Chapter 1. The origins of Mahāyāna, evidence and hypotheses¶

Essay 4 turns to Mahāyāna. Before its doctrines, however, an awkward historical problem needs examination. Most Mahāyāna scriptures, including the Prajñāpāramitā, Lotus, Avataṃsaka, Vimalakīrti, and Pure Land sūtras, are absent from the schools' early Āgamas. They began to appear around the turn of the Common Era, yet all open with "Thus have I heard" and claim to have been spoken personally by the Buddha. Where did they come from, and on what grounds do they claim to be his teaching? The present chapter first states the problem, then considers several explanations.

The terminology needs explanation because it concerns a principle adopted in the present book. "Hīnayāna," the "inferior vehicle," was coined by Mahāyāna to disparage others. Schools so described, such as Sarvāstivāda and Vibhajyavāda, never called themselves by that name, and the present book does not use it. Nor does it use "śrāvaka vehicle" or "śrāvaka disciples" as its own terms. Speaking of a "śrāvaka vehicle" already accepts the Mahāyāna classification that opposes a bodhisattva vehicle to it, conceding the framework of "vehicles" in advance. Yet people who followed the Buddha's teaching and sought liberation for themselves had never boarded any "vehicle." They were practitioners walking a path, taking themselves as an island and the teaching as an island. The vehicle framework was imposed afterward by Mahāyāna. The present book therefore uses precise names of schools or people whenever possible. When a collective term is necessary, it uses "practitioners of the path to liberation." Where people who hear the teaching must be distinguished from pratyekabuddhas, it uses "practitioners of the path to liberation through hearing the teaching" and "practitioners of the path to liberation through solitary awakening."

1. Why late scriptures claim to be the Buddha's word¶

We can begin with what can be established.

The discussion below places the more frequent appearance of the name Mahāyāna in inscriptions in the late fifth to early sixth century.

Gāndhārī fragments from around the first to second centuries are the earliest known physical Mahāyāna manuscripts. Written on birch bark, they were found in the region of present-day Afghanistan and Pakistan.¹ By around 179, Lokakṣema had translated the Daoxing Prajñāpāramitā, a Chinese version of the Eight Thousand Lines, and the Pratyutpanna Samādhi Sūtra. His translations provide a firm documentary date by which Mahāyāna scriptures existed.² Earlier dating can only be inferred. Linguistic features and layers of thought can establish relative chronology within a text, but cannot fix it to an absolute timeline.

The difficulty is that the scriptures are absent from the early Āgamas memorized independently over generations by different schools and available for comparison, as explained in 2.2. The Āgamas and Nikāyas preserved by the schools agree closely despite independent transmission because they share an earlier oral basis. The absence of Mahāyāna scriptures from those transmissions indicates that they do not belong to the same early stratum of recitation. On the historical evidence alone, they are new works that appeared later, rather than teachings already being recited during the Buddha's lifetime. Yet every one claims to be his word. The tension between late appearance and that claim is the core of the problem of Mahāyāna origins.

A quantitative fact also deserves attention. Drewes's count shows very few traces of Mahāyāna in archaeological inscriptions and donation records before the Gupta dynasty, approximately before the fifth century. The great majority of donors still belonged to mainstream schools.³ Several centuries separated the appearance of Mahāyāna scripture from the widespread adoption of Mahāyāna belief. The interval is itself evidence. Early Mahāyāna concerned a very small minority, rather than a mass movement sweeping through the saṅgha.

2. Legends and historical method, the nāga palace and three turnings of the wheel¶

Mahāyāna naturally supplied accounts of its own origins. The most famous has Nāgārjuna enter the nāga palace to retrieve scripture. The Buddha had taught the Prajñāpāramitā but instructed the nāga king to preserve it in his palace until conditions were ready for Nāgārjuna to bring it back to the human world, a story compared with another legend in 2.5. More generally, the Buddha was said to have taught Mahāyāna to bodhisattvas and gods in other realms. According to those accounts, contemporary practitioners of the path to liberation lacked the necessary capacity and had not heard the teachings. Later, classification by the "three turnings of the wheel" assigned the teachings of the path to liberation to a first turning, placing Mahāyāna at a higher level in subsequent turnings.

The approach of the present book applies the same standard as in 2.5. Such accounts are legends that authorize scripture and are not accepted as historical facts. We do not believe Nāgārjuna actually descended to a nāga palace, just as we do not believe the Buddha actually ascended to the Heaven of the Thirty-Three. The distinction made there must also be retained. The legend itself must be distinguished from the material to which it is attached. What matters is the evidential standing of that material. Mahāyāna scriptures lack early corroboration across schools, and their textual history points to late composition. Recognizing that does not diminish their intellectual value, as Section 5 explains.

3. Growth within the schools rather than a separate sect¶

A common misconception treats Mahāyāna as another "school" alongside Sarvāstivāda, Vibhajyavāda, and the other established schools, as though it emerged independently from a schism. The historical evidence describes a different pattern.

For a long period, Mahāyāna had no independent ordination lineage and did not split away from the saṅgha. It was closer to a current of thought within the existing schools. A monk could receive Sarvāstivāda ordination and live under its Vinaya while reciting Prajñāpāramitā and taking bodhisattva vows. Xuanzang's observations during his seventh-century journey and Yijing's slightly later record describe the same situation. Mahāyāna and non-Mahāyāna monks lived in one monastery under the same discipline, differing in the scriptures they read, the figures they worshiped, and the vows they took, rather than in ordination lineage.² Schopen's systematic studies of inscriptions further support the conclusion. In records of stūpa construction and relic offerings, recipient communities identified by affiliation usually bear mainstream school names. Inscriptions explicitly naming "Mahāyāna" as the recipient group remained rare for a long period. The name appeared more often only in the late fifth to early sixth centuries, several hundred years after the scriptures.⁴

An institutional distinction needs to be clear. Essay 3 explained that the schools differed centrally in their Vinayas and associated ordination lineages. "Splitting the saṅgha" (saṃghabheda) has a technical definition in the Vinaya. The definition current in the age of the schools concerned holding a separate uposatha ceremony within the same ordination precinct.⁵ Mahāyāna never established separate precincts or ordination lineages, so it never constituted schism in that disciplinary sense and was not an independent school. Calling it a "Mahāyāna sect" can therefore mislead. It was an intellectual movement across schools, rather than a branch defined by the Vinaya.

The movement's geographical distribution can also be traced, though the evidence remains fragmentary. Early Mahāyāna manuscript and inscriptional evidence is concentrated in northwestern India, in Gandhāra and Jibin, and the southern Deccan, in Andhra, around present-day Andhra Pradesh. Images possibly associated with Mahāyāna have been found at Nāgārjunakonda and Amarāvatī in Andhra, and tradition links Nāgārjuna himself to the region. Caution is needed, however. Geographical association does not establish a place of origin, and the relevant inscriptions cannot always be assigned to Mahāyāna with certainty. The evidence is still being assembled, and conclusions would be premature.

The map marks three geographical centers indicated by existing evidence. Gandhāra is where the earliest manuscripts were found. Andhra was a Mahāsāṃghika stronghold and an important region for Lokottaravāda, the doctrine of the supramundane Buddha. The area around Nālandā was a central region of coexistence between Mahāyāna and the schools, as recorded by Yijing and others.

The discussion identifies Andhra as an important region for Lokottaravāda and cautions that a geographical association does not establish a place of origin. The Chinese source map marks Amarāvatī and Nāgārjunakonda with the symbol for monasteries in which Mahāyāna and the schools coexisted. The changed symbols in this English image do not establish a different classification.

4. Modern hypotheses¶

Scholars have proposed several hypotheses about the people among whom Mahāyāna arose and the process of its emergence, without reaching a settled conclusion. The table gives an overview before the individual discussions.

The rise and decline of the hypothesis of lay origins. In the 1960s, Hirakawa Akira proposed that Mahāyāna began as a grassroots movement of lay followers making offerings around stūpas, in tension with the monastic community. The proposal has intuitive appeal. The bodhisattva path emphasizes the possibility of laypeople becoming Buddhas, as in the Vimalakīrti Sūtra, and stūpa worship was an important setting for lay devotion. Schopen's inscriptional studies weakened the hypothesis, however. Monastics account for a considerable proportion of donors around stūpas in the records. The few inscriptions that explicitly identify a school name refer to the recipient saṅgha, rather than donors identifying themselves. If Mahāyāna were a lay movement, the records should contain more traces of lay followers, but the evidence runs the other way. Nattier's study of the Inquiry of Ugra (Ugraparipṛcchā) is more direct. The bodhisattva path described there is extremely demanding, requiring renunciation, meditation, asceticism, and residence in the forest. It is entirely a way of life for elite monastics, far removed from popular lay devotion.⁶

The hypothesis of origins among forest meditators. The collective shift in work by Harrison, Nattier, Drewes, and others directs attention to a small number of unconventional monastics within the schools, especially meditators who chose to live apart in the forest. Institutionally, they held ordination in existing schools, but lived more strictly than ordinary monastery residents. They concentrated on samādhi and sometimes identified themselves as "sons of good family" or "bodhisattvas." They produced new scriptural texts, some directly expressing meditative experience in writing. The Pratyutpanna Samādhi Sūtra, for example, describes concentrated recollection of Amitābha and direct vision of the Buddha's body in meditation. On this evidence, the earliest forms of some Mahāyāna scriptures may have been reports of meditation or religious literature produced by groups practicing particular methods. Its proponents have themselves marked the limits of the hypothesis. When Harrison proposed it in 1995, he wrote, "My hypothesis, yet to be substantiated in detail." Boucher's 2008 study of the Rāṣṭrapālaparipṛcchā judged the additional ascetic practices to be a reformist movement within mainstream Buddhism, and forest dwelling to be a retreat taken when conditions in the monastery did not allow otherwise: "If they have to live in the forest now, it is because the conditions in the monastery do not make it possible to live as real monks otherwise." Drewes's 2010 count found that only two of the dozen or so texts linked to Lokakṣema, the Pratyutpanna Samādhi Sūtra and the Kāśyapaparivarta, actually advocate forest dwelling. The monastic side of the hypothesis is now generally accepted; the forest side is the position of a few sūtras.⁷

Worship of scripture. In 1975, Schopen identified a recurring theme in a group of Mahāyāna sūtras. Wherever a scripture is present, taught, or copied is itself equivalent to a shrine (caitya) containing the Buddha's relics. He suggested on that basis that early Mahāyāna was a loose federation of cults, each centered on its own text. The hypothesis has since been weakened. Drewes reread the passages and concluded that no evidence indicates that Mahāyāna sūtra shrines ever existed, that the sūtras recommend memorization more often than copying, and that writing was not peculiar to Mahāyāna. When Schopen reprinted the paper in 2005, he changed the "unambiguously negative attitude" of early Mahāyāna toward the stūpa cult to an "ambiguously negative" one. What remains is the weaker version, Gombrich's 1988 hypothesis about writing. The early canon was preserved by the saṅgha's organized oral transmission, a new text could not enter the curriculum of recitation, and anything inconsistent with the existing sūtras and Vinaya was discarded. Mahāyāna sūtras survived because they were written down. That version explains survival and textual form, not authorship.⁸ The dharmabhāṇaka hypothesis. The hypothesis begins with a word. Dharmabhāṇaka, which Lokakṣema rendered as "dharma master" or "sūtra master," occurs, by Drewes's 2011 count, thirty-seven times in the Sanskrit Eight Thousand Lines, sixty-one times in the Lotus Sūtra, fifty times in the Samādhirāja, and thirty-eight times in the Suvarṇaprabhāsa, yet "is not known to occur in non-Mahāyāna Indian Buddhist texts." Non-Mahāyāna literature calls a preacher a bhāṇaka or dharmakathika. Buddhas and celestial bodhisattvas who preach are not called dharmabhāṇakas; "the term is consistently used to refer to an actual human occupation." In the sūtras, dharmabhāṇakas memorize the text, transmit it orally to the students who follow them, sometimes have the students copy it, and preach in public on uposatha days and in monasteries, towns, and villages. The sūtras require that they be honored as Buddhas. Mahāyāna sūtras often entrust themselves to certain bodhisattvas and predict that they will return in a later age to reveal the sūtra, and the future revealers are usually described as dharmabhāṇakas. Drewes draws the inference: "Presuming that the authors of these texts presented themselves as reincarnations of the people about whom these predictions were made, these passages strongly suggest that they were dharmabhāṇakas." Shizutani Masao in 1954, MacQueen in 1982, and Harrison in 1993 had already identified dharmabhāṇakas as the compilers and disseminators of Mahāyāna sūtras, and Yinshun derived the figure from the reciter monks of the Vinaya. The hypothesis has limits of its own. Drewes himself says that what led dharmabhāṇakas to compose sūtras is unclear, and the samādhis in the sūtras are mostly presented as by-products of textual practice, the opposite of the hypothesis of revelation through meditation.⁹

The rise, abandonment, and revival of the school affiliation hypothesis. Another hypothesis falls outside the three accounts just discussed, predates all of them, and has recently been revived. It deserves separate notice. In the nineteenth century, Kern, Waddell, and Rhys Davids argued that Mahāyāna came from the Mahāsāṃghika branch. The hypothesis was abandoned in the late twentieth century, because it conflicted with another kind of evidence rather than simply lacking support. Harrison compared early Chinese translations of Mahāyāna sūtras and could identify no independent sect in them. Nor does any existing evidence show monks outside the schools' ordination lineages. Mahāyāna monks received ordination in the schools. Since Mahāyāna was not an ordination community, it cannot be aligned with one particular school. The point is the same as the conclusion of Section 3 of this chapter.

Karashima Seishi recently revived the hypothesis using new philological material. His argument follows a clear line. Schools disagree on whether Ajita and Maitreya are the same person. Theravāda and Sarvāstivāda say they are not. In Madhyama Āgama discourse 66, Ajita is predicted to become the wheel-turning king Śaṅkha, while Maitreya separately vows and is predicted to become a Buddha. They are distinct people. The Mahāsāṃghika Mahāvastu and the Saṃmitīya Saṃmitīyanikāyaśāstra say they are the same, with Ajita as the personal name and Maitreya as the family name. Several early Mahāyāna sūtras call the bodhisattva Maitreya "Ajita," suggesting that their authors belonged to a tradition identifying the two. Karashima also cites an inscription at the Kanaganahalli stūpa naming Ajita as the future Buddha.¹⁰

The material shows that a group of scriptures and a school share the same identification of the figures, while the desired conclusion is that the scriptures originated in that school. The inference must also answer counterevidence from ordination lineages, school affiliations stated in inscriptions, and the circulation of Mahāyāna sūtras across schools. The article does not directly address these objections. Each inferential step is sound, but the argument does not reach the step required by the conclusion. The mainstream response is explicit. In the collection he edited, Harrison denies a special relationship between Mahāyāna and Mahāsāṃghika, while Silk's extended review directly names the hypothesis.¹¹ The inscriptions also fail to support a single affiliation. Their editors judge that they indicate no particular school, and a subsequent specialized study concludes that at least two monastic communities coexisted at the site, one belonging to the Saṃmitīya group.¹²

The hypotheses need not exclude one another. A more likely situation is that small, differing circles within the schools developed independently in various places around particular meditations or scriptures. Early Mahāyāna was itself not a unified movement. It was closer to multiple small traditions with their own systems, later gathered under a common name and shared themes such as bodhicitta, prajñā, and emptiness. The position taken in the present book is that the three hypotheses answer three different questions. The forest hypothesis answers how some of these people lived, the writing hypothesis answers how the texts survived, and the dharmabhāṇaka hypothesis answers who produced them; choosing one of them as an exclusive theory of origins mistakes the question. The present book takes the dharmabhāṇaka hypothesis as its main line: the authors were monks within the schools whose occupation was preaching and transmitting sūtras, forest dwelling was the ideal of some of those sūtras, and writing was the condition under which all of them survived. The hypothesis of lay origins cannot be sustained on the available evidence. The overall question remains open, however. Mahāyāna had multiple origins rather than a single line of development. The sources and their mutual criticisms are set out in the second and third essays of the Chinese-language series 专题系列《谁写了大乘经》 (Who Wrote the Mahāyāna Sūtras).

How to read the table's "current standing" column. The last column records where the theories stand at present and will become outdated first. School affiliation is a ready example. Saying only "abandoned" is accurate, but gives a result whose reasons must be sought elsewhere. Once someone revives the hypothesis with new material, the status alone cannot show how the new argument relates to the old objections. First, establish whether the claim is new or revived. If it is revived, examine why it was abandoned. Second, examine the strength of the proponent's wording. Karashima uses "evidence," "clue," and "perhaps" throughout, rather than "proof." His title promises more evidence and a clue. He makes strong claims about affiliations that are impossible and weaker claims about affiliations that are possible. Third, establish whether anyone has responded directly. The last check is the easiest because whether an argument has received a response is verifiable, whereas whether it is correct usually is not.

The book therefore supplies reasons in each table entry as well as status. A judgment remembered only as a status collapses when a counterexample appears. When its reasons are remembered, a counterexample overturns only the particular reason it addresses.

5. How "the Buddha's word" was redefined¶

The central question is how late scriptures can call themselves the Buddha's word. The answer lies in Mahāyāna's redefinition of the criterion of buddhavacana itself, rather than in the legends of the nāga palace.

One route of redefinition moved the standard from origin to content. Any well-spoken teaching that accords with sound understanding and leads to liberation counts as the Buddha's word, or "what is well spoken is the Buddha's word." Its standing depends on whether it conveys the Buddha's teaching, rather than whether the historical Buddha personally spoke it at a particular place. The other route allowed the Buddha to "continue teaching." He was more than the historical figure who had died, being also the enduring dharma body. Teachings received by seeing a Buddha and hearing him in meditation, as in the revelations discussed above, therefore counted as his words. Together, the routes allowed new scriptures to qualify as the Buddha's word.

Both redefinitions have direct textual expressions in Mahāyāna scripture. In the Eight Thousand Lines of Prajñāpāramitā, Subhūti asks the Buddha where he should learn the perfection of wisdom. The Buddha answers, "From the Buddhas." "Buddha" no longer refers only to the one who died at Kuśinagara, but to multiple beings across time and space. In the Lotus Sūtra, the Buddha declares that he has not actually passed into final nirvāṇa. His passing is an "expedient manifestation" (upāyakauśalya), and he remains permanently on Vulture Peak, teaching. The claim effectively dissolves the historical death itself, making the Buddha's teaching temporally unending. Within the framework, new scriptures are understood as teachings that the "ever-present" Buddha continues to give bodhisattvas, rather than forgeries by later people.

Connecting the account with earlier discussions reveals a recurring structure. Each seeks authority for material absent from the early canon.

An outline of the answer to the question left in Essay 2 is now visible. Later developments could not fit within a closed early canon, and traditions devised different responses. Some invented scenes of transmission beyond the human world, others redefined what counted as the Buddha's word, and still others moved transmission outside writing altogether. The position taken in the present book is that, historically, Mahāyāna sūtras are new works from a later period, rather than the historical Buddha's words. Yet whether a text was personally spoken by the Buddha, whether it has intellectual value, and whether it remains consistent with the early account of liberation are distinct questions. A sūtra can be late and not spoken by the Buddha while still having intellectual value. It can also have intellectual value while departing from the early account of liberation. The questions must be kept separate.

Redefining the Buddha's word from provenance to accordance with sound understanding is itself a consequential intellectual act. It moves the criterion from who said something to whether it is right. Yet Mahāyāna supplies its own standard for determining what is right. Caution is needed because some Mahāyāna schools changed even the criterion of liberation, replacing arhat attainment with Buddhahood or positing nonabiding nirvāṇa, for example. "Accordance with sound understanding" is therefore not a neutral, ready-made criterion. Mahāyāna established it for itself. The later discussions of individual Mahāyāna schools follow the same approach. They set aside claims to ancient authority and take the doctrines seriously. They also continually make clear whether the doctrine still expresses the early account of liberation.

The next chapter turns to the early Mahāyāna scriptures themselves, examining the Prajñāpāramitā, Lotus, Avataṃsaka, Pure Land, and Vimalakīrti groups. The bodhisattva path as a new ideal of practice and its relation to the schools follow. The systems that made Mahāyāna into systematic philosophy are reserved for Essay 5 on Madhyamaka and Essay 6 on Yogācāra.

Revision note¶

Section 4 of this chapter was rewritten in October 2026. The first version took the combination of the forest-meditator hypothesis and the worship of scripture as the best documented explanatory framework. Rereading the sources for the Chinese-language series Who Wrote the Mahāyāna Sūtras changed three judgments. First, the proponents of the forest hypothesis have themselves limited it: Harrison called it a hypothesis yet to be substantiated, Boucher read forest dwelling as a retreat taken when conditions in the monastery did not allow otherwise, and Drewes counted only two texts advocating forest dwelling among those linked to Lokakṣema. Second, the worship of scripture has been weakened by Drewes, and Schopen has retreated; writing explains which sūtras survived, not who wrote them. Third, the term dharmabhāṇaka occurs only in Mahāyāna texts and refers to a real human occupation, the sūtras usually identify their predicted future revealers as dharmabhāṇakas, and Shizutani Masao, MacQueen, Harrison, and Drewes all identify dharmabhāṇakas as the compilers and disseminators of Mahāyāna sūtras. The chapter therefore now takes the dharmabhāṇaka hypothesis as its main line, reduces the forest hypothesis to a note on some sūtras, and reads the worship of scripture as a condition of survival. A row for the dharmabhāṇaka hypothesis has been added to the table and its "current standing" column revised, and the paragraphs on forest meditators and worship of scripture and the concluding paragraph of Section 4 have been rewritten. The sentences stating the first version's position are reproduced here for comparison: "The position taken in the present book is that the combined hypotheses of origins among forest meditators and worship of scripture currently offers the best documented explanatory framework. The hypothesis of lay origins is difficult to sustain on the available evidence. The overall question remains open, however. Mahāyāna had multiple origins rather than a single line of development." The argument and sources are given in the second and third essays of that series.

Further reading¶

• Paul Harrison. "Who Gets to Ride in the Great Vehicle? Self-Image and Identity among the Followers of the Early Mahāyāna." Journal of the International Association of Buddhist Studies 10, no. 1 (1987): 67–89.

References¶