confines of a vihara with hundreds of other monks, he could at least read about monks who did not.

While the ascetic ideal was likely invented to ameliorate the ascetic desires of the domesticated sangha, there is ample textual evidence that at least a few members of the sangha abandoned monasteries beginning in the mid to late first millennium ce. This coincided with the develop- ment of Tantric Buddhism. To be clear, there is no more archaeological evidence for Buddhist asceticism in the latter part of the first millennium than there is for the earlier periods. Rather, the commonalities in Tantric texts among Buddhists, Jains, Hindus, and, later, Muslims suggest inter- action and borrowing among practitioners of these different faiths. It is possible, given the ascetic character of Tantric texts, that this borrowing and blending occurred as ascetics of rival sects lived in proximity to one another, perhaps in cremation grounds and at the margins of Indian vil- lages. That said, at sites like Ellora in Maharashtra, there is clear evidence of major institutions of rival religious sects existing in close proximity to one another. It is possible, then, that the commonalities between Tantric texts were the product of the interaction between rival religious institu- tions, not ascetics in the peripheries of society. In either case, just as in ear- lier periods, Tantric texts were preserved and read in monastic contexts. It is also clear, based on numerous traveler accounts, that by the latter half of the first millennium ce, the great Buddhist monasteries in North India housed substantial numbers of Tantric scholars. Just as with Mahayana Buddhism, the study of Tantric Buddhism within Buddhist monasteries served as a safety valve for the expression of the ascetic desires of the communal sangha.

With the increasing seclusion of the sangha over the course of the first millennium ce, the Buddhist laity increasingly abandoned Buddhism in favor of other religious orders. This shift was fostered by the appropria- tion of Buddhist ritual and symbolism by nascent Hinduism and other religious sects. This is shown in both the gradual transformation of linga into forms resembling stupas, the assertion that the Buddha was an avatar of Vishnu, and the appropriation of Buddhist principles into the Tantric practices of rival religious orders. Given all of this, to the average layper- son, the shift from Buddhism to rival religious orders may not have felt like much of a shift at all. Lay Buddhists continued worshiping in much the same way as they always had, and often even in the same places they always had. Even in their conversion to rival religious orders, the laity con- tinued to be highly conservative. Overall, it was the laity, not the sangha, who preserved and practiced the earliest forms of Buddhism most faith- fully for the longest time—perhaps even after they nominally abandoned Buddhism for other faiths.

The increasing wealth of the Buddhist sangha in the early to mid-first

millennium ce, and the isolation that wealth allowed, fostered the decline of Buddhism in India in the late first millennium ce and the early sec- ond millennium ce. Buddhist monasteries first began to decline in the South, West, and Northwest of India, in part due to the adoption of Islam in the Persian Gulf and Central Asia. Buddhism had long facilitated trade relations in India, particularly with the Silk Road in Central Asia. As India’s trade partners increasingly adopted Islam, the economic value of Buddhist monasteries declined, with a corresponding decrease in finan- cial support. Combined with the abandonment of Buddhism by the laity, these economic factors led to the increasing abandonment of Buddhist institutions in the South, West, and Northwest toward the end of the first millennium ce.

Buddhism persisted, often with substantial state support, into the second millennium ce in the Gangetic Plain, Orissa, and the Northeast. In part this was due to the continued value of Buddhism in facilitating trade with still-Buddhist Himalayan and Southeast Asian states. With the invasion of North India by Central Asian Turks in the beginning of the second millennium, however, Buddhist monasteries in the Gangetic Plain, Orissa, and the Northeast were progressively deprived of their extensive landholdings. Without recourse to the support of the laity, the sangha abandoned their monasteries. By the end of the fifteenth century ce, Buddhism was all but gone in India, with the traces of Buddhism pre- served only in the writings, architecture, and ritual practices of Hinduism, Jainism, and Islam.


In recent years, Buddhist archaeologists have often complained that his- torians see archaeology as only useful for finding the cities and monaster- ies mentioned in Buddhist texts and, perhaps, fleshing out a few missing details (Aldenderfer 2008; Coningham 2001; Trautmann and Sinopoli 2002; Fogelin 2007b, 2013; Shaw 2007). To some degree, archaeologists have blamed historians for our own failings. Until the last few decades, archaeologists have accepted the view that religion is of the mind, beyond the reach of our materialist research. The development of the archaeology of religion, to my view, can be roughly divided into three phases. Initially, archaeologists of religion were forced to justify and defend their research to skeptical archaeologists—to establish religion as a legitimate venue for archaeological research. Next, archaeologists of religion needed to develop robust methods to study ancient religion. Among the primary insights in this phase was the recognition that religion is not simply what people think about, but rather something that people do. Archaeologists recognized that by doing religion, ancient people could legitimize authority over others and promote social solidarity among a population. Rather than the mere reflec- tion of societal concerns, archaeologists recognized that religion and ritual were instrumental in creating and perpetuating social practices.

With the archaeology of religion now a legitimate form of archaeo- logical research, and the methods for the archaeology of religion becom- ing systematized, archaeologists can now move into a third phase of research—contributing to the study of long-term religious history. This is precisely what I have attempted in this volume. But in doing so, I have realized that some of the basic claims from the second phase of research—including my own work—are limiting what the archaeology of religion can accomplish. While it is true that religion is something that peo- ple do, religion is also something that people think. Any effective study of long-term religious change must accommodate both the religious practices and the religious principles of ancient people. Symbols, doctrines, cosmolo- gies, and theology matter, and the archaeology of religion is poorer in their absence. In part, this requires the incorporation of historical sources, when available. But it also requires developing new archaeological approaches that are capable of incorporating symbolic and structural aspects of ancient societies. This does not mean that archaeologists should abandon the more practice-based approaches developed in the second phase of research, only that practice theory be balanced with more structural approaches. More so, these structural approaches need not be invented in whole cloth. Within the anthropological literature are numerous robust methods for studying

the symbolic and structural aspects of religion that can be applied to archaeological contexts. Here I have particularly relied on Peircian semiot- ics, but other approaches could also be used.

The standard critique of structural studies in anthropology is that they are unable to account for religious change since they postulate a superor- ganic system of belief that determines the behavior of individuals within any particular society. Rather than knowledgeable actors who alter and contest social practices, in structural studies people are automatons enact- ing cultural norms. To some degree, this critique of structuralism is cor- rect. Many structural studies have removed the agency of individuals. Further, the acceptance of this structural view of religion is precisely why archaeologists avoided the study of religion for so long. The rejection of the structural view of religion and the recognition that religion is something that people do were likely prerequisites for the development of the archae- ology of religion. But with these prerequisites met, it is now time to con- sider the real impact of structure—of religious symbols—in archaeological studies. This can be done by recognizing that the problem with structural approaches is not that they are structural, but rather the assumption that social structures are coherent. Once the idea of a structural coherency is abandoned, the dichotomy of structure and agency diminishes in value.

When structure is viewed as inherently inconsistent, incoherent, and fractious, assumptions of long-term stability no longer hold. If structural forms are incoherent or out of balance, it requires human effort—agency—to maintain them. Periods of stability are just as depen- dent on human agency as periods of change, and inaction can lead to change just as much as action can. The failure of traditional structural approaches to account for human change was not the product of their structural orientation, but rather their assumption that structure was a well-oiled, coherent machine. Structural incoherency can serve as the source for contestation and change, as different people seek to tip the bal- ance one way or another. In this archaeological history, I have identified a contradiction between the individual and group as fundamental to Indian Buddhism. But this does not mean that I think that this specific contra- diction is inherent to all societies. Rather, different societies will have dif- ferent tensions, contradictions, and social ruptures. Archaeologists and historians working in other contexts will need to identify the structural disjunctures particular to the societies they study. While the study of dis- juncture is valuable, it cannot be applied uncritically.

Viewing Indian Buddhism as shot through with contradictions, ten- sions, and disjunctures has allowed me to use numerous, often dis- cordant, social theories to more effectively study it. Critically, I did not

attempt to synthesize theories of practice, structure, and semiotics—I juggled them. I make no claim to resolving the dialectic. I frankly doubt it is resolvable. Theorists have been trying to balance the competing con- cerns of structure and agency, however conceived, for at least two or three millennia. Perhaps someone will finally solve this problem in my lifetime, but I wouldn’t bet on it. In the meantime, I am comfortable using differ- ent theories to address those portions of the archaeological and historical record of Buddhism for which they are most appropriate. If this means that I employ sometimes incompatible theories, so be it. In a world where societies are incoherent, theoretical incoherence is no vice.


I very much doubt that this first attempt at a large-scale integration of Buddhist archaeology and history will be the final word on Indian Buddhism. In fact, shortly after I completed the manuscript of this vol- ume, Robin Coningham and Kosh Prasad Acharya reported on excava- tions at Lumbini that, for the first time, provided solid archaeological evidence for pre-Mauryan Buddhism (Coningham et al. 2013). I was forced to return to the manuscript and delete the many times I had writ- ten that no pre-Mauryan Buddhist remains had ever been conclusively discovered. But new understandings of the history of Indian Buddhism are not solely dependent upon new archaeological discoveries. In my own research, the work of the textual scholar Gregory Schopen (1997, 2004, 2005) and the anthropologist Stanley Tambiah (1976, 1982, 1984) have greatly informed my archaeological interpretations. It is my hope that the archaeological history I have presented will be of similar value to scholars and students in Buddhist studies.

I do not expect that Buddhist textual scholars will agree with every- thing I have presented, but perhaps my archaeological history can inform alternative readings of ancient Buddhist texts—readings that lead to new Buddhist histories, different from both the standard history of Buddhist textual studies and the revisionist archaeological history I have presented here. For too long, the study of ancient Indian Buddhism has been sepa- rated into two walled gardens. In one, textual scholars have sat on benches reading ancient texts and debating their significance. In another, archae- ologists have been digging up the roses and generally making a mess of the place. Just as Siddhartha Gautama had to leave his father’s palace in Kapilavastu to gain enlightenment, archaeologists and textual scholars must abandon their walled gardens and begin to work together.



Agrawala, V. S. 1965. Indian Art. Prithivi Prakashan, Varanasi.

Aldenderfer, M. 2008. On Text, Materiality, and the Tibetan Buddhist Religious Architecture at Piyang. In Religion, Archaeology, and the Material World, edited by L. Fogelin, pp. 339–358. Center for Archaeological Investigations, Carbondale, IL.

Allchin, R. (editor) 1995. The Archaeology of Early Historic South Asia: The Emergence of Cities and States. Cambridge University Press, Cambridge.

Appadurai, A. (editor) 1986. The Social Life of Things. Cambridge University Press, Cambridge.

Asad, T. 2003. Formations of the Secular: Christianity, Islam, Modernity. Cultural Memory in the Present. Stanford University Press, Stanford, CA.

Bailey, G., and I. Mabbet. 2003. The Sociology of Early Buddhism. Cambridge University Press, Cambridge.

Bareau, A. 1974. La Parinirvana du Bouddha et la Naissance de la Religion Bouddique.

Bulletin de L’Ecole Francaise d’Extreme-Orient 64:275–299.

Barnes, G. L. 1999. Buddhist Landscapes of East Asia. In Archaeologies of Landscape, edited by W. Ashmore and A. B. Knapp, pp. 101–123. Blackwell, Malden, MA.

Barrett, D. E. 1954. Sculptures from Amaravati in the British Museum. Trustees of the British Museum, London,.

Barth, F. 1961. Nomads of South Persia. Little, Brown, New York.

Barua, B. 1979. Bharhut: Parts 1, 2, and 3. Indological Book House, Varanasi. Basham, A. L. 1967. The Wonder That Was India. Rupa, Calcutta.

———. 1981. The Evolution of the Concept of the Bodhisattva. In The Boddhisattva Doctrine in Buddhism, edited by L. S. Kawamura. Wilfred Laurier University Press, Waterloo, ON.

Bauer, A. A. 2002. Is What You See All You Get? Recognizing Meaning in Archaeology.

Journal of Social Archaeology 2:37–52.

Beal, S. 1908. Si-yu-ki: Buddhist Records of the Western World. 2 vols. Paragon Book Reprint, New York.

Bechert, H. (editor) 1995. When Did the Buddha Live? The Controversy on the Dating of the Historical Buddha. Sri Satguru Publications, Delhi.

Begley, V. 1983. Arikamedu Reconsidered. American Journal of Archaeology 87: 461–481.

Behrendt, K. A. 2007. The Art of Gandhara in the Metropolitan Museum of Art. Yale University Press, New Haven, CT.

Bell, C. 1992. Ritual Theory, Ritual Practice. Oxford University Press, New York.

———. 1997. Ritual: Perspectives and Dimensions. Oxford University Press, Oxford.

( 235 )

Bloch, M. 1977. The Past in the Present. Man 12:278–292.

———. 1986. From Blessing to Violence: History and Ideology in the Circumcision Ritual of the Merina of Madagascar. Cambridge University Press, Cambridge.

———. 1989. Ritual, History, and Power: Selected Papers in Anthropology. Athlon Press, London.

Boivin, N. 2008. Material Cultures, Material Minds: The Role of Things in Human Thought, Society and Evolution. Cambridge University Press, Cambridge.

Bourdieu, P. 1977. An Outline of a Theory of Practice. Cambridge University Press, Cambridge.

Bradley, R. 1991. Ritual, Time, and History. World Archaeology 23:209–219.

———. 1998. The Significance of Monuments: On the Shaping of Human Experience in Neolithic and Bronze Age Europe. Routledge, London.

———. 2000. An Archaeology of Natural Places. Routledge, London.

———. 2005. Ritual and Domestic Life in Prehistoric Europe. Routledge, London. Breckenridge, C. A., and P. van der Veer (editors). 1993. Orientalism and the

Postcolonial Predicament: Perspectives on South Asia. University of Pennsylvania Press, Philadelphia.

Briant, P. 2002. From Cyrus to Alexander: A History of the Persian Empire. Translated by

P. T. Daniels. Eisenbrauns, Winona Lake, IN.

Brown, P. 1965. Indian Architecture: The Buddhist and Hindu Periods. D.B. Taraporevela and Sons, Bombay.

Brück, J. 1999. Ritual and Rationality: Some Problems of Interpretation in European Archaeology. European Journal of Archaeology 2(3):313–344.

Burgess, J. (1882) 1972. Notes on the Amaravati Stupa. Prithivi Prakashan, Varanasi. Caldwell, R. 1856. A Comparative Grammar of the Dravidian or South-Indian Family of

Languages. Harrison, London.

Campbell, A. D. 1816. A Grammar of the Teloogoo Language, Commonly Termed the Gentoo, Peculiar to the Hindoos Inhabiting the North Eastern Provinces of the Indian Peninsula. College Press of Fort St. George, Madras.

Carrithers, M. 1979. The Modern Ascetics of Lanka and the Pattern of Change in Buddhism. Man 14(2):294–310.

———. 1983. The Forest Monks of Sri Lanka: An Anthropological and Historical Study.

Oxford University Press, Delhi.

Casson, L. 1989. The Periplus Maris Erythraei. Princeton University Press, Princeton, NJ. Chakrabarti, D. K. 1995. Post-Mauryan States of Mainland South Asia (c. bc 185–ad 320). In The Archaeology of Early Historic South Asia: The Emergence of Cities

and States, edited by R. Allchin, pp. 99–122. Cambridge University Press, Cambridge.

——— 1997. Colonial Indology: Sociopolitics of the Ancient Indian Past. Munshiram Manoharlal, Delhi.

———. 2001. A History of Indian Archaeology: From the Beginning to 1947. Munshiram Manoharlal, Delhi.

———. 2001. The Archaeology of Hinduism. In Archaeology and World Religion, edited by T. Insoll, pp. 33–60. Routledge, New York.

Chauley, G. C. 2000. Excavated Remains at Lalitagiri, A Buddhist Site in Orissa. In Archaeology of Orissa, edited by K. K. Basa and P. Mohanty, pp. 441–455. Pratibha Prakashan, Delhi.

Chesson, M. S. (editor). 2001. Social Memory, Identity, and Death: Anthropological Perspectives on Mortuary Rituals. American Anthropological Association, Arlington, VA.

Chimpa, L., and A. Chattopadhaya. 2010. Taranatha’s History of Buddhism in India.

Motilal Banarsidas, Delhi.

Comaroff, J. 1985. Body of Power, Spirit of Resistance: The Culture and History of a South African People. University of Chicago Press, Chicago.

Coningham, R. 1995. Monks, Caves, and Kings: A Reassessment of the Nature of Early Buddhism in Sri Lanka. World Archaeology 27:266–281.

———. 2001. The Archaeology of Buddhism. In Archaeology and World Religion, edited by T. Insoll, pp. 61–95. Routledge, London.

Coningham, R. A. E., K. P. Acharya, K. M. Strickland, C. E. Davis, M. J. Manuel,

I. A. Simpson, K. Gilliland, J. Tremblay, T. C. Kinnaird, and D. C. W. Sanderson. 2013. The Earliest Buddhist Shrine: Excavating the Birthplace of the Buddha, Lumbini (Nepal). Antiquity 87(338):1104–1123.

Connerton, P. 1989. How Societies Remember. Cambridge University Press, Cambridge. Coomaraswamy, A. K. 1927. History of Indian and Indonesian Art. E. Weyhe, New York. Cunningham, A. (1854) 1997. The Bhilsa Topes or Buddhist Monuments of Central India.

Munshiram Manoharlal, New Delhi.

———. (1876) 1962. The Stupa of Bharhut: A Buddhist Monument Ornamented with Numerous Sculptures. Indological Book House, Varanasi.

———. (1892) 1998. Mahabodhi or the Great Buddhist Temple under the Bodhi Tree at Buddha-Gaya. Munshiram Manoharlal, New Delhi.

Daniel, E. V. 1984. Fluid Signs: Being a Person the Tamil Way. University of California Press, Berkeley.

Davids, T. W. R. 1910. Buddhism: Being a Sketch of the Life and Teachings of Gautama, the Buddha. Society for Promoting Christian Knowledge, London.

Davids, T. W. R., and C. A. F. R. Davids (1910) 2007. Dialogues of the Buddha. Motilal Banarsidass, New Delhi.

Dehejia, V. 1972. Early Buddhist Rock Temples. Cornell University Press, Ithaca, NY.

———. 1997. Discourse in Early Buddhist Art: Visual Narratives of India. Munshiram Manoharlal, New Delhi.

DeMarrais, E. 2004. Materialization of Culture. In Rethinking Materiality: Engagement of Mind with the Material World, edited by E. DeMarrais, C. Gosden, and

C. Renfrew, pp. 11–22. McDonald Institute, Cambridge.

DeMarrais, E., L. J. Castillo, and T. Earle. 1996. Ideology, Materialization, and Power Strategies. Current Anthropology 37(1):15–86.

Dirks, N. B. 1993. The Hollow Crown: Ethnohistory of an Indian Kingdom. University of Michigan Press, Ann Arbor.

———. 2001. Castes of Mind: Colonialism and the Making of Modern India. Princeton University Press, Princeton, NJ.

Dirks, N. B., G. Eley, and S. B. Ortner (editors). 1993. Culture/Power/History: A Reader in Contemporary Social Theory. Princeton University Press, Princeton, NJ.

Dobres, M. A. 2000. Technology and Social Agency. Blackwell, Malden.

Douglas, M. 1966. Purity and Danger: An Analysis of the Concepts of Pollution and Taboo.

Routledge, New York.

Dundas, P. 1992. The Jains. Routledge, New York.

Durkheim, E. (1893) 1984. The Division of Labor in Society. Translated by W. D. Halls.

Free Press, New York.

———. (1897) 1951. Suicide: A Study in Sociology. Translated by J. A. Spaulding and

G. Simpson. Free Press, Glencoe, IL.

———. (1915) 1995. The Elementary Forms of Religious Life. Translated by K. E. Fields.

Free Press, New York.

Dutt, S. 1988. Buddhist Monks and Monasteries of India. Motilal Banarsidass, Delhi. Eaton, R. M. 2000. Temple Desecration and Indo-Muslim States. Journal of Islamic

Studies 11(3):283–319.

Eco, U. 1976. A Theory of Semiotics. Advances in Semiotics. Indiana University Press, Bloomington.

Eliade, M. 1961. Images and Symbols: Studies in Religious Symbolism. Princeton University Press, Princeton, NJ.

Ellis, F. W. 1816. Note to the Introduction. In A Grammar of the Teloogoo Language, edited by A. D. Campbell. College Press of Fort St. George, Madras.

Elster, J. (editor) 1986. Karl Marx: A Reader. Cambridge University Press, Cambridge. Erdosy, G. 1995. City States of North India and Pakistan at the Time of the Buddha. In The Archaeology of Early Historic South Asia: The Emergence of Cities and States,

edited by R. Allchin, pp. 99–122. Cambridge University Press, Cambridge.

Fergusson, J., and J. Burgess. (1880) 1988. The Cave Temples of India. Munshiram Manoharlal Publishers, New Delhi.

Fogelin, L. 2003. Ritual and Presentation in Early Buddhist Religious Architecture.

Asian Perspectives 42(1):129–154.

———. 2006. Archaeology of Early Buddhism. AltaMira Press, Lanham, MD.

———. 2007a. The Archaeology of Religious Ritual. Annual Reviews in Anthropology


———. 2007b. History, Ethnography, and Essentialism: The Archaeology of Religion and Ritual in South Asia. In The Archaeology of Ritual, edited by E. Kyriakidis,

pp. 23–42. Cotsen Institute of Archaeology, Los Angeles.

———. 2007c. Inference to the Best Explanation: A Common and Effective Form of Archaeological Reasoning. American Antiquity 72(4):603–625.

———. 2008a. Religion, Archaeology, and the Material World. Center for Archaeological Investigations, Carbondale, IL.

———. 2008b. Introduction: Methods for the Archaeology of Religion. In Religion, Archaeology, and the Material World, edited by L. Fogelin, pp. 1–14. Center for Archaeological Investigations, Carbondale, IL.

———. 2008c. Delegitimizing Religion: The Archaeology of Religion as . . . Archaeology. In Belief in the Past, edited by K. Hayes-Gilpin and D. S. Whitely, pp. 129–142. Left Coast Press, Walnut Creek, CA.

———. 2011. Ignoring the Problem: Spatial Strategies for Ameliorating Social Contradictions in Early South Asian Buddhism. Journal of Social Archaeology 11(2):194–211.

———. 2012. Material Practice and the Metamorphosis of a Sign: Early Buddhist Stupas and the Origin of Mahayana Buddhism. Asian Perspectives 51(2):278–310.

———. 2013. The Place of Veneration in Buddhist Mortuary Sites. In Oxford Handbook of the Archaeology of Death and Burial, edited by S. Tarlow and

L. Nilsson-Stutz, pp. 227–240. Oxford University Press, Oxford.

Foucault, M. 1972. The Archaeology of Knowledge. Translated by A. Sheridan. Pantheon Books, New York.

———. 1977. Discipline and Punish. Translated by A. Sheridan. Vintage Books, New York.

———. 1990a. The History of Sexuality, Volume 1: Introduction. Translated by

R. Hurley. Vintage Books, New York.

———. 1990b. The History of Sexuality, Volume 2: The Use of Pleasure. Translated by

R. Hurley. Vintage Books, New York.

———. 1990c. The History of Sexuality, Volume 3: The Care of Self. Translated by

R. Hurley. Vintage Books, New York.

Fowles, S. M. 2008. Steps Toward an Archaeology of Taboo. In Religion, Archaeology, and the Material World, edited by L. Fogelin. Center for Archaeological Investigations, Carbondale, IL.

———. 2013. An Archaeology of Doings: Secularism and the Study of Pueblo Religion.

School of Advanced Research, Santa Fe.

Frauwallner, E. 1956. The Earliest Vinaya and the Beginnings of Buddhist Literature.

Serie Orientale Roma 8. Is. M. E. O., Rome.

Fritz, J. M. 1986. Vijayanagara: Authority and Meaning of a South Indian Imperial Capital. American Anthropologist 88:44–55.

Fritz, J. M., and G. Michell. 1989. Interpreting the Plan of a Medieval Hindu Capital: Vijayanagara. World Archaeology 19:105–129.

Fuller, C. J. 1992. The Camphor Flame: Popular Hinduism and Society in India. Princeton University Press, Princeton, NJ.

Fussman, G. 1988. Central and Provincial Administration in Ancient India: The Problem of the Mauryan Empire. The Indian Historical Review 14:43–72.

Geertz, C. 1957. Ritual and Social Change: A Javanese Example. American Anthropologist 59(1):32–54.

———. 1973. The Interpretation of Cultures. Basic Books, New York.

Gell, A. 1998. Art and Agency: An Anthropological Theory. Oxford University Press, Oxford.

Ghosh, A. 1959. Nalanda. Archaeological Survey of India, New Delhi.

Giddens, A. 1979. Central Problems in Social Theory: Action, Structure, and Contradiction in Social Analysis. University of California Press, Berkeley.

———. 1984. The Constitution of Society: Outline of a Theory of Structuration. Polity, London.

Giles, H. A. 1923. Gems of Chinese Literature. B. Quaritch, London.

Gramsci, A. 2011. Prison Notebooks: Volumes 1–3. Translated by J. A. Buttigieg.

Columbia University Press, New York.

Guha, R., and G. C. Spivak (editors). 1988. Selected Subaltern Studies. Oxford University Press, New York.

Gupta, S. P. 1980. The Roots of Indian Art. B. R. Publishing, Delhi.

Habib, I. and F. Habib. 1990. Mapping the Mauryan Empire. Proceedings of the Indian History Congress 1989–90, 57–79.

Hawkes, J. 2008. Bharhut: A Re-Assessment. South Asian Studies 24:1–14.

———. 2009. The Wider Archaeological Contexts of the Buddhist Stupa Site of Bharhut. In Buddhist Stupas in South Asia, edited by A. Shimada and J. Hawkes. Oxford University Press, Oxford.

Hayes, K. A. 1993. When Is a Symbol Archaeologically Meaningful? Meaning, Function, and Prehistoric Visual Arts. In Archaeological Theory: Who Sets the Agenda, edited by N. Yoffee and A. Sherratt, pp. 81–92. Cambridge University Press, Cambridge.

Heitzman, J. 1997. Gifts of Power: Lordship in an Early Indian State. Oxford University Press, Delhi.

Hobsbawm, E., and T. Ranger (editors). 1983. The Invention of Tradition. Cambridge University Press, Cambridge.

Horkeimer, M., and T. W. Adorno. 2007. Dialectic of Enlightenment. Translated by

E. Jephcott. Stanford University Press, Palo Alto, CA.

Howey, M. C. L., and J. M. O’Shea. 2006. Bear’s Journey and the Study of Ritual in Archaeology. American Antiquity 71(2):261–282.

Hume, I. N. 1964. Handmaiden to History. North Carolina Historical Review 41(2): 215–225.

Humphrey, C., and J. Laidlaw. 1994. The Archetypal Actions of Ritual: A Theory of Ritual Illustrated by the Jain Rite of Worship. Oxford University Press, Oxford.

Inden, R. B. 1990. Imagining India. Basil Blackwell, Cambridge, MA.

Insoll, T. (editor) 2001. Archaeology and World Religion. Routledge, London. Insoll, T. 2004. Archaeology, Religion, Ritual. Routledge, London.

Johansen, P. G., and A. M. Bauer (editors). 2011. The Archaeology of Politics: The Materiality of Political Practice and Action in the Past. Cambridge Scholars, Newcastle upon Tyne.

Jones, W. 1824. Discourses Delivered before the Asiatic Society: And Miscellaneous Papers on the Religion, Poetry, Literature, etc. of the Nations of India. C. S. Arnold, London.

Keane, W. 1997. Signs of Recognition: Powers and Hazards of Representation in an Indonesian Society. University of California Press, Berkeley.

Kemp, B. 1995. How Religious Were the Ancient Egyptians? Cambridge Archaeological Journal 5:25–54.

Kenoyer, J. M. 1997. Early City-States in South Asia: Comparing the Harappan Phase and Early Historic Period. In The Archaeology of City-States: Cross-Cultural Approaches, edited by D. L. Nichols and T. H. Charlton. Smithsonian Institution Press, Washington, DC.

———. 1998. Ancient Cities of the Indus Valley Civilization. Oxford University Press, Islamabad.

Kosambi, D. D. 1965. The Culture and Civilization of Ancient India in Historical Outline.

Routledge and Kegan Paul, London.

Kyriakidis, E. (editor). 2007. The Archaeology of Ritual. Cotsen Institute of Archaeology, Los Angeles.

Lahiri, N. 1992. The Archaeology of Indian Trade Routes (up to c. 200 b.c.). Oxford University Press, Delhi.

———. 2000. Archaeology and Identity in Colonial India. Antiquity 74:687–692.

———. 2005. Finding Forgotten Cities: How the Indus Civilization was Discovered.

Permanent Black, Delhi.

Lal, B. B. 1955. Excavations at Hastinapura and Other Explorations in the Upper Ganga and Sutlej Basins. Ancient India 16:5–151.

Lamotte, E. 1960. Manjusri. T’oung Pao 48:3–4.

———. 1988. History of Indian Buddhism from the Origins to the Saka Era. Translated by S. Webb-Boin. Institute Orientaliste, Universite de Louvain, Louvain-la-Nueve.

Latour, B. 1993. We Have Never Been Modern. Translated by C. Porter. Harvard

University Press, Cambridge, MA.

———. 1999. Pandora’s Hope: Essays on the Reality of Science Studies. Harvard University Press, Cambridge, MA.

———. 2005. Reassembling the Social: An Introduction to Actor-Network Theory. Oxford University Press, Oxford.

Leidy, D. P. 2008. The Art of Buddhism: An Introduction to Its History and Meaning.

Shambhala, Boston.

Lévi-Strauss, C. 1963. Structural Anthropology. Translated by C. Jacobson and

B. G. Schoepf. Basic Books, New York.

———. 1983. Structural Anthropology (vol. 2). University of Chicago Press ed.

Translated by M. Layton. University of Chicago Press, Chicago, IL.

Liu, X. 1988. Ancient India and Ancient China: Trade and Religious Exchanges, AD 1–600.

Oxford University Press, Delhi.

Lockwood, M., and A. V. Bhat. 1981. The Farce of Drunken Sport. Christian Literature Society, Chennai.

Lopez, D. S. J. 1988. The Heart Sutra Explained: Indian and Tibetan Commentaries. State University of New York Press, Albany.

———. 1995. Buddhism in Practice. Princeton University Press, Princeton, NJ.

———. 1996. Elaborations on Emptiness: Uses of the Heart Sutra. Princeton University Press, Princeton, NJ.

———. 2001. The Story of Buddhism: A Concise Guide to Its History and Teachings.

Harper San Francisco, San Francisco.

Luders, H. 1912. A List of Brahmi Inscriptions. Epigraphia Indica 10.

Majumdar, N. G. 1937. The Bajaur Casket of the Reign of Menander. Epigraphia Indica


Malinowski, B. (1925) 1948. Magic, Science and Religion. Doubleday, New York. Marshall, J. 1931. Mohenjo-Daro and the Indus Civilization. Arthur Probsthain, London.

———. (1951) 1975. Taxila: An Illustrated Account of Archaeological Excavations Carried Out at Taxila under the Orders of the Government of India Between the Years 1913 and 1934. 3 vols. Motilal Banarsidass, Delhi.

Marshall, J., and A. Foucher. 1983. The Monuments of Sanchi. 3 vols. Swati Publications, Delhi.

Marx, K. (1844) 1970. A Critique of Hegel’s Philosophy of Law. Translated by A. Jolin and J. O’Malley. Cambridge University Press, Cambridge.

———. (1845–1846) 1998. The German Ideology. Prometheus Books, Amherst.

———. (1869) 1963. The 18th Brumaire of Louis Bonaparte. International Publishers, New York.

McCrindle, J. W. (1901) 1979. Ancient India as Described in Classical Literature.

Munshiram Manoharlal, Delhi.

———. (1926) 2000. Ancient India as Described by Megasthenes and Arrian. Munshiram Manoharlal, Delhi.

———. (1927) 2000. Ancient India as Described by Ptolemy. Munshiram Manoharlal, Delhi.

Mertz, E., and R. J. Parmentier (editors). 1985. Semiotic Mediation: Sociocultural and Psychological Perspectives. Academic Press, New York.

Meskell, L. 2005. Archaeologies of Materiality. Blackwell, Malden, MA; Oxford. Miller, D. 1985. Artefacts as Categories: A Study of Ceramic Variability in Central

India. New Studies in Archaeology. Cambridge University Press, Cambridge; New York.

———. 1987. Material Culture and Mass Consumption. Social Archaeology. Blackwell, Oxford; New York.

———. 1998. Material Cultures: Why Some Things Matter. University of Chicago Press, Chicago.

———. 2005. Materiality. Duke University Press, Durham, NC.

Mills, B. J., and W. H. Walker (editors). 2008. Memory Work: Archaeologies of Material Practice. School for Advanced Research, Santa Fe.

Mitra, D. 1971. Buddhist Monuments. Sahitya Samsad, Calcutta.

———. 1981. Ratnagiri (1958–61): Vol. 1. Memoirs of the Archaeological Survey of India no. 80. Archaeological Survey of India, New Delhi.

———. 1983. Ratnagiri (1958–61): Vol. 2. Memoirs of the Archaeological Survey of India no. 80. Archaeological Survey of India, New Delhi.

Moore, J. D. 1996. Architecture and Power in the Ancient Andes. Cambridge University Press, Cambridge.

Morrison, K. D. 1995. Trade, Urbanism, and Agricultural Expansion: Buddhist Monastic Institutions and the State in Early Historic Western Deccan. World Archaeology 27(2):203–221.

———. 2005. Brahmagiri Revisited: A Re-analysis of the South Indian Sequence. In South Asian Archaeology 2001, edited by C. Jarrige and V. Lefèvre, vol. 1,

pp. 271–275. Éditions Recherche sur les Civilisations-ADPF, Paris.

Munier, C. 1998. Sacred Rocks and Buddhist Caves in Thailand. White Lotus Press, Bangkok.

Nagaraju, S. 1981. Buddhist Architecture of Western India. Agam Kala Prakashan, Delhi.

Narain, A. K., and T. N. Roy. 1976. Excavations at Rajghat, 1957–58, 1960–65. Banaras Hindu University, Varanasi.

Narain, A. K., and P. Singh. 1977. Excavations at Rajghat. Banares Hindu University, Varanasi.

Nigam, J. S. 2000. Notes on the Excavation of Udayagiri and Khalkattapatna. In Archaeology of Orissa, edited by K. K. Basa and P. Mohanty, pp. 495–506. Pratibha Prakashan, Delhi.

Oldenberg, H. 1882. Buddha: His life, His Doctrine, His Order. Translated by W. Hoey.

Williams and Norgate, London.

Ortman, S. G. 2000. Conceptual Metaphor in the Archaeological Record: Methods and an Example from the American Southwest. American Antiquity 65(4): 613–645.

Ortner, S. B. 1973. On Key Symbols. American Anthropologist 75:1338–1346.

———. 1984. Theory in Anthropology since the Sixties. Comparative Studies in Society and History 26(1):126–166.

———. 1989. High Religion: A Cultural and Political History of Sherpa Buddhism.

Princeton University Press, Princeton, NJ.

Pant, P. C., and V. Jayaswal. 1990. Ancient Stone Quarries of Chunnar: An Appraisal.

Pragdhara: Journal of the UP State Archaeological Organization 1(1):49–52. Parmentier, R. J. 1997. The Pragmatic Semiotics of Culture. Semiotica 116:1–115. Pauketat, T. R. 2001. Practice and History in Archaeology. Anthropological Theory


Peirce, C. S., C. Hartschorne, P. Weiss, and A. W. Burks. 1931–1958. Collected Papers of Charles Sanders Peirce. Harvard University Press, Cambridge, MA.

Peirce, C. S., N. Houser, J. R. Eller, A. C. Lewis, C. L. Clark, and D. B. Davis. 1998. The Essential Peirce: Selected Philosophical Writings: 1893–1913. Indiana University Press, Bloomington.

Peirce, C. S., N. Houser, and C. J. W. Kloesel. 1992. The Essential Peirce: Selected Philosophical Writings: 1867–1893. Indiana University Press, Bloomington.

Piggot, S. 1943. The Earliest Buddhist Shrines. Antiquity 17:1–10.

Plunket, P. (editor). 2002. Domestic Ritual in Ancient Mesoamerica. Cotsen Institute of Archaeology, Los Angeles.

Poonacha, K. P. 2007. Excavations at Mahastupa, Kanaganahalli, Chitapur Taluk, Gulbarga District, Karnataka. Archaeological Survey of India, New Delhi.

Possehl, G. L. 2002. The Indus Civilization. Altamira Press, Walnut Creek, CA. Prasad, N. R. V. 1994. Bavikonda: A Buddhist Site in North Coastal Andhra Pradesh.

Department of Archaeology and Museums, Government of Andhra Pradesh, Hyderabad.

Prasad, R. C. 1987. Archaeology of Champa and Vikramasila. Ramanand Vidya Bhawan, Delhi.

Preucel, R. W. 2006. Archaeological Semiotics. Blackwell, Malden, MA.

Pruecel, R. W., and A. A. Bauer. 2001. Archaeological Pragmatics. Norwegian Archaeological Review 34:85–96.

Ramachandra Rao, P. R. 2002. Amaravati. Youth Advancement, Tourism & Cultural Dept. Govt. of Andhra Pradesh, Hyderabad.

Ramachandran, T. N. 1953. Nagarjunakonda, 1938. Memoirs of the Archaeological Survey of India 71.

Ray, H. P. 1986. Monastery and Guild: Commerce under the Satavahanas. Oxford University Press, Delhi.

———. 1994. Kanheri: The Archaeology of an Early Buddhist Pilgrimage Center in Western India. World Archaeology 26:35–46.

———. 1999. Archaeology of Seafaring. Pragati Publications, Delhi.

Rea, A. 1907–1908. A Buddhist Monastery on the Sankaram Hills. M.A.S.T., Calcutta. Renfrew, C. 1985. The Archaeology of Cult: The Sanctuary at Phylakopi. Thames and

Hudson, London.

Renfrew, C., and E. Zubrow (editors). 1994. The Ancient Mind: Elements of Cognitive Archaeology. Cambridge University Press, Cambridge.

Rongxi, L. 1997. The Great Tang Dynasty Record of Western Regions. Numata Center for Buddhist Translation and Research, Berkeley.

———. 2000. Buddhist Monastic Traditions of Southern Asia: A Record of the Inner Law Sent Home from the South Seas by Sramana Yijing. Numata Center for Buddhist Translation and Research, Berkeley.

Rongxi, L., and A. Dalia. 2002. Lives of Great Monks. Numata Center for Buddhist Translation and Research, Berkeley.

Rowlands, M. 1993. The Role of Memory in the Transmission of Culture. World Archaeology 25:141–151.

Sahlins, M. 2002. Waiting for Foucault, Still. Prickly Paradigm Press, Chicago.

Sahni, D. R. 1937. Archaeological Remains and Excavations at Bairat. Jaipur State, Jaipur.

Said, E. W. 1978. Orientalism. Vintage Books, New York.

Sarkar, H. 1966. Studies in Early Buddhist Architecture of India. Munshiram Manoharlal, New Delhi.

Sarkar, H., and B. N. Mishra. 1966. Nagarjunakonda. Archaeological Survey of India, New Delhi.

Sastri, N. 1966. A History of South India from Prehistoric Times to the Fall of Vijayanagara.

3rd ed. Oxford University Press, Madras.

Sastry, V. V. K., B. Subrahmanyam, and N. R. K. Rao. 1992. Thotlakonda: A Buddhist Site in Andhra Pradesh. Department of Archaeology and Museums, The Government of Andhra Pradesh, Hyderabad.

Saussure, F. d., C. Bally, A. Sechehaye, and A. Riedlinger. 1983. Course in General Linguistics. Translated by R. Harris. Duckworth, London.

Schaffer, J. G. 1984. The Indo-Aryan Invasions: Cultural Myth and Archaeological Reality. In The People of South Asia: The Biological Anthropology of India, Pakistan, and Nepal, edited by J. R. Lukas, pp. 77–88. Plenum Press, New York.

Schiffer, M. B. 1976. Behavioral Archeology. Studies in Archeology. Academic Press, New York.

———. 1995. Behavioral Archaeology: First Principles. University of Utah Press, Salt lake City.

Schmidt, P. R., and B. B. Mapunda. 1997. Ideology and the Archaeological Record in Africa: Interpreting Symbolism in Iron Smelting Technology. Journal of Anthropological Archaeology 16(1):73–102.

Schoff, W. H. 1974. The Periplus of the Erythrean Sea. Munshiram Manoharlal, Delhi. Schopen, G. 1997. Bones, Stones, and Buddhist Monks: Collected Papers on the

Archaeology, Epigraphy, and Texts of Monastic Buddhism in India. Studies in the Buddhist Traditions. University of Hawai’i Press, Honolulu.

———. 2000. The Mahayana and the Middle Period in Indian Buddhism: Through a Chinese Looking Glass. Eastern Buddhist 32(2):1–25.

———. 2004. Buddhist Monks and Business Matter: Still More Papers on Monastic Buddhism in India. University of Hawai’i Press, Honolulu.

———. 2005. Figments and Fragments of Mahayana Buddhism in India. University of Hawaii Press, Honolulu.

Sewell, R. (1880) 1973. Report of the Amaravati Tope and Excavations on Its Site in 1877.

Bhartiya Pub. House, [Varanasi].

Sharma, G. R. 1960. The Excavations at Kausambi (1957–59). Department of Ancient History, Culture and Archaeology, University of Allahabad, Allahabad.

———. 1969. Excavations at Kausambi (1949–50). Memoirs of the Archaeological Survey of India. Manager of Publications, Delhi.

Sharma, R. S. 1980. Indian Feudalism. Macmillan, Madras.

Shastri, A. M. 1997. Vakatakas: Sources and History. Aryan Books International, New Delhi.

Shaw, J. 2000. Sanchi and Its Archaeological Landscape: Buddhist Monasteries, Settlements and Irrigation Works in Central India. Antiquity 74:775–756.

———. 2002. Methodological and Theoretical Frameworks for the Archaeological Survey of Central India. Archaeological Review from Cambridge. June 2002.

———. 2007. Buddhist Landscapes in Central India: Sanchi Hill and Archaeologies of Religious and Social Change, c. Third century bc to Fifth Century ad. British Association of South Asian Studies, London.

———. 2013a. Archaeology of Religious Change: Introduction. World Archaeology


———. 2013b. Archaeologies of Buddhist Propagation in Ancient India: “Ritual” and “Practical” Models of Religious Change. World Archaeology 45(1):83–108.

Shaw, J., and J. Sutcliffe. 2001. Ancient Irrigation Works in the Sanchi Area: An Archaeological and Hydrological Investigation. South Asian Studies 17:55–75.

Shimada, A. 2012. Early Buddhist Architecture in Context: The Great Stupa at Amaravati (ca. 300 bce–300 ce). Brill, Leiden.

Shimada, A., and J. Hawkes (editors). 2009. Buddhist Stupas in South Asia. Oxford University Press, Oxford.

Silverstein, M. 1976. Shifters, Linguistic Categories, and Cultural Descriptions. In Meaning and Anthropology, edited by K. Basso and H. Selby. University of New Mexico Press, Albuquerque.

Singer, M. 1978. For a Semiotic Anthropology. In Sight, Sound and Sense, edited by

T. A. Sebeok, pp. 202–231. Indiana University Press, Bloomington.

Singh, U. 2004. Discovery of Ancient India: Early Archaeologists and the Beginnings of Archaeology. Permanent Black, Delhi.

———. 2008. A History of Ancient and Early Medieval India. Pearson Longman, Delhi. Sinha, B. P., and S. R. Roy. 1969. Vaisali Excavations (1958–62). Directorate of

Archaeology and Museums, Patna.

Sinopoli, C. M. 2001. On the Edge of Empire: Form and Substance in the Satavahana Dynasty. In Empires: Perspectives from Archaeology and History, edited by S.

E. Alcock, T. N. Daltroy, K. D. Morrison, and C. M. Sinopoli, pp. 155–178. Cambridge University Press, Cambridge.

Smith, M. 2001. The Archaeology of an Early Historic Town in Central India. British

Archaeological Reports, Oxford.

Smith, M. L. 1997. Strong Economies, Weak Polities: The Archaeology of Central India in the Early Centuries a.d. Dissertation, University of Michigan.

Spink, W. M. 2006. Ajanta: History and Development. Brill, Leiden.

Srinivas, M. N. 1966. Social Change In Modern India. University of California Press, Berkeley.

———. 1989. The Cohesive Role of Sanskritization. In The Cohesive Role of Sanskritization,

pp. 56–72. Oxford University Press, Delhi.

Stewart, M. L. 1989. Nalanda Mahavira: A Study of an Indian Pala Period Buddhist Site and British Historical Archaeology, 1861–1938. B.A.R., Oxford.

Strenski, I. 1983. On Generalized Exchange and the Domestication of the Sangha.

Man 18(3):463–477.

Strong, J. S. 1983. The Legend of King Asoka. Princeton University Press, Princeton, NJ.

———. 1994. Images of Asoka: Some Indian and Sri Lankan Legends and Their Development. In King Asoka and Buddhism: Historical and Literary Studies, edited by A. Seneviratne, pp. 99–126. Buddhist Publication Society, Kandy.

———. 2004. Relics of the Buddha. Princeton University Press, Princeton, NJ. Subrahmanyam, R. 1964. Salihundam—A Buddhist Site in Andhra Pradesh. Andhra

Pradesh Government Archaeological Series, no. 17. Government of Andhra Pradesh, Hyderabad.

Subrahmanyam, R., K. D. Banerjee, M. D. Khare, B. V. Rao, H. Sarkar, R. Singh,

R. V. Joshi, S. B. Lal, V. V. Rao, K. R. Srinivasan, and K. Totadri. 1975. Nagarju- nakonda (1954–60): Volume 1. Archaeological Survey of India, New Delhi.

Sugandhi, N. 2003. Context, Content, and Composition: Questions of Intended Meaning and the Asokan Edicts. Asian Perspectives 42(2):224–246.

———. 2008. Between the Patterns of History: Rethinking Mauryan Imperial Interaction in the Southern Deccan. University of Chicago, Chicago.

Tambiah, S. J. 1976. World Conqueror and World Renouncer: A Study of Buddhism and Polity in Thailand Against a Historical Background. Cambridge Studies in Social Anthropology 15. Cambridge University Press, Cambridge [Eng.]; New York.

———. 1982. The Renouncer: His Individuality and his Community. In Way of Life: King Householder, Renouncer; Essays in Honor of Louis Dumont, edited by

T. N. Madden, pp. 299–321. Motilal Banarsidas, New Delhi.

———. 1984. The Buddhist Saints of the Forest and the Cult of Amulets. Cambridge University Press, Cambridge.

Thapar, R. 1966. A History of India: Volume 1. Penguin Books, London.

———. 1984. From Lineage to State: Social Formations in the Mid-First Millennium b.c. in the Ganga Valley. Oxford University Press, Delhi.

———. 1997. Asoka and the Decline of the Mauryas: With a New Afterword, Bibliography, and Index. Oxford University Press, Delhi.

———. 2000. Cultural Pasts: Essays in Early Indian History. Oxford University Press, Delhi.

———. 2002. Early India: From the Origins to ad 1300. University of California Press, Berkeley.

Tilley, C. 1999. Metaphor and Material Culture. Blackwell, London.

Trainor, K. 1997. Relics, Ritual, and Representation: Rematerializing the Sri Lankan Theravada Tradition. Cambridge University Press, Cambridge.

Trautmann, T. R. 1971. Kautilya and the Arthasastra: A Statistical Investigation of the Authorship and Evolution of the Text. Brill, Leiden.

———. 1997. Aryans and British India. University of California Press, Berkeley.

Trautmann, T. R., and C. M. Sinopoli. 2002. In the Beginning Was the Word: Excavating the Relations Between History and Archaeology in South Asia. Journal of the Economic and Social History of the Orient 45(4):492–523.

Turner, V. 1967. Forest of Symbols: Aspects of Ndembu Ritual. Cornell University Press, Ithaca, NY.

Van Dyke, R. M., and S. E. Alcock (editors). 2003. Archaeologies of Memory. Blackwell, Malden.

van Gennep, A. 1960. The Rites of Passage. Routledge & Kegan Paul, London. Venclová, N. 1998. Mšecké Zehrovice in Bohemia: Archaeological Background to a Celtic

Hero. Kronos B. Y. Editions, Sceaux.

Walker, W. H. 1996. Ritual Deposits: Another Perspective. In River of Change: Prehistory of the Middle Little Colorado River, edited by C. E. Adams, pp. 75–91. University of Arizona, Tucson.

———. 1998. Where Are the Witches of Prehistory? Journal of Archaeological Method and Theory 5:245–308.

———. 1999. Ritual, Life Histories, and the Afterlives of People and Things. Journal

of the Southwest 41:383–405.

———. 2002. Stratigraphy and Practical Reason. American Anthropologist


Wallace, A. F. C. 1956. Revitalization Movements. American Anthropologist


———. 1966. Culture and Personality. Random House, New York.

Weber, M. 1978. The Routinization of Charisma. In Economy and Society, edited by

G. Roth and C. Wittich, vol. 1, pp. 246–254. University of California Press, Berkeley.

———. (1905) 1958. The Protestant Ethic and the Spirit of Capitalism. Scribner,

New York.

Wheeler, R. E. M. 1948. Brahmagiri and Chandravalli 1947: Megalithic and Other Cultures in Chitaldurg District, Mysore State. Ancient India 4:180–310.

———. 1954. Rome Beyond the Imperial Frontiers. George Bell and Sons, London. Wheeler, R. E. M., A. Ghosh, and K. Deva. 1946. Arikamedu: An Indo-Roman Trading

Station on the East Coast of India. Ancient India 2:17–125.

White, D. G. (editor). 2000. Tantra in Practice. Princeton University Press, Princeton, NJ.

Whitley, D. S., and K. Hays-Gilpin (editors). 2008. Belief in the Past: Theoretical Approaches to the Archaeology of Religion. Left Coast Press, Walnut Creek, CA.

Willis, M. 2009. The Archaeology of Hindu Ritual. Cambridge University Press, Cambridge.


Adorno, Theodor, 38 agency. See practice theory

Ajanta, 138, 149, 157–163, 166, 175,

191, 199

Ajivikas, 26, 74, 82, 89–90

Alexander the Great, 18, 21, 75

Amaravati, 22, 115–120, 123, 181, 185,


Ananda, 110, 177

anthropological archaeology, 12–13

Anuradhapura, 25

Arikamedu, 21

Arthashastra, 17, 204

Aryans, 10–11, 147

Asad, Talal, 50–53 asceticism

archaeological evidence of, 30–31, 110, 176–177, 208,

226, 229

of the Buddha, 2, 4, 146

communal/collective, 33, 105,

178–179, 226–227

ideal, 3, 33, 173–179, 228–229

origin, 32–33, 173–179, 226

wandering/solitary, 2, 74, 98–100,

178, 205, 209–211, 224 (see also

forest monks)

Ashoka, 75–81

inscriptions of, 20, 26, 77–82, 85–87,

90–93, 148

legitimations of, 71–72, 81,

86–89, 103

redistribution of relics, 23, 83

support of Buddhism, 81, 85, 88,

102, 183

Ashokavadana, 76

attenuation, 136–141, 143, 170–172,

183–185, 192, 194, 215, 228

Aurangabad, 139, 161, 163–165,


Avalokiteshvara, 151, 157

Bagh, 161, 163

Bairat, 90–93, 96–97, 101–102,

120, 122

Bala, 156, 168

Bavikonda, 127, 132, 166, 199

Bedsa, 126, 139, 142, 199

Bell, Catherine, 38, 49–50, 174

Bhaja, 27, 125–126, 130–131, 165,

199, 217

Bharhut, 21–22, 113–123, 175, 181,

186, 198

Bloch, Maurice, 38, 49

Bodh Gaya, 21, 23–25, 71, 117,

183–185, 195–196, 200, 217, 223

Bodhi tree. See Bodh Gaya Boivin, Nicole, 60

Bradley, Richard, 5, 49, 53

Brahmagiri, 22

Brahmanism, 17, 74, 94, 147 Buddha images

as an invention of tradition, 32–33, 177, 228

Gandharan, 28, 155, 157

Mathuran, 28, 155, 157

origin, 2, 28–29, 155–158, 168–176,

185, 195

as signs, 88, 162, 170–173, 175, 194

taboo on, 28–29

in viharas, 163–164, 166, 178, 186,

191, 213–214, 218

Buddhamitra, 156, 168

Carrithers, Michael, 4, 98–100, 210

Chalukyas, 205, 221

( 247 )

Chunar quaries, 86–89, 91

circumambulation, 25, 91–92, 96–97,

120–123, 134–135, 144–145,


Coningham, Robin, 22–23, 83, 227, 233

contradictions, 63–69, 97–101, 133,

178, 232

Cunningham, Alexander, 21, 114,

128, 195

Dehejia, Vidya, 118, 138

Delhi Sultanate, 17, 220

Dhamek stupa, 184–185

Dharmarajika stupa, 115–116, 119, 123,

181, 185, 198

dialectic, 35–37, 48, 54, 59, 69, 173, 233

disjuncture. See contradictions Douglas, Mary, 46, 206

Dravidian languages, 10–11

Durkheim, Emile, 36, 42–45, 47–48, 50,

69, 92, 96

early Buddhism, 7–8

Eaton, Richard, 18, 220–222

Ellora, 161, 163–164, 190–194, 204,

210, 219, 229

epigraphy, 19–20, 112–113

false consciousness, 38–39, 42

Faxian, 18–19, 31, 151, 177–178,

197, 215

forest monks, 2–3, 31, 98–101,

154–155, 175–179, 202, 205, 210,

229. See also asceticism, solitary Foucault, Michel, 62, 65–66

Fowles, Severine, 51–52, 118

Frankfurt School, 36, 38

Fritz, John, 5

Ganesha, 187, 200, 221

Geertz, Clifford, 48, 54, 63–68

Gell, Alfred, 59–61

Ghaznavids, 219–221

Ghurids, 220–222

Giddens, Anthony, 63, 68

Gramsci, Antonio, 38

Gudimallam linga, 192–193

guilds, 20–21, 28, 106, 108–109, 135

Guptas, 147–150, 169, 188, 196

Gurjara-Pratiharas, 203–205, 219

Hawkes, Jason, 22

Heart Sutra, 152–154

hegemony, 38–39

higher criticism, 15–16 Hinayana Buddhism. See early

Buddhism Hinduism, 147, 187–188

syncretism, 186–197, 230

re-occupation of Buddhist sites, 197–200

conversion to, 2, 33, 180,

200–201, 204

See also Shiva; Vishnu Hobsbawm, Eric, 31, 173–174

Huns, 149

ignoring the problem, 67–69, 101, 105,

131, 133

implied mass, 135–144, 162, 171–172,


incoherence of religion, 36, 62–69,


Indika of Megasthenes, 18

Indo-European languages, 9–11 inscriptions. See epigraphy Insoll, Timothy, 52

invention of tradition, 31–33, 146,

173–177, 221, 226, 228

Jaggadala, 204, 217

Jainism, 74, 95, 108, 180, 190,

199, 203–204, 209–211,

219–224, 230

Jamalgarhi, 159

Jaulian, 128, 160, 199

Jivakarama monastery, 82

Junnar, 126, 139

Kalawan, 128, 159, 199

Kalinga, 77–79, 106, 147

Kanaganahalli stupa, 116, 181, 198

Kanheri, 126, 139, 193

Karla, 126, 131, 139, 194, 200, 217

Kausambi, 21, 83, 156, 168, 175

Kharavela, King, 106, 147

Kondane, 126

Kondivte, 89–93, 96–97, 101–102,

120–122, 130–131, 139

Kushanas, 106–107, 147, 149, 156

Kushinagar, 23

Index ( 249 )

Lalitagiri, 127, 161, 182, 222

land, gifts of 33, 147–150, 169, 200,


Latour, Bruno, 59–60

linga, 188–194, 200

Lomas Rishi, 26, 90–93, 102, 133

Lopez, Donald, 9, 152–154, 206, 208

Lumbini, 22–25, 83, 88–93, 96, 102,

183, 197, 227

Mahabharata, 17, 147, 188 Mahabodhi temple. See Bodh Gaya

Mahaparinibbana-sutta, 23, 83, 93–94,

102, 110, 119

Mahayana Buddhism, 7–8, 29–30,

150–155, 169–179, 228

Bodhisattvas, 98, 151–154, 157, 207 Mahmud of Ghazni, 219

Mainamati, 215

Maitreya, 157

Malinowski, Bronislaw, 63–65, 68

mandalas, 207, 218

Manikwala, 159

Marshall, John, 114–115, 128–129, 133

Marx, Karl, 35–39, 43–44, 48, 61, 69

materiality, 32, 58–62, 85, 101, 167,

169–173, 217

Mattavilasa Prahasana, 209

Mauryas, 75–82, 103, 147

columns/pillars, 83, 86–89, 91,

115, 148

See also Ashoka

memory. See invention of tradition Mitra, Debala, 24–25, 115, 195

Mohra-Moradu, 128, 159

Moore, Jerry, 41–42, 86, 166, 185, 222

Morrison, Kathleen, 22, 126

Muhammad Ghori, 205, 220, 223

Mulasarvastivada-vinaya, 111

Nagarjuna, 152, 154, 161, 173

Nagarjunakonda, 158–161, 166–167,

175, 199

Nalanda, 19, 26, 204, 206, 208,

211–219, 222–223

Naropa, story of, 208 Nasik, 150

Ndembu, 46–47

Northern Black Polished ware, 84 nuns, Buddhist, 8, 26, 77, 156, 226

Odantapuri, 204, 217, 222

Palas, 203–205, 211, 217

Pali Canon, 15, 77, 103, 109

Peirce, Charles Sanders, 36, 55–58, 61,

88, 117–118, 137, 143

Perfection of Wisdom Sutras, 151–154 Periplus of the Erythraean Sea, 18 pilgrimage, 2, 22–25, 31, 102, 119, 123,

168, 177

Pippala, 128, 159

practice theory, 36, 38, 61–62, 136,

171, 231

pradakshina. See circumambulation

pratyekabuddhas, 100–101, 178

Puranas, 17, 147, 149, 188

Rajgir, 31, 82, 177

Ramatirthan, 127

Ramayana, 17, 147, 188

Ranger, Terrence, 31–32, 173–174

Rashtrakutas, 203–205

Ratnagiri, 143, 182, 222

Ray, Himanshu, 22, 28, 108, 126

refectories, 129–130

relics, of the Buddha, 23–25, 70, 77, 83,

102, 111

as icons of the Buddha, 94–95, 116–119, 141–145, 171–172, 192,

200, 218

Renfrew, Colin, 5

rites of passage, 45–46, 53

Rome, 21, 32, 173

routinization of charisma, 40, 97, 103,

105, 144

sacred and profane, 42–43, 50, 52–53

Sahlins, Marshall, 34

Sakas, 107

Salihundam, 124, 127, 132, 161, 199

Sanchi, 113–123, 144, 198, 218

architecture, 24, 26, 171, 185–186

iconography, 29, 87

Sangam literature, Tamil, 17 Sankaram, 124, 127, 161, 199–200

Sanskrit, 9–11, 15

Sarnath, 23–24, 71, 87–89, 164, 171,

181, 211, 213–214, 218–219

Buddha images, 156–160, 168, 175, 185

Dhamek stupa, 183–185

Satavahanas, 17, 106, 141

scholasticism, monastic, 2–3, 173,

176–179, 202–205, 207–210,

219, 224

Schopen, Gregory, 15, 19–20, 94, 108,

112–113, 119, 154–158, 233

semiotics, 54–58, 61–62, 88, 93–96,

117–118, 142–144, 169–173. See

also Peirce, Charles Sanders Sena Empire, 205

Shaw, Julia, 22, 28, 114–115, 144

Shimada, Akira, 22, 115

Shingardar stupa, 185, 199 Shinkot relic casket, 94 Shiva, 187–189, 192–194

shravakas, 100, 178

Shravasti, 83, 155, 168, 175

Shungas, 106

Silk Roads, 109, 177, 219, 230

Singh, Upinder, 147–148, 187

social evolution, 13

social solidarity, 42–45, 48, 62, 96,

122, 231

Somapura, 183, 204, 215–219, 222

South Asian studies, 12–13 Spink, Walter, 162

Strong, John, 76–77, 81 subjects and objects, 59–62 Sudama, 26, 90–93, 102, 133

Sugandhi, Namita, 77, 86

Sutta Pitaka, 82

Takht-I-Bahi, 127–128, 159–160,

199, 219

Tambiah, Stanley, 98–101, 205,

210, 233

Tantric Buddhism, 154, 201–203,

205–210, 212, 217–218, 224,


Taxila, 21, 115–116, 127–133, 159–160,

181, 185, 198–199

temple desecration, 220–223

Thapar, Romila, 17, 22, 27–28, 73–74,

108, 148, 188

Theravada Buddhism, 7, 109

Thotlakonda, 22, 126–127, 132,

166–167, 199, 217

Trautmann, Thomas, 10, 17

Trobriand Islands, 63–64

Turner, Victor, 45–47, 55, 69, 92

Udayagiri, 182, 192–194, 222

Vaisali, 70, 83–85, 94, 96–97, 102, 113

Vajrayana Buddhism. See Tantric Buddhism

Vakatakas, 147, 149–150, 161, 169

Van Gennep, Arnold, 45–47

Vedas, 9, 14, 17–18, 188

Vedic ritual, 74, 149, 188

Vikramasila, 183, 215–218, 222

Vishnu, 187–189, 200, 221, 230

Weber, Max, 36, 39–42, 44, 69,

108, 170

Wheeler, Mortimer, 21–22

White, David, 207, 209

Xuanzang, 18–19, 197–198, 202,


Yijing, 18–19, 204, 211–215