<<

A STUDY OF THE COMPOSITION, TRANSMISSION, AND DEVELOPMENT OF THE KĀŚYAPAPARIVARTA

A Thesis Presented to the Faculty of the Graduate School of Cornell University In Partial Fulfillment of the Requirements for the Degree of Master of Arts

by Kittipong Vongagsorn August 2020

© 2020 Kittipong Vongagsorn

ABSTRACT

This thesis is a study of the Kāśyapaparivarta, an early Mahāyāna sūtra which was translated into Chinese between as early as the second and tenth centuries. It has two related goals: to investigate the textual history of the Kāśyapaparivarta by examining various versions that have been preserved in manuscripts, Chinese and Tibetan translations; and to study the stylistic features of the Sanskrit version of the Kāśyapaparivarta to see if they can tell us about the method of the composition of Mahāyāna sūtras. My central argument is that Mahāyāna sūtras were composed in written form which is different from the early that were composed orally. However, the oral/aural tradition was still the primary concern of Mahāyāna composers.

I start with the investigation of all extant versions of the Kāśyapaparivarta to illustrate the picture of the popularity of this text. Then, I compare all versions by using the methodology of textual criticism. I propose that the Kāśyapaparivarta has three stages of development: the early stage, the middle stage, and the final stage. Each stage shows some changes to the text in terms of structure, wording, and length. I propose that this variation might be the effect of writing that was used in the composition and transmission of Mahāyāna sūtras. In the second part of this thesis, I examine the stylistic features of the Sanskrit version of the Kāśyapaparivarta. I begin with the investigation of the theory that Mahāyāna sūtras were composed in written form. I argue that although Mahāyāna sūtras may have been composed in written form, the oral/aural tradition was still the central concern of Mahāyāna authors. I then examine the stylistic features of the text to find out to what extent the oral/aural tradition influenced the composition and transmission of the Kāśyapaparivarta. By analyzing the stylistic features, I conclude that the significance of the oral/aural culture never decreased in Mahāyāna . Most of the stylistic features were modeled on the early Buddhis texts. However, the inconsistency and the innovation of some stylistic features of the Kāśyapaparivarta might indeed suggest that it was composed in written form.

ii BIOGRAPHICAL SKETCH

Kittipong Vongagsorn was born and raised in Thailand. He received his first Bachelor’s degree from Silpakorn University, Thailand in 2014. In 2015 he won a scholarship from the Dhammachai International Research Institute (DIRI) to pursue higher degrees in . He went to the University of Washington and received his Post-Baccalaureate degree in South Asian Language and Literature (Sanskrit) in 2018. He received his M.A. in Asian Studies from Cornell University in 2020.

iii

ACKNOWLEDGEMENT

This thesis would not have been completed without the great support from many people. It is my privilege to acknowledge and thank those who are particularly significant for my studies. First and foremost, I wish to thank the Dhammachai International Research Institute (DIRI) for granting me a scholarship to pursue graduate degrees in Buddhist Studies abroad. Without their generous financial support and encouragement, it would be impossible for me to go this far. I am wholeheartedly thankful to be a part of their research project on . There are several professors throughout my studies who deserve special recognition for setting me on the path into the exciting world of the Sanskrit language. Professor Richard Salomon is my ādi-ācārya of Sanskrit who is always willing to answer any question no matter how absurd they are. His knowledge and curiosity about languages have always inspired me to better myself. Professor Collet Cox has been a compassionate teacher who always gives me very generous advice. Dr. Timothy Lenz guided me to another exciting world, the world of . Dr. Joseph Marino has always been my generous teacher and friend both at the University of Washington and Cornell. I also wish to thank professors at Cornell who have been my wonderful teachers and mentors throughout my studies. Professor Daniel Boucher has been my kindest teacher and advisor. He introduced me to the world of Religious Studies, another important world for my studies. He primarily inspired me to do this project and also guided me throughout. I could not have found any better professor to teach me all the Buddhist languages:

Chinese, Pāli, Tibetan and Sanskrit. Professor Lawrence McCrea led me to yet another world, the world of Indian epistemology. He made me see that Buddhism is not just Sūtra and , but also the sophisticated Śāstra. I would like to thank Phrasudhammayanavithes (Most Venerable Sudhammo), the chairman of DIRI who gave me the special opportunity to study abroad. Dr.Chanida Jantrasrisalai and all senior researchers of DIRI have been my generous mentors since the beginning of my studies. Thanks to Claire, Bruno, Liyu, Tarini, Andrew and my friends at the

iv

Department of Asian Studies with whom I have spent uncountable time discussing Asian Studies not limited to Buddhism, and who have helped me edit my papers and this thesis. Finally, if my study would make any benefit to the world of Buddhist Studies, I would like to dedicate it to Most Venerable Dhammajayo and Most Venerable Dattajīvo who are my spiritual teachers, and to my parents whom I owe my most special thanks. Without their unconditional love, generous support, kind encouragement and hard work, I would not have made it this far.

v

TABLE OF CONTENTS

BIOGRAPHICAL SKETCH ...... iii

ACKNOWLEDGEMENT ...... iv

TABLE OF CONTENTS ...... vi

LIST OF TABLES ...... viii

LIST OF ABBREVIATIONS ...... ix

PART ONE: The History of the Kāśyapaparivarta ...... 1 1. Introduction ...... 1 2. Overview of the Structure of the Sūtra ...... 4 3. Historical Background ...... 8 3.1 Date ...... 8 3.2 Name of the text ...... 10 3.3 Kāśyapaparivarta in the Mahāratnakūṭa Collection ...... 12 4. The Textual Development of the Kāśyapaparivarta ...... 14 4.1 Versions of the text ...... 14 4.1.1 Versions without verses ...... 15 4.1.2 Versions with verses ...... 18 4.1.3 Quotations in other works ...... 19 4.2 Three stages of textual development ...... 22 5. Discussion ...... 38

PART TWO: The Composition and Transmission of the Kāśyapaparivarta ...... 44 1. Introduction ...... 44 1.1 The Composition and Transmission of Pre-Mahāyāna Literature ...... 47 1.2 The Composition and Transmission of Mahāyāna Scriptures ...... 48 2. Stylistic Features in the Kāśyapaparivarta ...... 55 2.1 Formulaic expressions ...... 55 2.1.1 Formula of numerical lists ...... 56 2.1.2 formula of similes ...... 59 2.2 Repetitions ...... 61 2.3 Other Stylistic Features ...... 67 2.3.1 The arrangement of lexical strings in the waxing syllable principle ...... 67 2.3.2 Sets of related subjects ...... 74 2.3.3 Numerical order ...... 75 3. Discussion ...... 77

CONCLUSION ...... 84 vi

BIBLIOGRAPHY ...... 91

vii

LIST OF TABLES

Table 1: The comparison of sections 150-166 in each version ...... 30

viii

LIST OF ABBREVIATIONS

BHS Buddhist Hybrid Sanskrit KP Kāśyapaparivarta P Pāli Skt Sanskrit

T Taishō shinshū daizōkyō; texts are referred to by serial number followed by volume, page, register, and line numbers.

ix

PART ONE: The History of the Kāśyapaparivarta

1. Introduction

The Kāśyapaparivarta (hereafter KP) is one of the earliest translations of Mahāyāna texts into the Chinese language. It was translated in the 2nd century CE by Lokakṣema who was the first translator of Mahāyāna sūtras.1 Hence, this sūtra has been marked by scholars as an early

Mahāyāna sūtra by its terminus ante quem. However, it is uncertain how early its Indic archetype really is, since it could have been composed a few centuries or even a few years before the translation. The task of dating Mahāyāna scriptures is challenging and by all means uncertain.

Despite being uncertain of its early history, we are certain that after the KP’s composition, it became very popular in Central Asia, particularly in Khotan area where most of manuscripts were found. There are five complete Chinese translations dating from late 2nd century to 10th century CE, two Tibetan translations: one old version that remains only some fragments discovered in Dunhuang and one complete translation in the Kanjur collections, some quotations in Khotanese, one almost complete Sanskrit manuscript, and some earlier Sanskrit fragments.

Due to the variety of its sources, the KP is an interesting source for scholars of Buddhist studies who are captivated by either the study of the development of Mahāyāna literature or the philological study of Mahāyāna texts. Along these lines, I will be examining all versions of the

KP to study the development of the text and relationships between each version.

1 According to Harrison 1993, 135-169, there are 11 sūtras that are considered to be Lokakṣema’s translations. Among these sūtras, nine texts are preserved, and two texts are lost. The nine extant sūtras are Aṣṭasāhasrikāprajñāpāramitāsūtra, Pratyutpannabuddhasaṃmukhāvasthitasamādhisūtra, Drumakinnararājaparipṛcchāsūtra, Ajātaśatrukaukṛtyavinodanāsūtra, Avataṃsakasūtra, Lokānuvartanāsūtra, The Sūtra of Mañjuśrī's Questions Concerning the Career, Kāśyapaparivarta, and Akṣobhyatathāgatasyavyūhasūtra. The two lost sūtras are Śūraṃgamasamādhisūtra and the Sūtra of Illumination Meditation (Guanming sanmei jing 光明三昧經). 1

Several scholars have already examined multiple aspects of this text. Baron A. von Staël-

Holstein (1926), the pioneer scholar of the study of the KP, transcribed the Sanskrit manuscripts preserved in the St. Petersburg collection into Roman script and divided the text into 166 sections to facilitate the comparative study of the Tibetan and Chinese versions. His sectional divisions became the model of reference and has been employed by later scholars in their studies as well as in this study. Friedrich Weller translated all four Chinese versions and the Tibetan translation in relation to the Sanskrit manuscript into German.2 Weller also studied the

Mongolian version of this text with an emphasis on the linguistic perspective.3 Since the

Mongolian version is generally not different from the Tibetan translation, it is not included in this study. Weller also produced an index to Indian versions of each translation in two volumes, which is very useful for the comparative study of this text.4 Pāsādika reconstructed the missing parts of the Sanskrit manuscripts and then translated it into English.5 Some other scholars, such as Jonathan Silk, have focused on the philological aspects of the text as well as its commentaries.6 Even though this text has been studied by many scholars, none have ever studied the stylistic features of this text nor done a thorough comparative study of each version.7

The purpose of this present study is to understand the relationships between different versions of the KP and the impact of the writing on the composition and transmission of

2 See Weller 1962; 1964; 1966a; 1966b; [1968/69] 1987. 3 See Weller 1961a; 1961b; 1965. 4 See Weller 1933/35. 5 See Pāsādika 2015. 6 See Silk 2009a. 7 Silk 1994, 25 wrote in footnote 1 of his Ph.D. dissertation that he and Prof. Nagao were preparing a critical edition of the Kāśyapaparivarta in Sanskrit, Tibetan, and Chinese with annotated English translation. He expected to publish it in 1995. However, the project seems not to have been completed; we have only his unpublished translation of the Sanskrit version. The critical edition of the KP will definitely be very beneficial to the study of this text and the nature of early Mahāyāna sūtras. Vorobyova-Desyatovskaya 2002 has re-transliterated the Sanskrit manuscript which was primarily did by von Staël-Holstein. She also included other Sanskrit fragments and their comparison. However, she did not consider the Chinese and the Tibetan translations in her book. 2

Mahāyāna texts, which contributed to the textual variation in different versions as well as stylistic features of this text. The study is separated into two parts. In the first part, I will examine the differences between variant versions of the KP from the perspective of textual transmission by using the method of textual criticism. By conducting a thorough comparative study of each version of the text, I will suggest the way the text changed over time, and the effects of the writing on the textual transmission on these various versions.

This emphasis on understanding the impact of the writing on the transmission of

Mahāyāna sūtras and on the stylistic features leads me to the second part of the study. In that part, I will focus on the composition and transmission of the text. In doing so, it becomes evident that though some of the stylistic features of Mahāyāna sūtras are similar to their predecessors’, such as in the Pāli Canon, some characteristics are different and unique in

Mahāyāna literature. Scholars have suggested that some differences between Mahāyāna sūtras and the early Buddhist texts are the result of the change of mediums of the composition and transmission, specifically, from orality to literacy. To support their theory that Mahāyāna sūtras were composed and transmitted in written form, scholars look into two different places for evidence: the historical context and the stylistic features of the texts. While historical context has been studied by eminent scholars in Mahāyāna studies, I will scrutinize stylistic features of

Mahāyāna sūtras by using the KP as a case study.8 I will borrow the methodology employed by

Mark Allon (1997a) in his study of style and function of the Dīghanikāya in the Pāli Canon. In this study, Allon concludes that the stylistic features in the Pāli Canon confirm its oral

8 Very few scholars have investigated Mahāyāna sūtras as literature and scrutinized the literary devices used by Mahāyāna authors in their composition. Among those who have is Cole 2005, who studies 4 Mahāyāna sūtras: the Lotus Sūtra, the Diamond Sūtra, the Tathāgātagarbha Sūtra and the Vimalakīrtinirdeśa. Osto 2015, 115-135 follows Cole’s methodology in his study of Prajñāpāramitā strata. However, none of them pay attention to the stylistic features of Mahāyāna sūtras. 3

composition and transmission. Likewise, I will examine the stylistic features in the KP and compare them with the oral features in the Pāli Canon to find out to what extent they are similar to or different from the Pāli Canon. The scope of my research is limited to only one text, which should not be regarded as representative of all Mahāyāna sūtras. But considering the significance of this text and its status as an early Mahāyāna sūtra, a close investigation of its stylistic features will shed some light on the study of the composition and transmission of Mahāyāna sūtras.

2. Overview of the Structure of the Sūtra

The KP is one of many Mahāyāna sūtras dealing with the ethics of and their practices. Its core teaching depicts some important philosophical concepts of and

Yogācāra schools of Mahāyāna Buddhism.9 It is considered to be one of the early Mahāyāna scriptures because of the date of its first translation into Chinese, in the 2nd century CE, and because its content does not show many of the innovative developments of Mahāyāna Buddhism, but merely gives the general ideas of encouraging people to be a bodhisattva. It explains what a monk should do and should not do in order to become a real bodhisattva, and touches on the superiority of being a bodhisattva to being śrāvaka monks, while criticizing renunciants and monks who do not follow monastic practices seriously.

The text follows the basic traditional structure of early Buddhist sūtras. It starts with the opening formula or nidāna telling time, place, and persons that are involved in the sūtra. Then the Buddha, the main speaker of the sūtra, tells Kāśyapa, his disciple, the instruction of bodhisattvas. The first part of the text (§1-22) after the opening passage is presented in numerical lists. It typically starts with the phrase “These four qualities, Kāśyapa, lead to…” (Skt. catvāra

9 Vorobyova-Desyatovskaya 2002, vii. 4

ime kāśyapa dharmāḥ…, … saṃvartante). There are 11 pairs of lists, each one has 4 points of instruction. Each list is paired with another to create positives and negatives, for example, the first list (§1) discusses four qualities that cause the loss of bodhisattva wisdom, while the second list (§2) deals with four qualities that lead to the attainment of bodhisattva wisdom. This part of the sūtra resembles Catukkanipāta of the Aṅguttaranikāya in the Pāli canon (Skt.

Ekottarārikāgama). The list of topics includes ways to increase wisdom, good and bad friends of bodhisattvas, and the paths of bodhisattvas to attain the perfect enlightenment, etc.

Subsequently, the Buddha summarizes these numerical instructions with the thirty-two qualities of bodhisattvas (§23-28). This number is, of course, not a random number. It reflects the physical marks that all Buddhas have. The authors of the KP attempt to show that if Buddhas has thirty-two physical marks, bodhisattvas–future Buddhas–also need to have these thirty-two qualities to be distinct from śrāvakas. The thirty-two qualities of bodhisattvas are qualities such as compassion for all beings, generosity, firm intention, and non-desire for the Hīnayāna.

The next part (§29-51) of the sūtra explains the superiority of bodhisattvas by using a series of similes. The purpose of the use of similes is to “make known” (Skt. vijñāpayet) the good qualities of bodhisattvas.10 The typical formula of this part is: “It is like…, in that way…”

(Skt. tadyathā… evameva…). The Buddha compares bodhisattvas with four elements (§29-32).

Their mind is well tamed like tamed elephants (§37). They are untainted like lotuses are not stained by water (§38). They are beneficial for other sentient beings (§44, §47). This series of similes affects the imagination of the audience listening to the sūtra. It helps the audience to

10 upamopanyāsanirdeśās te kāśyapa nirdekṣyāmi yair upamopanyāsanirdeśebhiḥ bodhisatvo mahāsatvaguṇān vijñāpayet. (von Stäel-Holstein 1926, §29) 5

better comprehend the concepts that the Buddha is explaining.11 After the long list of similes, the sūtra starts talking about the philosophical concepts of early Buddhism and the innovative ideas of the Mahāyāna. It mentions the (Skt. madhyamā pratipad) (§52-59), an idea which already existed in the early Buddhist texts. However, it is now explained in Mahāyāna perspective. For instance, “The idea of permanence, Kāśyapa, is one end, the idea of , Kāśyapa, is a second end. The middle between these two permanence and impermanence (…) this is called, Kāśyapa, the Middle Way” (§56).12 Beside this, it mentions

Dependent Origination (Skt. pratītyasamutpāda) which is also a core concept in early Buddhist texts (§61-63). Yet, the mention of this well-known idea leads to the concept of emptiness (Skt.

śūnyatā), which is a concept that is emphasized in Mahāyāna Buddhism (§64). In this part we can see the rhetorical move applied by the authors of the KP. It starts with recognizable Buddhist concepts that can be found in early Buddhist texts, but we need to understand them through a

Mahāyāna lens.

The next part consists of extended similes that explain the complex concepts found in the earlier part by using metaphorical language such as the concept of emptiness (§64-66) and the concept of karma and defilement (§71). It also mentions the inferiority of the Śrāvakas whose

“roots of good are little” (§77) and who are not the “real sons of the Tathāgata” (§80).

Sometimes, the object of comparison is technical and requires specific knowledge of Indian myths, such as the Karaviṅka bird whose sound is sweeter than other birds. This bird is

11 Marino 2017, 8-11 argues that the metaphorical language such as an extended simile is used to “capture and engage the imagination of the audience in order to convey otherwise abstruse concepts in more readily comprehensible terms.” He also contends that the use of similes is a pedagogical strategy that is found everywhere in the Pāli Canon. 12 nityam iti kāśyapa ayam eko 'ntaḥ anityam iti kāśyapa ayaṃ dvitīyo 'ntaḥ yad etayor dvayo nityānityayor madhyaṃ (…) iyam ucyate kāśyapa madhyamā pratipad (von Stäel-Holstein 1926, §56). 6

compared to a bodhisattva who generates the initial aspiration to be a Buddha; his aspiration already surpasses that of all śrāvakas (§84).

Next, the sūtra uses the numerical lists like the first part of the text. However, in this part each list has only two points of teaching, for example: the two empty spaces of renunciants

(§112); the two impurities of renunciants (§114); and the two incurable illnesses of renunciants

(§119). This part is not intended to teach only the bodhisattva audience but admonishes all renunciants (Skt. pravrajita) in general. Then, the Buddha discusses in verse some general

Buddhist teachings, including non-self, the correct path, the purification of mind, and precepts, to the assembly (§137). Some monks in the assembly become liberated from the depravities while five hundred monks who possess the meditative absorption cannot understand these basic

Buddhist teaching and leave the assembly to reside in the forest (§138).

After the five hundred monks leave the assembly because of their ignorance, the Buddha asks Subhūti to explain the to them and bring them back to him but Subhūti rejects this proposal, saying that since the monks do not listen to the Buddha, they would not listen to him.

The Buddha then uses his supernatural power to create two magically manifested monks to meet with the five hundred deserters on their way to the forest. After conversing with these two conjured monks about nirvāṇa (Skt. parinirvāsyati) (§142-144), the five hundred monks become liberated from the depravities and come back to the Buddha. Subhūti asks them what they were taught by the two magically manifested monks. They explain the teaching in detail. Finally, other people in the assembly achieve the dharmic eye (Skt. dharmacakṣu) (§149). Lokakṣema’s version ends here, while a story of bodhisattva Samantāloka was added in the Qin, Song,

Sanskrit and Tibetan versions. His presence in the last part of the sūtra is obviously an additional

7

part which was inserted later. Its content bears the characteristic of later Mahāyāna sūtras such as the cult of book and the one will receive after reciting only one verse of this sūtra (§159).

This outline of the KP shows the flow of the narrative and the authors’ use of different rhetorical styles. The plot is not complicated but the method of persuasion is clear. When the text mentions complex innovative concepts of Mahāyāna, it is followed by similes in order to make these ideas comprehensible for their audience. In addition, the authors of the KP managed the rhetorical moves in each part of the composition well. Starting from general practices of bodhisattvas, followed by metaphors, they then mentioned the concept of emptiness by referring to general concepts in early Buddhist text first. Finally, they concluded that everyone could achieve their goal if they opened their mind and used the right lens to listen to the Buddha’s teachings. The right lens is the lens of Mahāyāna.

3. Historical Background

3.1 Date

Dating Buddhist scriptures is usually a complicated and uncertain task. First of all, neither Mainstream nor Mahāyāna scriptures record their date of composition. The notion of dating itself might not have been significant in Indian history. Second, even texts that were clearly composed later, such as Mahāyāna , still claim that they were spoken by the historical Buddha because they require his authority as the speaker of the sūtras. Therefore, finding a precise date of composition of Buddhist scriptures is a complex, if not impossible, task.

Scholars usually use two methods to date Buddhist scriptures. First, they look at the earliest manuscripts of a text and examine linguistic characteristics, style of writing, paleographic features, and even carbon-date the folios to find the approximate date of the 8

manuscript. However, the problem of this method is that sometimes we do not have early extant manuscripts, even though the date of composition of the text should far precedes the date of manuscripts that we have, such as the case of the Pāli Canon for which extant manuscripts are quite late. The second method is to rely on the Chinese translations. With this method, we can roughly date when that text was first translated. Nonetheless, both methods can only provide the terminus ante quem of the text. They show only that that text existed by the time the manuscripts were written or translated into Chinese. The date of composition, however, can be years or centuries earlier than the date of transcription or translation.

In the case of the KP, the Sanskrit manuscripts are much later than the Chinese translation of Lokakṣema. The complete manuscripts in St. Petersburg’s collection are dated between the 7th and 8th centuries CE.13 The earlier, but incomplete Sanskrit fragments in the

Hoernle Collection and the Mannerheim Collection, are dated in between the 3rd and 5th century

CE, roughly coinciding with the date of the Jin and the Qin translations.14 Therefore, our terminus ante quem is the late 2nd century CE, which is the date when the KP was translated into

Chinese for the first time by Lokakṣema. This translation of the KP is one of the earliest translations of Mahāyāna scriptures. According to Sengyou’s biography of Lokakṣema,

Lokakṣema came to Luoyang at the end of the reign of Emperor Huang (147-168 CE). However,

13 The date of this manuscript is different between von Staël-Holstein 1926, vii and Vorobyova-Desyatovskaya 2002, vii. While the former dates this manuscript in between 9th-10th century CE, the latter thinks it should be dated in between 7th-8th century CE. One thing we can assure is that this Sanskrit manuscript was circulated before or about the time of the last Chinese translation of the Song dynasty (985-986 CE) because this Chinese translation seems to relate to the Sanskrit version. 14 de Jong 1977, 247 references Kuno’s study in 1938 that these fragments correspond more closely to the Jin and the Qin versions than other versions, therefore, “[h]e concludes that this text must have been in existence in the 3rd- 5th centuries A.D.” 9

he did not start his career until the reign of the Emperor Ling. His translation, then, took place between 178-189 CE.15

However, Pāsādika dates this text back to the 1st century BC. He compares this text to the

Aṣṭasāhasrikā Prajñāpāramitā which was translated into Chinese in the same period. He argues that the language in the prose parts of the KP seems more archaic than the prose parts in the

Aṣṭasāhasrikā Prajñāpāramitā.16 This date, however, cannot be certain. First of all, the language can be archaized in later period as it will be shown in the case of the KP. Secondly, some parts of the KP which discusses emptiness might have been borrowed from the Aṣṭasāhasrikā

Prajñāpāramitā. This concept of emptiness comes with Subhūti who can be regarded as a representative of this concept. This might suggest that when the KP was composed or compiled, the Aṣṭasāhasrikā Prajñāpāramitā had already been popularized or known among Mahāyāna communities. The idea that the KP was composed before or even in the same period as the

Aṣṭasāhasrikā Prajñāpāramitā is not convincing. Yet, we cannot ascertain how long after the

Aṣṭasāhasrikā Prajñāpāramitā the KP was composed. Unless we find older manuscripts in the future, the oldest evidence of the existence of the KP is still the date of Lokakṣema’s translation.

3.2 Name of the text

The name “Kāśyapaparivarta” is generally used by modern scholars to refer to this text.

It comes from the Tibetan translation in the Kanjur collections where this text is called “Ārya-

Kāśyapaparivarta-nāma-mahāyāna-sūtra.” However, this is not the earliest or the only name known for this text. In Indian sources such as the Śikṣāsamuccaya, when quoting passages from

15 Nattier 2008, 73. 16 Pāsādika 1980, 49-50. 10

this text it is always referred to as Ratnakūṭa, which later became the name of the collection of

Mahāyāna texts existing only the Chinese and the Tibetan canons. The name Kāśyapaparivarta did not appear in any early Sanskrit works.

The earliest name we can trace is in Lokakṣema’s translation. The colophon of the text identifies it as Weiri (yue) monibao jing 遺⽇(⽈)摩尼寶經. Karashima suspects that the character ri ⽇ in Lokakṣema’s translation might be a false transliteration of yue ⽈ which is the transliteration of *vulla of its underlying language.17 This word is equal to [Vai]pulya in

Sanskrit which is a category of Buddhist teachings.18 From the reconstruction of the name of this text, it might have been called *Vaipulya-maṇiratna-sūtra in its early stage.19

However, in the Jin version, this text is called *Mahāyāna-Ratnakūṭa-sūtra ( baoyan jing 摩訶衍寶嚴經). The word “Vaipulya” was shifted to “Mahāyāna” and the name of the text became “Ratnakūṭa” as it was known to most of Indian sources.20 In the Qin version, the text is called *Samantāloka-bodhisatva-parivarta (Puming pusa jing 普明菩薩會). Bodhisattva

Samantāloka appears at the last scene of the sūtra. This part does not exist in the Han version which confirms that this might be a new name of the sūtra used only in this version.

17 Karashima 2015, 118-119. See also Nattier 2008, 76. 18 Vaipulya is one of nine (or twelve genres) in early Buddhist texts. It is known in the Pāli Canon as Vedalla which are composed of sutta, geyya, veyyākaraṇa, gāthā, , , jātaka, abbhutadhamma, and vedalla. It is a type of scripture in form of questions and answer either between the Buddha and his disciples or between disciples themselves. However, in Mahāyāna scriptures, sometimes it can be used as Vaipulya/Vepulla, Vaitulya/Vetulla, or Vaipulya/Vepulla. Skilling 2013, 69-162 suggests that the term “Vaipulya” or “Vepulla” derives from the earlier form “Vaidalya/Vedalla” and “Vetulya/Vetulla”, the terms that are confused in some Mahāyāna scriptures. Vaipulya and Vaitulya are used to be employed only in certain contexts. Later, Vaipulya won over other terms and became the most familiar term. The other two terms were gradually deleted from Mahāyāna corpus. The Lokakṣema’s translation shows that in his time, the KP was still called Vaipulya like many other Mahāyāna scriptures. 19 Although the name in Lokakṣema’s translation should be reconstructed as maniratna, I suspect that it might actually be a loose translation of ratnakūṭa instead of maniratna which Karashima reconstructed. However, there is no clear evidence to prove it. Therefore, I follow Karashima’s reconstruction here. 20 Skilling 2013, 90 explains that Mahāyāna often call themselves as Vaidalya/Vaitulya/Vaipulya which became ones of many epithets of Mahāyāna sūtras. The name such as “Mahāyāna Vaidalya Sūtra” is widely added at the end of Mahāyāna sūtras to identify that this sūtra belongs to Mahāyāna. 11

In the old Tibetan version, the text is called Ratnakūṭa like other Indian sources as well as the Jin version, while in the new Tibetan translation in the Kanjur collections, it is called

ĀryaKāśyapaparivarta-nāma-mahāyāna-sūtra.21 However, we cannot assume that the name

“Kāśyapaparivarta” was invented in the Chinese and Tibetan traditions only when it was included in the Mahāratnakūṭa collection because this collection only exists in the Chinese and the Tibetan canons. Apple has shown recently that the first Indian text that refers to the KP as

Ārya-Kāśyapaparivarta instead of Ratnakūṭa is the Ratnakaraṇḍodghāṭamadhyamakopadeśa written by Dīpaṃkaraśrījñāna in the year 1037 CE. He then concludes that “We can infer, therefore, that a shift of the title from Ratnakūṭa to Kāśyapaparivarta occurred between the 8th to

10th centuries in Indian sources.”22 And the shift of the name might not be related to the compilation of the Mahāratnakūṭa collection in China but it already happened in India.

Therefore, we have here three stages of the development of the name of this text. Firstly, in the Lokakṣema’s translation, it was called *Vaipulya-maṇiratna-sūtra. Secondly, the word vaipulya was changed later to mahāyāna which is another epithet of vaipulya. At this time, the sūtra was called *Ratnakūṭa, which was the common name among most of Indian sources such as the Śikṣāsamuccaya as well as in the old Tibetan version. Finally, the sūtra was called

“Kāśyapaparivarta” in later Sanskrit texts, as well as in the new Tibetan translation in the Kanjur collections.

3.3 Kāśyapaparivarta in the Mahāratnakūṭa Collection

21 Apple 2017, 206. 22 Apple 2017, 208. 12

The KP is one of forty-nine texts in the Mahāratnakūṭa collection, including some important texts such as Sukhāvativyūha, Rāṣṭrapālaparipṛcchā, Ugraparipṛcchā,

Śrīmālādevīsiṃhanādasūtra, and Bodhisattvapiṭaka. The name Ratnakūṭa was used only to refer to the KP earlier in Indian anthologies such as the Śikṣāsamuccaya and the Sūtrasamuccaya. The

Mahāratnakūṭa as a collection of scriptures only exists in the Chinese Tripiṭaka and the Tibetan

Kanjurs. The Mahāratnakūṭa first appeared as a collection of texts in China in Tang Dynasty by the South Indian monk , who brought a collection of Sanskrit manuscripts from India to China in 693 CE. He was requested by Emperor Zhongzong of Tang to translate the

Mahāratnakūṭa into Chinese language. Bodhiruci completed this translation in 713 CE. In his edition, he revised twenty-three texts, retranslated fifteen inadequately translated texts, and translated eleven new texts that had never been translated before.23

Little is known about the reason for the compilation of the Mahāratnakūṭa collection.

According to Pagel, most of texts in this collection, like most of Mahāyāna texts, share a common theme on bodhisattva path.24 Therefore, it is not certain that this would be the initial factor of the compilation. Pagel also observes eight general ideas in the Mahāratnakūṭa collection: the training of lay-bodhisattva path; the training of monastic bodhisattvas; the training on the difference between ordinary or worldly morality; the discourses on the jātaka tales; the amplifications of miracle and magic elements; the myth of buddhas and their buddha-fields; the veneration of caitya and vision of all buddhas; and the concept of female bodhisattvas.25 These general themes show that the Mahāratnakūṭa collection contains texts with an emphasis on the

23 Pedersen 1980, 60. See also Forte 2002. 24 Pagel 1992, 44. 25 Pagel 1992, 46-7. 13

full range of Mahāyāna concepts. And it does not bear any special characteristic which can be identified as a significant thread of unification.

The KP is directly related to the Mahāratnakūṭa collection. The name of the collection used to be the name of the text itself before it was shifted to Kāśyapaparivarta in later period.

The KP was incorporated into the Mahāratnakūṭa collection only in China and Tibet. In India, it was circulated as an independent text as we have some extant translations in Chinese language.

The Sanskrit manuscript in St. Petersburg and the Song translation are not attached to the

Mahāratnakūṭa collection even though they are dated to after the compilation of the

Mahāratnakūṭa collection. Only the Tibetan translation in the Kanjur collections is incorporated in this collection. It is likely that the archetype of the Tibetan translation might have been an independent text as well, but it was integrated into the collection when it was translated into the

Tibetan following the Chinese tradition.

4. The Textual Development of the Kāśyapaparivarta

4.1 Versions of the text

The KP exists in many different versions in the Sanskrit, Chinese, and Tibetan translations. Although the versions could be divided into three groups based on language, I find it more suitable to categorize the versions into three groups by using their structure as this also suggests their stage of development. The first group comprises the versions that have no verse structure and is the older group while the second group includes the later versions that do have verses. The last group is made up of quotations in other Mahāyāna scriptures.

14

4.1.1 Versions without verses

T 350 Weiri(yue) monibao jing 遺⽇(⽈)摩尼寶經 (The Han Version)

The Weiri(yue) monibao jing 遺⽇(⽈)摩尼寶經 (*Vaipulya-maṇiratna-sūtra) was translated by Lokakṣema (Zhi Loujiachen ⽀婁迦讖) in the 2nd century CE.26 His translation style is slightly different from , a pioneer translator who came to Luoyang roughly two decades before him. He coined new terms that had never been used in An Shigao’s translations. He usually chose to transcribe Indian technical terms into Chinese words instead of translating them. There are also multiple transcriptions in his translation of the KP. He usually omitted the opening phrase “,” which is considered to be one of the most important parts of

Buddhist scriptures because it authorizes the status of the sūtra as the word of the Buddha.27

Also, he rendered all verses into Chinese prose even though the Indic archetype might have been composed in verse.28 This is can be seen in section 136 of the KP. All other versions of the KP have this paragraph in verse while Lokakṣema rendered it in prose. Presumably, this paragraph might have been originally composed in verse, but Lokakṣema chose to render it in Chinese prose, which is his own style of translation. His translation has 127 sections and it shows the first

26 According to Nattier 2008, 75, Lokakṣema was a monk from Yuezhi or Kushan area in the Northwest of India. He arrived in Luoyang during the reign of the Emperor Huan (147-168 CE) but he only started translating Buddhist scriptures from 178-189 CE in the reign of Emperor Ling. He translated many Mahāyāna scriptures such as Aṣṭasāhasrikā Prajñāpāramitā and Śūraṃgamasamādhi-sūtra which are two of the most important Mahāyāna scriptures. It cannot be assumed, however, that the Mahāyāna was the most popular in the Kushan area where he came from. Nattier suggests that the reason he chose to translate Mahāyāna scriptures might have been “his own interest in Mahāyāna sūtras.” 27 Nattier 2008, 75. This must be considered as his personal style rather than evidence of his archetype not having it since most of his translations do not start with the typical nidāna phrase. In Nattier’s note 173, she observes that Lokakṣema omitted the phrase “Thus have I heard” in 6 translations: Aṣṭasāhasrikā Prajñāpāramitā (T 224), Pratyutpannabuddha Saṃmukhāvasthita Samādhi Sūtra (T 418), Lokānuvartanāsutra (T 807), The Sūtra of mañjuśrī’s Questions Concerning the Bodhisattva Career (T 458), Kāśyapaparivarta (T 350), and the Sūtra concerning the Perfect Enlightenment of Amitābha (T 362). 28 Nattier 2008, 76 notes that “Lokakṣema renders Indian verse passage into Chinese prose, doing so even in the case of the Neizang bai bao jing 內臟百寶經 (T 807), whose Indic language sources-text was probably entirely inverses.” 15

phase of the development of this text. Even though the overall content is not different from other versions, the length of his translation is much shorter.

T 351 Moheyan baoyan jing 摩訶衍寶嚴經 (The Jin Version)

The Moheyan baoyan jing (*Mahāyāna-ratnakūṭa-sūtra) was translated in the Jin dynasty (265-420 CE). The translator was anonymous. It has 143 sections in total, all of which are in prose except section 136 which is in verse. This solitary section in verse was not inserted later like the summary verses in the later versions but was originally written in verse. The content of this version is close to the Qin version and could have been translated from related archetypes.

T 310 (section 43) Puming pusa hui 普明菩薩會 (The Qin version)

The Puming pusa hui (*Samantāloka-bodhisattva-parivarta) was translated in the Qin dynasty (351-431 CE). The translator is also anonymous. It has 144 sections—only one section more than the Jin version. The structure and content of this version are similar to the Jin version.

However, the last part of the text is different and may be a piece of evidence for us to be able to trace to its relationship to other versions. This version was collected as one of the forty-nine sūtras of the Mahāratnakūṭa collection.

T 659 (chapter 7) Dasheng baoyun jing ⼤乘寶雲經 (The Ratnamegha version)

The Dasheng baoyun jing (*Mahāyāna-Ratnamegha-sūtra) was translated by

*Maṇḍalasena (Mantuoluoxian 曼陀羅仙) and *Saṅghabhara (Sengjiapoluo 僧伽婆羅) in the 6th

16

century.29 The KP was incorporated in the Ratnamegha sūtra as the seventh chapter. Like other versions which were translated before the 9th century CE, this version does not have verses except for only paragraph 136 which might have been originally written in verse. Its content is quite similar to that of the Jin and the Qin versions, but the language of translation is much closer to the Qin version. It might actually be a revision of the Qin version.

The Old Tibetan Version from Dunhuang

The old Tibetan manuscripts of the KP were found in Dunhuang. They belong to the

Stein and Pelliot collections. The title of the old Tibetan version is Ārya-Ratnakūṭa-nāma- mahāyāna-sūtra which is different from the new version in the Kanjur collections. It also lacks the uniformity of vocabulary used for translation which suggests that it was translated before the imperial codification of the Mahāvyupatti in 814 CE. This old Tibetan version does not have verses, which is similar to the earlier recensions of this text. The seven fragments contain sections 0-63 and 98-135 of Staël-Holstein’s edition.30

The R. Hoernle Collection and the Mannerheim Collection

The two fragments of the KP in the R. Hoernle collection were first identified by the

Japanese scholar Horyu Kuno. In his article “Sanskrit Fragments of the Kāśyapaparivarta,” de

Jong finds that these fragments belong to the same manuscript as the one in the Mannerheim

Collection edited by J.N. Reuter. These fragments, three in total, correspond to sections 128-136 of the Staël-Holstein edition. This Sanskrit version of the KP is closer to the Chinese translations

29 Silk 2010, 900. 30 See Apple 2017, 205-230 for more information and a transcription of the old Tibetan version. 17

of the Qin and the Jin versions. It, too, is also not written in verse summary. Therefore, this version of the text can be dated to around the 3rd-5th centuries CE.31

The Indikatusaya Copper Plaques

The Indikatusaya copper plaques were discovered in a stūpa in . They are dated to between the 8th and 9th centuries CE. The version represented in these plaques is also not written in verse. According to Paranavitana, the scribe of these plaques was not familiar with the

Sanskrit language and he therefore made a few mistakes in his inscriptions. All of the plaques only preserve short passages of thirteen syllables or fewer. These passages correspond to sections

40, 63, 65, 71, 79-80, 88, 105, 106, 134, 138, and 161 in Staël-Holstein’s edition.32

4.1.2 Versions with verses

The Sanskrit version of the St. Petersburg’s manuscript (the Sanskrit version)

The Sanskrit version of the St. Petersburg’s manuscript of the KP was found in the

Khotan area. It can be dated around the 7th-8th centuries CE. The Sanskrit manuscript of the KP is located in the collection of the St. Petersburg Branch of the Institute of Oriental Studies. The manuscript is almost complete; only 5 sections out of 166 sections are missing. It was first studied by Alexander von Staël-Holstein, who published an edition in 1926 that included a transcription of the Sanskrit manuscript with its parallels in Chinese and Tibetan. In the year

31 de Jong 1977, 249-55. 32 Paranavitana 1933, 199-212. 18

2002, Vorobyova-Desyatovskaya in collaboration with Karashima and Kudo published another edition of the same manuscript with photographs of all 74 plates.33

T. 352 Dajiaye wen dabaoji zhengfa jing ⼤迦葉問⼤寶積正法經 (the Song version)

The Dajiaye wen dabaoji zhengfa jing (*Mahākāśyapa-paripṛcchā-Mahāratnakūṭa- saddharma-sūtra) was translated by Dānapāla (shihu 施護), who was active during the Song dynasty, roughly in the year 986.34 His translation has 161 sections. The missing four paragraphs are due to a defect in the original Sanskrit manuscript that was used by Dānapāla.35 The content corresponds to the Sanskrit and Tibetan versions, which represent the last stage of the development of this text.

The Tibetan translation in the Kanjur collections (the Tibetan version)

This version was translated by Jinamitra, Śīlendrabodhi, and Ye-shes-sde in the 9th century CE. This version is a revision of an earlier version with new translation vocabulary. The content usually follows the Sanskrit version which can be considered as the last phase of the development of the KP.

4.1.3 Quotations in other works

As I have already stated, the KP is a significant sūtra of Mahāyāna Buddhism. Not only was it preserved in the many aforementioned versions, but it also quoted in many other

33 Vorobyova-Desyatovskaya 2002, vii-xi. 34 Harrison 1993, 164. 35 Staël-Holstein 1926, x. 19

Mahāyāna scriptures. The KP is quoted eight times in the Śikṣāsamuccaya of Śāntideva (sections

3, 4, 5, 6, 11, 15, 24, 25, 97-102, 128). It is also found in chapter 24 of the

Mahāyānasūtrālaṃkāra written by Asaṅga. Sections 11, 97, 98, 99, and 102 are quoted in

Prajñākaramati’s commentary of the Bodhicaryāvatāra (Bodhicaryāvatārapañjika).36 It is interesting that the quotations in these works are not always identical to the extant Sanskrit version. First of all, both the Śikṣāsamuccaya and the Bodhicaryāvatārapañjika do not quote the summary verses which were probably added in later version.37 Secondly, there are some difference in word choices. For example, in section 98 while the extant Sanskrit version uses pradīpārcciḥsadṛśaṃ, the Śikṣāsamuccaya and the Bodhicaryāvatārapañjika use dīpārcciḥsadṛśaṃ; and in section 102 when the extant Sanskrit version uses tannātītaṃ, the

Śikṣāsamuccaya and the Bodhicaryāvatārapañjika use tannaivātītaṃ. In general, the quotations in the Bodhicaryāvatārapañjika usually agree with those in the Śikṣāsamuccaya except in section

11 where the Bodhicaryāvatārapañjika has saṃprakāśanā while both the Śikṣāsamuccaya and the extant Sanskrit version has only prakāśanā. Moreover, all quotations of the KP in the

Bodhicaryāvatārapañjika also appear in the Śikṣāsamuccaya. This is very likely that

Prañākaramati quoted these passages from the Śikṣāsamuccaya, and not from a different recension of the KP.

De Jong also listed works that quoted sections from the KP by referring to research done by Japanese scholars who had identified more quotations in other Sanskrit works: the

Prasannapadā, the Madhyāntavibhāgaṭīkā, the Laṅkāvatārasūtra, and the Ratnagotravibhāga.

Besides these works, there are some other works that are preserved only in Chinese and Tibetan:

36 von Staël-Holstein 1926: XVI. 37 The addition of summary verses will be discussed in 4.2 20

the Mahāyānasaṃgraha, the Daśabhūmivibhāṣa (T 1522), the Prajñāpāramitopadeśa (T 1509), the Mahāyānāvatāra (T 1634), the Anuttarāśrayasūtra (T 669), the Mahāparinirvāṇasūtra (T

374-377), Foxing lun 佛性論 (T 1610), and Ji zhufabao zuishangyi lun 集諸法寶最上義論 (T 1638).38

Moreover, Pāsādika also found more quotations in the Saṃdhinirmocanasūtra, the

Vimalakīrtinirdeśa, and the Sūtrasamuccaya.39 This increases the total number of works with quotations from the KP to almost twenty texts.

Not only are quotations from the KP found in Sanskrit works and Chinese and Tibetan translations which were originally rendered from Sanskrit manuscripts, but the KP is also quoted in other works that were originally written in Central Asian languages. An old Khotanese translation contains a portion of the list of thirty-two qualities of a bodhisattva. This old

Khotanese version seems to be a translation of the prose part of the Sanskrit version of the KP which is in sections 23-25. The Khotanese translation does not follow the Sanskrit word by word but seems to use a fair amount of interpretative translations. Many technical terms such as upāyakauśalya and ārūpyadhātu are rendered in Khotanese words rather than using loanwords from Sanskrit.40

Besides the Khotanese translation, the KP is also quoted in the Book of Zambasta twice.

The first passage is found in the Book of Zambasta chapter 8 verses 38-39. They are the quotations from the KP sections 98-99, which deal with the nature of the mind.41 The second passage is found in the Book of Zambasta chapter 8 verses 40-46. It is quoted from the KP section 69. Only verse 44 is a direct quotation of the KP section 69, whereas the rest seems to be

38 de Jong 1977, 254. 39 Pāsādika 1980, 49. 40 Maggi 2015, 101-43. 41 Martini 2011, 139-45. 21

an abridgement and summary of the overall content from sections 64-71 in the KP.42 This quotation is particularly about the Middle Path. Besides these quotations, there are also two similes used in the Book of Zambasta that are similar to the KP. Even though these similes could possibly have been drawn from other sources, Martini believes that they are more likely to have come from the KP because this text was circulated very early in Central Asia.43

4.2 Three stages of textual development

Most of Mahāyāna sūtras, if not all, were not composed at once but were compiled and inserted with more elements from the beginning of their composition until the versions that were preserved in manuscripts and in the Chinese and Tibetan translations which come down to us.

Conze also states that Mahāyāna scriptures are “a collective work which has been subjected to additions and alterations in the course of centuries.”44 The task to identify the earliest version of any Mahāyāna sūtra, which represents the first stage of development of the movement, seems to be, at least with available evidence, impossible. Nattier has written that,

Any study of early Mahāyāna Buddhism is subject to one overarching constraint: the absence of any written sources that could document directly the nascent phase of these new ideas and practices. To put it another way, the initial stages of the development of ideas about the practice of the bodhisattva path took place off- camera, and only after the basic ideas associated with this practice had undergone considerable development were the earliest texts that we now refer to as Mahāyāna sūtras composed. Rather than showing us the incipient phase of Mahāyāna thinking, these scriptures already represent a somewhat later phase of

42 Martini 2011, 146-8. 43 Martini 2011, 154. 44 Conze 1952, 251. This claim is widely accepted by Mahāyāna scholars. Skilling 2003, 103 has also written that “The texts of the Bodhisattvayāna/Mahāyāna are not uniformly ‘late’. Each corpus, each body of texts, has many layers or strata, and as a whole embodies centuries of development.” However, there is no clear evidence to support this claim. We cannot ensure that there are no Mahāyāna texts at all that come down to us in their original form. 22

development, in which the viability of the bodhisattva path (at least for some members of the Buddhist community) is already taken for granted.45

Nattier’s statement points to some valuable observations. First, we do not have any record of the initial phase of the emergence of Mahāyāna Buddhism. Second, the earliest

Mahāyāna translation into the Chinese language which we consider to be the oldest version might not represent the earliest phase of the rise of the movement. Silk also affirms that “[I]t already represents a considerable degree of sophistication and development, rather than recording the first few rough steps toward an expression of a new and raw set of ideas.”46

Therefore, it is difficult to presume the character of the earliest Mahāyāna ideas. What we can do to understand the beginning phase of Mahāyāna literature is to trace significant threads of the early Mahāyāna literature from its textual nature that is illustrated by its descendants or many extant versions that are preserved in different languages.

There are common principles for evaluating these descendant versions of Mahāyāna sūtras in Chinese translations also suggested by Nattier:

First, when we have both earlier and later recensions of a given sūtra it is generally the case that the later text will be the longer one. There are exceptions of course, and many of these involve interpolations arguably added in China; but by and large growth rather than shrinkage seems to be the norm (…). Second, when relatively early Indic-language manuscripts are available for comparison, Chinese translators can sometimes be proven to be in agreement with theses earlier versions (and thus acquitted of the charge of having abbreviated their texts). The contents of the Schøyen collection of Buddhist manuscripts, which includes text ranging from the 2nd to 9th century in date and includes fragments of several Mahāyāna sūtras, promise to provide us with priceless new comparative data in this regard.47

45 Nattier 2000, 77. 46 Silk 2002, 374. 47 Nattier 2003, 61. 23

I agree with Nattier that later recensions of the same Mahāyāna text are generally longer than the earlier versions due to the process of expansion which is normal in Mahāyāna literary culture. By applying this evaluation and the timeline of translation to the case of the KP, the development of this text can be separated into three stages: the early stage, the middle stage, and the final stage.

The extant versions can only be evidence of the last two stages. The early stage of this text as well as most of Mahāyāna sūtras have to be considered from the linguistic evidence by using the method of literary criticism.

Early stage: the primitive text

As was noted above, Lokakṣema’s Chinese translation48 is not the original version of the sūtra. This version already shows the development of the text that could have happened decades or even centuries before his translation. Harrison suggests that Lokakṣema’s translations can only illustrate the early middle period of Mahāyāna composition.49 Most of the original composition, if not all, took place off-camera, which means that we do not have the evidence of its primary production. Jan Nattier proposes the process of sūtra production which she calls

“sūtrafication”, saying that some sūtras might primarily have been just teachings of some teachers or . Later on, these teachings were compiled, framed as sūtras, and became

48 When I mention any translation in this study, I refer to its archetype in the Sanskrit or Prakrit languages. For example, when I state that the Jin version is similar to the Qin version, I mean that the Sanskrit manuscript which the Jin version was translated from is similar to the manuscript of the Qin version. However, there is an exception in some cases that Chinese translators might have adapted the earlier Chinese translations to their own translations. This makes the two versions similar without the relationship of the underlying Indic versions. For example, in the case of the Qin and the Ratnamegha versions where, I think, the Ratnamegha version in Chinese might be a revision of the Qin translation. In this case it is difficult to decide whether the underlying Indic versions of these two translations are also related or only the Chinese translations are related. However, there are some more detail that points to the fact that the underlying Indic archetypes of these two versions are, indeed, related. The detail will be discussed later. 49 Harrison 1995, 56. 24

canonical texts. This process could have occurred in oral form.50 However, Cole argues that “this model seems plausible enough for some of the simpler Mahāyāna sūtras.”51 Given the fact that the KP is not a very long sūtra and it does not have a complex narrative plot like other long

Mahāyāna sūtras, it is possible that the KP could have had its origin in some Mahāyānas teachers’ oral teachings as suggested by Nattier, which were then compiled and extended later over the course of a millennium as I will show later. But in this oral stage, this text might not have been called a sūtra yet. In other words, it did not become a sūtra until it was compiled in a complete sūtra form which might have occurred in the written tradition. The underdeveloped nidāna in Lokakṣema’s translation that lacks an important phrase “Thus have I heard” might also support this claim.

Another interesting topic about the origins of Mahāyāna sūtra production is the language of the composition and transmission. We know that the early Buddhist texts were orally composed and transmitted in Prakrit languages. Until the 2nd century CE, some of them were gradually rendered into Sanskrit language through the process of Sanskritization. This gradual process is shown in the studies of Gāndhārī manuscripts in the North-west of India and Central

Asia.52 Some scholars suggest that the early Mahāyāna scriptures might have followed the same pattern of development. Karashima proposes that “Early Mahāyāna scriptures were originally in

Prakrit not in Sanskrit.”53 This claim is supported by many discoveries of Mahāyāna manuscripts written in Gāndhārī languages. One of the most important of these manuscripts is the manuscript of Aṣṭasāhasrikā Prajñāpāramitā. Not only does it confirm that some early Mahāyāna sūtras

50 Nattier 2003, 11-12, footnote 3. 51 Cole 2005, 17. 52 Salomon 2017, 239-267 suggests 4 stages of the evolution of written āgama collections. In stages two and three, there is evidence of the gradual process of Sanskritization in Gāndhārī manuscripts, especially in stage three. The later the manuscripts are, the more Sanskritized they are, too. 53 Karashima 2015, 113. 25

were composed and transmitted in Prakrit, but it also confirms that the underlying Indic version of Lokakṣema’s Chinese translation was indeed shorter than the later versions in Chinese,

Tibetan, or Sanskrit. These versions have been lengthened rather than Lokakṣema having abbreviated his archetypes. Falk and Karashima have found that,

Lokakṣema did not abbreviate a text similar to the Sanskrit version, but rather translated a version already slightly enlarged in comparison to our manuscript, so that our manuscript can be regarded as representing the forerunner to the one Lokakṣema knew.54

From this evidence, it is likely that Lokakṣema could have also translated the KP from

Gāndhārī manuscripts. In fact, Weller has studied the underlying Indic language of Lokakṣema’s translation. He indicates that “ihrer eine Anzahl weise auf das Prakrit hin, nicht auf das

Sanskrit.”55 I have also observed that this Prakrit language could be Gāndhārī rather than other

Prakrit. First of all, Lokakṣema came from the Yuezhi or Kushan area in the 2nd century CE where the Gāndhārī language was predominantly spoken. Secondly, the transliterations of proper names and technical terms show that this Prakrit language distinguishes between s, ś, and ṣ sounds,56 which is the distinct characteristic of Gāndhārī language. In his book A Glossary of

Lokakṣema’s Translation of Aṣṭasāhasrikā Prajñāpāramitā, Karashima also finds some traces of

Gāndhārī language in the Chinese transliteration of Lokakṣema’s translation. For example, dasa- ajie 怛薩阿竭 *tasa-agada < BHS. tathāgata, bunali 不那利 *puṇ(ṇ)ari < Skt.puṇḍarīka, and

54 Falk and Karashima 2012, 20. 55 Weller [1969] 1987, 1162. 56 For example, s in sanmei 三昧 [sam mwojH] *samā’i < Skt. samādhi, youpose 優婆塞 [‘juw ba sok] *upāsaka/upāsa’a < Skt. upāsaka; ś in shi 釋 [syek] *śak(ka) < Skt. śakra, ouhegousheluo 漚惒拘舍羅 [‘owH hwa kju syaeH la] *uvā’akośa(l)la < Skt. upāyakauśalya; and ṣ in jiasha 袈裟 [kae srae] *kāṣā’a > Skt. kāṣāya, shamen 沙⾨ [srae mwon] *ṣamaṇa < Skt. śramaṇa. See Weller [1969] 1987, 1162-1175 for more detail on the transliteration of proper names in the Han version. 26

jiuyuan 鳩垣 *kuvaṇ(ṇ)a < BHS. kumbhāṇḍa(ka). The phonetic alteration of these three words are attested in Gāndhārī manuscripts such as the Dharmapada and Niya documents.57

If my observation is correct and Lokakṣema translated the KP from the Gāndhārī language, it is then plausible that the early stage of the development of the KP is no different from the other early Buddhist texts. It might have been some teachings of Mahāyāna teachers which were compiled later as a complete and formal sūtra. The early versions might have been composed and compiled in the Prakrit language before it was sanskritized in later period. This model proposed by Nattier might work with the Ugrapariṛcchā as well as other early Mahāyāna sūtras such as the KP.

Middle stage: The variants

Most of the extant versions of the KP can only represent the middle stage of the development starting from Lokakṣema’s translation in the Han dynasty, followed by the Jin, the

Qin, and the Ratnamegha versions. Lokakṣema’s version shows the early middle stage of the textual development with its simple and short style. However, the content of this version is only slightly different from other later versions. The compilation of the sūtra had happened some time before it was translated into Chinese, yet, we cannot assume that this version was the only version that circulated in the 2nd century CE, nor can we assume that it was the direct ancestor of later versions since many textual differences are found in later versions.

The question about recensions and archetypes of Mahāyāna sūtras is difficult if not impossible to answer. The common issue that all comparative studies of Buddhist texts with different recensions share is that no version can be directly linked to others. In other words, not

57 Karashima 2010, 760. 27

all recensions share the same ancestors even though they are the same text or variants of a text that were circulated under the same name. Zimmermann made an observation in his comparative study of various versions of the Tathāgatagarbhasūtra, saying that “Besides the obvious interpolations, the relation between the two recensions is not an ancestral one. Neither of the two can be said to have grown out of the other recension. In many instances the recensions are simply different in content – sometimes so different that it is hard to imagine that a common archetype had ever existed.”58 This fact seems to be true for the case of the KP as well as many other

Mahāyāna sūtras.

The KP can be considered to be more complicated than the Tathāgatagarbhasūtra. The more recensions there are, the more complex the relationships between recensions. We are here a little fortunate in that we have a clear boundary between two versions of the KP. We know that the versions written in verse are the later versions, comprised of the Song version, the Sanskrit version, and the Tibetan version. The other versions without verses such as Lokakṣema’s version, the Jin version, the Qin version, and the Ratnamegha version are earlier versions.

Regardless of additional summary verses in each section and some differences of wording, these two groups of the text always agree with each other in terms of their overall content.

Even though an examination of Lokakṣema’s translation in detail indicates that it preserves the early characteristics of Mahāyāna sūtras, we cannot consider Lokakṣema’s version as a direct ancestor of any later versions since the differences are too great to allow this assumption. Weller has suggested that the archetypes of extant versions of the KP might have been in two lines, “Dabei spaltete sich die Überlieferung des Werkes in zwei große Stränge auf.

Derer einer wird durch die Han-Übersetzung dargestellt, der andere durch die restlichen

58 Zimmermann 1999, 146. 28

Fasungen des Werkes.”59 It is not necessarily the case that the Jin and the Qin versions are both direct descendants of the Indic text that Lokakṣema had in front of him. It is plausible that there may be three or more separate strands of textual tradition rather than one which later split into two or more. The later versions might have come down from completely different archetypes: each one from its own archetype. Every version has its own distinct features that are different from the others.

The last part of the sūtra (§150-165) which was added later can be useful when considering the relationship between each recension. The Han version which is the oldest version does not have this part. It ends with section 149 and then skips sections 150-165, jumping to the ending section (§166), which could be the format of the primitive version. It is not likely that

Lokakṣema had deleted this part (§150-165) of the sūtra because the content of the sūtra is complete in his version. Section 149 in the Han version says that,

At that time, Subhūti questioned on the affaires which the five hundred monks could reply. At that time, one million and two hundred thousand human beings, and nāgas all attained the Stream-Enterer. One thousand and three hundred monks all attained Arhatship.60

Then, this is followed by the last section (§166) which is the ending section of other versions as well.

When the Buddha spoke the sūtra, monks, nuns, laymen, laywomen, gods, humans, nāgas, all rejoiced. They paid respect in front of the Buddha and left.61

59 Weller [1969] 1987, 1146. 60 All translations in this study, otherwise stated, are my translation. T 350, 12: 194a14-a16: 時須菩提問事以所可報五百人。爾時百二十萬人。及諸天鬼神龍皆得須陀洹道。千三百比丘皆得阿羅漢道. 61 T 350, 12: 194a16-1: 佛説經已。比丘比丘尼優婆塞優婆夷。諸天世人鬼。神龍皆歡喜。前爲佛作禮而去 29

From this format we can see that the end of the sūtra is complete. It is more likely that the later versions added more content in between sections 150-165. Because this part was added later, it is different in all versions, and can be a source to trace their relationship. Even though the Jin and the Qin versions have something in common—they were translated in almost the same period, they do not have verses, and their content is quite similar—the differences at the end of these two versions suggest that they might have come from different archetypes. The Jin version did not have sections 150-156, while Qin version did not have sections 157-165. At some point, someone might have decided to combine the two different versions, and this became the archetype of the Sanskrit version since it has both parts from the Jin and the Qin versions (150-

156 from the Jin and 157-165 from the Qin).

Table 1: The comparison of sections 150-166 in each version

section Jin Qin Ratnamegha Sanskrit/ Notes Song/Tibetan 150 x x x 151 x x x 152 x x x 153 x x x 154 x x x 155 x x x 156 x x x 157 x x x 158 x 159 x x x The Ratnamegha version has some additional phrases that are unlike the Jin version. 160 x x 161 x x 162 x x 163 x x 164 x This section is summary verses of prose sections 161-163. 165 x

30

166 x x x

The problem is the Indic archetype of the Ratnamegha version. This translation has something in common with the Qin version. First, while other Chinese translations use numbers

1, 2, 3, and 4 (⼀, ⼆, 三, 四)62 to explain each aspect of teachings in the first part (sections 1-23), the Ratnamegha and the Qin versions do not have these numbers. Second, the Ratnamegha translation maintains the translation style of the Qin version. It seems that the translators of the

Ratnamegha had access to the Qin translation and translated their version by consulting with the

Qin version. For example,

The Qin version:

復次迦葉。菩薩有四法。所⽣善法増⻑不失。何謂爲四。捨離邪法求正經典六波羅蜜 菩薩法藏。⼼無憍慢於諸衆⽣謙卑下下。如法得施知量知⾜。離諸邪命安住聖種。不 出他⼈罪過虚實不 求⼈短。若於諸法⼼不通達作如是念。佛法無量隨衆所樂⽽爲演 説。唯佛所知非我所解。以佛爲證不⽣違逆。迦葉。是爲菩薩四法所⽣善法増⻑不 失。63

Again, Kāśyapa, a bodhisattva has these four qualities by which the wholesome qualities he has produced will increase and not decrease. What constitutes four? He discards evil and strives for true sūtras which are the six perfections [in] the Bodhisattva’s Basket. His heart has no pride toward sentient beings, and he is very humble. He knows the amount and when it is enough to receive alms in accordance with his dharma. He abandons all wrong livelihood and satisfies the holy . He does not produce someone’s sin either false or real and he does not seek for the stumbling of others. If he does not understand all dharmas, he thinks like this ‘the dharma of the Buddha is limitless. He preaches beings in accordance with their pleasure. Only the Buddha knows, I do not.’ He regards the Buddha as clear and does not contradict him. Kāśyapa, these are four

62 The use of the number was obviously inserted later in the Chinese translation process. Presumably, the translators did not want to be confusing. Different versions can have different ways of phrasal division so that sometimes the content in each number 1-4 is not similar in different versions. For example, in section 16 the phrase concerning nirvāṇa is included in number 3 in the Jin version but it is included in number 2 in Song version. This suggests that the translators of the Jin version and the Song version interpreted this passage differently. 63 T. 310, 12: 632a16-a24. (§von Stäel-Holstein 1926, §6) 31

qualities of a bodhisattva by which the wholesome qualities he has produced will increase and not decrease.

The Ratnamegha version:

善男子。具有四法。菩薩摩訶薩生生世世増長善根。轉勝明淨不復退失。何謂爲四。 捨離一切不淨之法。勤求一切清淨正法。夫正法者。所謂六波羅蜜菩薩法藏。於諸衆 生心無憍慢。猶若家犬。如法受施而自知量。捨離一切諸邪慢等。安心知足住於聖 種。不説他人罪過虚實。不求人短。若於諸法 心不通達。於諸不毀謗當作是念。佛法 無量爲諸衆生隨機所説甚深難解。唯佛自知非我所解。是故不生違逆之心。64

Son of a good family, a Bodhisattva Mahāsattva, endowed with these four qualities, will increase wholesome roots life after life. He will be brighter and purer without ever transgressing again. What constitutes four? He discards all impure dharmas and strives for all pure, true dharmas. These true dharmas are, namely, the six perfections [in] the Bodhisattva’s Basket. His heart has no pride toward sentient beings like a house dog. He is satisfied with the alms that he receives in accordance with his dharma. He discards all false pride. His calm mind knows to abide with the saintly attitudes. He does not accuse other people of what is real or unreal. He does not seek for the stumbling of others. If he does not understand all dharmas, he should not disparage all [dharmas], but should think ‘the Buddha bears eye-witness. The dharma of the Buddha is limitless. He preaches the sentient beings in according to their nature which is extremely difficult to understand. Only the Buddha knows, I do not. Yet, his mind does not contradict him.

From this passage, we notice that the Qin and the Ratnamegha versions are stylistically parallel to each other. Not only the content, but the vocabulary (in the bold letters) is also similar or sometimes identical. This same style of translation occurs throughout the whole text. For this reason, the Ratnamegha is rather a revised version of the Qin translation with some additional detail and some changes of word choices.

64 T. 659, 16: 276c04-c12, c14-c17. The Ratnamegha translation seems to have a scribal error in this part because the last part of the passage “於諸不毀謗當作是念。佛法無量爲諸衆生隨機所説甚深難解。唯佛自知非我所解。是故不生違逆之心” was moved to the middle of the next section. I am quite sure that this is not the error or variants from the Indic manuscripts but rather the error in the process of copying of the Chinese version since the content of this part does not fit well with the content of the other passage. Here, I edited the passage and moved it back to the supposedly original version in order to compare with the Qin version. Also, when we can identify that the Ratnamegha version imitates the language of the Qin version, we are sure that this scribal error occurred in the transmission of the Chinese version. 32

More importantly, the Ratnamegha has sections 150-156 similar to the Qin version, while the Jin version does not have this part (Table 1). However, it also has sections 157 and 159 which are similar to the Jin version. In this case, it might be that the Qin and the Ratnamegha versions might have descended from the same ancestral line. But the Ratnamegha might have been interpolated from the Jin line, and borrowed sections 157 and 159 from it. This situation is common in Buddhist textual tradition. It shows that the people who owned different versions of the same text were in touch with each other, and they accepted the variants by combining them.

Final stage: the harmonized version

The two stages mentioned above which I considered to be early and middle stages have one thing in common: they do not have verses except the section 136. The origin of the verse part in the later versions has been speculated to have come about from two possibilities. The first possibility is that there were two different versions—one with verses and one without—being circulated at the same time since the early period. The second scenario is that the summarizing verses on each section in the Sanskrit, Song, and Tibetan versions were inserted later.65 It was first noticed by de Jong that “The Kāśyapaparivarta, in which the verse parts are later than the prose parts, offers an interesting example of a text in which the verses, written in Buddhist

Hybrid Sanskrit, are definitely later than prose parts, the language of which is much closer to standard Sanskrit.”66 Later, Silk hypothesizes that “the verses we find in the unique, nearly complete Sanskrit manuscript of the KP (and evidenced by several translations) are not written in a genuine dialect, but are rather a result of intentional archaization.”67 This addition might have

65 Silk 2009b, 182. 66 de Jong 1977, 255. 67 Silk 2009b, 181. 33

happened “at some point during or before the 7th century.”68 If Silk is right about the period of the addition of verses, this would also be the date of the last stage of the development of the KP.

As we saw above, the extant Sanskrit text incorporates sections found for the first time in the Jin and Qin translations and thus may represent a conflation of previously independent versions. When conflated, the two translations representing two redactions may have been combined without any attempt to resolve anomalies. This conflation might be considered as a process of harmonization of the text. Besides this attempt to compile all existing versions at this time, the compilers then decided to insert verses written in an artificial Middle Indic style, which is nowadays knows as Buddhist Hybrid Sanskrit, into the prose archetype in order to make the sūtra look archaic as hypothesized by Silk. This process needs not have occurred in the same period. The conflation might have taken place before the insertion of the verses. This version with verses could have been regarded as a standard version which became popularized and overshadowed all other earlier versions before finally replacing them completely in some area.

However, as was the case with Mahāyāna scriptures, even after this harmonization, more passages were inserted later. In sections 21-21 and 50-51, the Sanskrit version, the Tibetan version, and the Song version have additional prose passages without summary verses that do not appear in the Han, Jin, or Qin versions. This fact might suggest that these passages were inserted later. More interestingly, these sections do not have summary verses at the end like other prose passages. This might also suggest that they were added after the stage of conflation and versification, which pushes the addition of these passages to an even later period.

In addition to this, the Sanskrit archetypes of the Song and Tibetan versions are not identical to the extant Sanskrit manuscript, but rather they are descendants of this extant Sanskrit

68 Silk 2009b, 182. 34

version. There are some minor differences between these two versions and the Sanskrit version that could be considered as interpolations, such as the opening passage in section one. In the

Sanskrit versions (and all earlier versions), it appears that the Buddha addressed Mahākāśyapa without the passage describing Kāśyapa approaching him before he started preaching with the first set of four qualities that would lead to the loss of a bodhisattva’s knowledge. But in the

Song and the Tibetan versions, there is a traditional opening passage that introduces the character of Kāśyapa before the Buddha addressed Kāśyapa:

爾時尊者⼤迦葉波在⼤衆中安詳⽽坐爾時世尊告迦葉⾔69

“At that time, Venerable Mahākāśyapa in the middle of the saṃgha was seated with serene manner. At that time the World Honored one addressed Kāśyapa saying . . .”

de'i tshe na tshe dang ldan pa 'od srung chen po 'khor de nyid du 'dus par gyur nas 'dug go de nas bcom ldan 'das kyis tshe dang ldan pa 'od srung chen po la bka' stsal pa

At that time, Venerable Mahākāśyapa, having gathered with the group, was seated in such manner. At that time the Blessed One addressed him saying . . .”

These two passages do not appear in the extant Sanskrit version. Moreover, in section 60 the Tibetan version and the Song version agree with each other against the Sanskrit manuscript which has a shorter passage. While the Sanskrit manuscript only explains that the idea of existence is one extreme and the idea of non-existence is another extreme with the middle way being in between the two, the Tibetan version and the Song version add more details about no- seeing, no-abiding, being devoid of consciousness, and devoid of characteristics. It is possible that the Song and the Tibetan versions might have had more details added later that the extant

Sanskrit version did not have. These additional passages in the Song and the Tibetan versions

69 T 352, Vol.12, 200c20. (von Stäel-Holstein 1926, §1) 35

may have been added after the harmonization of the text, and this might suggest that the extant

Sanskrit version might be older than the Song and the Tibetan versions.

In addition to some differences between the Sanskrit manuscript and later translations in the Chinese and Tibetan languages, the Tibetan version also shows some minor differences from the Song version. In section 140, the Sanskrit version has:

te tataś cyuta samānā ekacittaprasādena kālagatāḥ trāyastriṃśeṣu deveṣūpapannāḥ ten' eva hetunā iha mama śāsane pravrajitāḥ70

When they died from there, they died with a single faithful thought and were reborn among gods of the Heaven of Thirty-Three Deities. For that very reason they ordained here in my teaching.

However, the Tibetan has a sentence telling that when they died from the heaven, they were reborn in the human realm before ordaining:

de dag 'chi ba'i dus byas nas dad pa sems gcig rnyed pa des sum cu rtsa gsum pa'i lha rnams kyi nang du skyes so de nas shi 'phos nas 'dir te rgyu de nyid kyis nga'I bstan pa 'di la rab tu byung ste

After they died, because they attained a single faithful mind, they were born in the realm of the Thirty-Three Deities. After that, they transmigrated from the heaven realm, and they were born here. Because of that very reason they ordained [as a monk] in my teaching.

In this context, we might expect that the Song version that usually agrees with the Tibetan version might again agree with the Tibetan version in this sentence. But the Song version agrees with the Sanskrit version against the Tibetan version. This also confirms that it is not that the

Sanskrit version dropped this sentence or missed copying because the Song translation does not have this phrase. This sentence, indeed, does not exist in some versions. The Song version says:

於命終後求⽣忉利天宮。爲如是事於佛教中⽽求出家

70 von Stäel-Holstein 1926, §140. 36

After they died, they sought in the heaven of Thirty-Three Deities. Because of this very reason, they sought to leave home in the teaching of the Buddha.

More interestingly, the earlier translations—Jin, Qin and Ratnamegha—all have this sentence.

Even though there are some minor differences in wording (the human realm and the

Jambudvīpa), they confirm that this sentence exists in some earlier versions. The first passage is from the Jin translation, and the second passage is from the Qin translation:71

以是因縁。⾝壞命終⽣忉利天。從彼命終還⽣⼈間。於我法中出家學道

For this reason, after they died, they were born in the Heaven of the Thirty-Three Deities. When they died from there, they were reborn in the human realm. Then they left home to study the way in my teaching.

以是善心命終之後生忉利天。忉利天終生閻浮提。於我法中而得出家

Because of their good mind, when they died, they were born in the Heaven of the Thirty-Three Deities. After [dying from] the Heaven of the Thirty-Three Deities, they were born in the Jambudvīpa. Then they left home for my teaching.

This case could suggest that the Tibetan version might have consulted with the earlier versions, while the Sanskrit version and the Song version did not conflate the sentence.

In summary, it can be noticed that the incorporation of different versions happened repeatedly throughout the history of the text. First, the Ratnamegha version might be evidence of the interpolation of the ancestor of the Jin’s and the Qin’s archetypical lines. Then, the Sanskrit version appears to conflate parts of the Jin archetype and the Qin archetype. Presumably,

Buddhist communities where these different versions were circulated might have used different versions, yet they were in touch with each other and had access to other versions of the same

71 This passage in the Ratnamegha translation is almost identical to the Qin version so I do not quote it here. 37

text. Instead of refusing these variant texts, they tried to combine them, producing a compromise from different versions. The textual variation in the early period, especially in Mahāyāna societies, was accepted. The conflation between versions occurred often, which makes the relationships between them difficult to identify.

5. Discussion

The task of finding an Urtext of any Mahāyāna sūtra is an arduous one, and most of the time, if not all of the time, is impossible. It is not that Mahāyāna sūtras do not have an original model. The fact is that versions changed over a thousand years. In the case of the KP, the earliest version that we have is Lokakṣema’s translation in the late second century. This version could have evolved to a substantial degree before the translation. Even though it is the oldest version and it bears some archaic features, it can only represent the early middle stage of the development of the text.

The KP already shows us that general approaches to dating the versions of Buddhist texts by considering the archaic characteristics of their language might not always work. The summary verses in the KP of the Sanskrit version that were written in an artificial Middle Indic language or Buddhist Hybrid Sanskrit were proven to have been added later and the prose part, which is in almost perfect Sanskrit, is the older part. This fact does not follow the basic observation of textual criticism of Buddhist texts where oftentimes the verse part written in Buddhist Hybrid

Sanskrit is the older part since it usually maintains the archaic form of the texts. Also, one might expect that the versification was done for the purpose of oral memorization, as Gombrich states:

“The stricter the metre, the easier to preserve the wording.”72 However, it is proven from the case

72 Gombrich [1988] 2005, 76. 38

of the KP that this might not always be the case. When the process of versification occurred, which may have been around 7th-8th century CE, the text would already have been written down for centuries for the sake of preservation. Instead of functioning as memory-aid, the versification might serve as a process of archaization of the text and more importantly as the legitimization of the text. Most of Mahāyāna sūtras have both prose and verse combined. It is possible that the one who decided to add verses in the KP might try to imitate this characteristic in the belief that it is a significant feature of Mahāyāna sūtras. Therefore, when examining the history of any text we need to note that each sūtra has a different development and that a particular principle of textual criticism does not always work.

For these reasons, the study of the textual history of the KP seems to support Schopen’s claim that Mahāyāna sūtras “had in their early phases no fixed, standard form; such a form—if ever achieved—occurred only in their late phases.”73 In the case of the KP, it seems that at one point around 7th-8th century CE there was an attempt to harmonize the sūtra by combining two or more versions and then adding summary verses to the prose passages. The evidence of this conflation is the three later versions with verses: the Sanskrit version, the Song version and the

Tibetan version. These three versions preserve almost identical content and wording; only some minor differences between the three are found.74 The additional phrases in the Tibetan version which is similar to the Jin and the Qin versions might show that the community that circulated

73 Schopen 2009, 215. 74 Boucher 2008, 109. In his comparative studies of versions of the Rāṣṭapālaparipṛcchā, he finds that ’s translation is different from other later versions. He notes that “[w]ith Jñānagupta’s Chinese translation of the late sixth century, the Rāṣṭapāla undergoes far fewer changes by the time of its ninth-century Tibetan translation, its late tenth-century Chinese translation by Dānapāla, and its much later Nepalese Sanskrit manuscripts.” He proposes that the significant development of this text might happen in between 270-550 C.E. It is interesting that this date might be the beginning date of the development of the KP, too. 39

the Sanskrit archetype of the Tibetan version had access to earlier versions which were still preserved or were even in use in some Mahāyāna communities.

Even though we might not be able to find out what the Urtext of the KP originally looked like with the evidence that we have, it is still significant to verify the relationships between texts that were circulated under the same name with some variation in their content and language.

These variants represent the communities who composed, enlarged, and circulated them. They can also illustrate the relationships between one Mahāyāna community and the others. The concept of Mahāyāna Buddhism as different communities has been discussed by Silk. He has posited that the idea that there was only one singular Mahāyāna community and movement seems not to be right. He stated that “we must—provisionally—suppose each scripture to represent a different community, a different Mahāyāna.”75 He also proposed that “each reading of a work which produces a new interpretation allows, although it does not necessitate, the creation of a new community. Radical re-readings, which amount to rewritings, may indeed create new communities.”76 From the textual evidence of the KP, we can assume that different communities that had different versions of the text had been in contact with each other. They might even have read or listened to other versions and borrowed some parts that their version did not have and added them into their version. This could be an attempt to level their texts for the same standard. The additional parts usually did not affect the core meaning of the entire text, but they fulfilled some passages that might not have clear meaning or might cause confusion. This standard text was circulated in Mahāyāna communities and became the archetype of all later

75 Silk 2002, 369. 76 Silk 2002, 370. 40

versions. But again, as is the nature of textual transmission, even versions that were widely accepted later were subject to change throughout the course of time.

Another significant aspect in the study of the textual history of Buddhist texts is understanding their transmission process. Early Buddhist texts emphasize the transmission of exact words even if the meaning is different77 while Mahāyāna sūtras focus on the same meaning even though the wording is somehow different.78 The later Mahāyāna authors inserted verses in the KP without hesitation, and this version was accepted and became popular at least in a certain area where it was translated into Chinese and Tibetan in later period. This same situation also occurred with other Mahāyāna texts such as the Rāṣṭapālaparipṛcchā of which later versions in

Tibetan and Chinese translations have some degree of additional verses and prose passages.79

This additional part is quite long and cannot be due to errors in transmission but must rather be an intentional insertion. This might suggest that while verbatim transmission is important in the early Buddhist oral tradition, it is not emphasized in the Mahāyāna tradition where written texts were usually utilized to preserve and transmit texts. When Mahāyāna preachers read the texts aloud for their audience, they could probably pull some clichés from their memory or even give

77 To support this idea, Anālayo 2007, 9 notes that “The emphasis on verbatim transmission in the early Buddhist oral tradition can even be detected in some transmission errors, where at times in otherwise closely similar Pāli and Sanskrit passages the counterpart to a particular term shows close phonetic similarity but a considerably different meaning”. He suggests that “In such cases, it seems as if the attempt of the reciters to precisely remember has preserved formal aspects, even though the meaning was lost.” 78 Schopen 2009, 206 observes that even though there are variant copies of Bhaiṣajyaguru-sūtra which cannot be identified as the same text, they still basically have the same content. He notes that “those differences do not usually affect the general content, but the specific expression. (…) [T]hey say the same thing may be another matter.” 79 Boucher 2008, 108-110 finds that the earliest version of the Rāṣṭapālaparipṛcchā which was translated by Dharmarakṣa in the third century CE differs significantly from the later versions such as the extant Sanskrit manuscripts, the Tibetan and later Chinese translations. He notes that “[t]here are numerous passages and whole sections in the fully elaborated text that have no parallel in his [Dharmarakṣa’s] third-century translation.” Some of these missing parts are prose passages but the majority of them are verses. This seems to agree with the textual development of the KP where verse sections were added later. Boucher adds that “[i]n all, approximately 50 percent of our extant Sanskrit recension is not represented in our earliest Chinee translation. Most noticeably omitted are some 248-353 verses, reminding us that the often assumed historical priority of metrical sections in Mahāyāna sūtras may need to be qualified.” 41

some oral explanation which later became a part of the texts, and at the same time created variants.

One speculation on the nature of written culture in the early age of textual transmission is that some texts were written as notes for memorization. Mahāyāna teachers as well as

Mainstream monks might have only taken notes of the texts they wanted to teach and used their written notes as a guideline when they preached or recited the texts. When these texts were transmitted to their disciples, some interpolation and addition might have happened from the explanation they heard from their teachers. This type of written text can be seen in some very early Gāndhārī manuscripts. Lenz has studied Gāndhārī manuscripts of Avadāna stories. He finds that most of these manuscripts were written as only abbreviated stories. Each story has a sentence saying that “the reader should know the story that is being summarized and should be able to expand that summary with little difficulty (…), filling in all the details and twists and turns of plot necessary to make an expanded narrative.”80 From the same evidence, Salomon has remarked that “[s]uch notations create the impression that the scribe had written the stories for his personal use. They may have served as memory prompts or lecture notes for oral presentations, rather than being formal written texts intended to be read by others, like the primary texts on each of the scrolls.”81 When a monk preached the same text by using the same notes, he might not have used the same wording every time. His disciples could have memorized variations which preserved the same content. Therefore, the words and phrases might not be identical—in other words they were not transmitted verbatim—but the core content is similar.

80 Lenz 2010, XIV. 81 Salomon 2018, 232. 42

Therefore, it is plausible that one of reasons none of the versions of the KP are identical, and the fact that some versions can bear substantial differences might have to do with the medium of the textual transmission. Scholars have proposed that the survival of Mahāyāna texts depended chiefly on the use of written texts. The stylistic features of the KP also show that it might have been composed and transmitted in written form. The topic of the oral and literary characteristics of the KP will be discussed in detail in the second part of this study. However, I want to emphasize here one characteristic of written texts in Indian culture, namely that it is easy to interpolate by adding, cutting, or changing some passages in the text. In the oral tradition, particularly in the early Buddhist tradition, texts were recited orally in a group which means that an individual monk could not change any passage by himself alone since his recitation would be different from the entire group. The change might have been generated by a group of reciters.

However, in written tradition any scribe could change some passages in the text intentionally or unintentionally when he made copies. No one could be sure which version was the correct one.

This might be why we have many variants of Mahāyāna texts, and none of them are identical.

However, this is only one speculation that I attempt to explain in terms of what happened to the history of Mahāyāna textual transmission. Details of oral and literary style in the KP will be discussed in part two of this study.

43

PART TWO: The Composition and Transmission of the Kāśyapaparivarta

1. Introduction

In part one, I pointed out that the variation between versions of the KP might be the result of the medium used in the process of composition and transmission. I propose that in the Indian situation the use of writing might have facilitated the process of addition, interpolation, and conflation. Even though these incidences are usually understood to have occurred in the oral tradition, it is more likely that in the early Buddhist culture the written tradition was more susceptible to change.82 The early Indian culture emphasizes the verbatim transmission of texts, one of the prime examples being the , which is the model of oral transmission in India.

Every male must learn to recite the Vedas and some subsidiary treatises from a very early age up to 36 years.83 They can not only recite the entire body of Vedic texts from beginning to end without mistakes, but also recite the texts backward. Interestingly, it is not necessary for these to understand the meaning of the texts. The only mission they have is to memorize and recite the exact sounds of the Vedas for the sake of liturgy; the meaning can be learned later.84 Buddhist monks might have modeled their texts on the Vedic literature.85

However, since the Buddhist monks were not trained to recite Buddhist texts when they were young and some were non-brahmins who might not have been accustomed to the memorization of a large body of texts, the oral features of the Pāli canon such as mnemonic lists, stock passages and redundancy would have helped them to succeed in verbatim transmission of the

82 Gombrich [1988] 2005, 78 has written that the that were held after the death of the Buddha decided which discourses monks should memorize. He states that “it must have been difficult, if not impossible, to slip a new text into the curriculum.” 83 Gombrich [1988] 2005, 75. 84 Rocher 1994, 10. 85 Gombrich [1988] 2005, 75. 44

Buddha’s teachings.86 Even when writing was later used in the process of textual composition and transmission, there would have been only a small number of people who could read and write. These literate people could change their texts intentionally and unintentionally when copying and transmitting them to the next generation without anyone checking their mistakes.

However, most Indian monks would have been more familiar with oral recitation of scripture, which was done communally. Because the early Buddhist culture emphasized the verbatim transmission of their texts, when one recited a text incorrectly or forgot some part of the text, the group would correct his mistakes. Any change then would have to be done by the whole community, not individually.87

In fact, scholars have proposed that the advent of writing played a significant role in the emergence and the success of the composition and transmission of Mahāyāna scriptures in contrast to their predecessors who relied primarily upon the oral tradition. This argument was first pointed out by Gregory Schopen (1975). Then wrote an article entitled

“How Mahāyāna Began” in the year 1988 theorizing that the rise and survival of Mahāyāna scriptures relied heavily on the use of writing. Beside these two classic works on this particular topic, a few scholars such as Donald S. Lopez (1995), David McMahan (1998), Paul Harrison

(2003), and recently Alan Cole (2005) have attempted to fully or partially support and prove this theory with more evidence in wider perspectives. In addition, some Gāndhārī manuscripts recently discovered in northern and eastern which can be dated as early as the 1st century BCE can also provide us with more evidence that the written texts were circulated

86 Allon 1997a, 363. 87 Gombrich [1988] 2005, 78 claims that the change that happened in the early Buddhist texts is more likely to occur unintentionally. He says that “That is not to claim that no change occurred; but the changes must have been mostly unintentional, due to lapse of memory and to the contamination of texts as someone’s memory slipped from one text to another.” 45

in a very early period and probably as early as the rise of Mahāyāna Buddhism as Gombrich has hypothesized.88

However, a significant point that cannot be discarded is the importance of the oral/aural culture in Mahāyāna scriptures. Even though writing was employed in the composition and transmission at the beginning of Mahāyāna Buddhism, its scriptures still relied substantially on the oral tradition. Nattier compared different versions of the Ugraparipṛcchā and observed that some different passages in each version might be the result of interpolation. She claims that these interpolations were introduced by the oral tradition, which was still important in the Indian culture even though writing was already employed in the composition and transmission of

Buddhist texts.

Though the apparent taboo against recording in writing had already been broken as early as the 1st century BCE, the memorization of scriptures has continued to be a crucial element in Buddhist communities down to today. An oral text can be dictated and recorded by one or more scribes; a written text can be memorized and then recited orally. The boundary between oral and written Buddhist texts was clearly an open one, with frequent border crossing in both directions.”89

Nattier’s claim cannot immediately be verified, however the study of the stylistic features in

Mahāyāna texts will shed brighter light on her hypothesis. This part of my study will focus on the stylistic features of the KP which may evidence the influence of oral tradition in written texts. By examining the stylistics features of the KP, I will argue that although Mahāyāna texts

88 Gombrich ([1988] 2005, 80) suggests that “There has long been a general consensus that the earliest surviving Mahāyāna texts go back to the second or first century B.C. This chronology, albeit imprecise, clearly fits the time when writing came more into use and it was possible to commit large texts to writing. Maybe this had something to do with better materials.” However, the earliest evidence of the existence of Mahāyāna scripture is the Gāndhārī manuscript of the Aṣṭasāhasrikā Prajñāpāramitā which is dated to around first century CE. It is possible that the beginning of this text is as early as the first century BCE (Falk and Karashima 2012, 19). 89 Nattier 2003, 59. 46

might have been composed and transmitted in written form, they were not intended to be read alone quietly. These texts were composed for the sake of oral recitation and public teaching. As a result, the authors of these texts were aware of compositional styles that would facilitate oral recitation and aural comprehension.

1.1 The Composition and Transmission of Pre-Mahāyāna Literature

Scholars have widely agreed that the early Buddhist texts including many in the Pāli canon were composed and transmitted orally in their earliest stage. The historical context and the internal evidence both suggest that they were the product of oral tradition. In his study The Oral

Composition and Transmission of Early Buddhist Texts, Mark Allon has illustrated five reasons that convince us that the early Buddhist texts were composed and transmitted orally: (1) In the

Pāli nikāyas, there is no reference to writing or a literate tradition but only to the oral recitation of the texts. (2) The few passages that do mention writing in the Vinaya do not refer to writing down the texts in order to preserve them. (3) No rules in the Vinaya mention the use of writing.

(4) There is no material evidence to demonstrate the preservation of texts in written form in India in the early stage of Buddhism. Finally, (5) certain stylistic features used in the early Buddhist texts are characteristic of texts that are intended to be composed and transmitted orally.90

Moreover, Allon has thoroughly examined the stylistic features of a long sūtra in the

Dīghanikāya and their function in memorization.91 He finds that features of early Buddhist texts such as oral formulas, repetitions, numerical lists, and waxing syllable counts unquestionably confirm that they were composed orally. These stylistic features were developed to help monks

90 Allon 1997b, 39. 91 Allon 1997b, 43-49. 47

and nuns memorize texts as correctly as possible. Allon’s methodology is also applied in this study in order to compare the stylistic features of the early Buddhist texts with Mahāyāna sūtras.

More detail on his study as well as the oral stylistic features in the Dīghanikāya will be discussed later in this study.

1.2 The Composition and Transmission of Mahāyāna Scriptures

While there is little doubt about the original oral tradition of the early Buddhist texts, it is more complicated when it comes to the origins of Mahāyāna sūtras. The question of whether or not they were composed and transmitted orally has been discussed for decades without a definitive answer. On the one hand, Mahāyāna sūtras represent a novel compositional style that points to a change from orality to literacy. On the other hand, they also preserve some traditional characteristics that are considered to be elements of oral production such as stock phrases and formulaic passages. There is a debate between a main group of scholars who believe that

Mahāyāna scriptures were composed and transmitted in written form and another group of scholars who argue that they were composed orally.

The discussion on the origins of the composition of Mahāyāna sūtras is based on two types of evidence: the historical context and the textual context. The former refers to the period when Mahāyāna sūtras were composed, the invention of writing in India, and some developments in Mahāyāna Buddhism that are dependent on writing such as the cult of the book.

The latter refers to the characteristics of Mahāyāna sūtras such as their stylistic features, the complexity of narratives, the length of the texts, and the self-awareness of texts as texts. With all of this evidence, many scholars have come to the conclusion that Mahāyāna sūtras were not

48

composed and transmitted orally like their predecessors but rather in written form, a novel technology that arose in India in about the same period as the rise of Mahāyāna.

Gregory Schopen is the first scholar to bring up the discussion about the written tradition and its significance in Mahāyāna Buddhism.92 In his oft-cited article “The Phrase ‘sa pṛthivīpradeśaś caityabhūto bhavet’ in the Vajracchedikā: Notes on the Cult of the Book in

Mahāyāna,” he points out that Mahāyāna Buddhism invented a new cult which he calls “the cult of the book.” In response to Hirakawa’s theory (1963) on the origin of Mahāyāna that the relic stūpa was the religious center for Mahāyāna Buddhists, Schopen argues that Mahāyāna’s emphasis was on the cult of the book. He focuses on the phrase sa pṛthivīpradeśaś caityabhūto bhavet which appears as a formulaic phrase in many Mahāyāna sūtras and posits that this phrase should be considered in its literal meaning, not figuratively. That is to say, Mahāyāna authors considered the spot where a sūtra was preached or a physical book was set up as a real caitya, which was as important as the relic stūpas that might have been mainly worshipped and controlled by the Mainstream Buddhists. In many Mahāyāna sūtras, physical books and Dharma preachers are considered to be the center of focus for cultic activities such as veneration with flowers and incense. The veneration of physical books could not have existed if there had not been writing. Thus, we can say that this new cultic practice was developed after the fact that

Mahāyāna sūtras were written down in a physical book form.

Richard Gombrich writes an article entitled “How Mahāyāna Began,” hypothesizing that the use of writing was the primary factor for the survival of Mahāyāna Buddhism.93 The early

Mahāyāna texts would have been lost since the saṃgha would have not preserved them orally in

92 Schopen 1975, 147-181. 93 Gombrich [1988] 2005, 74-80. 49

the same way as the old canonical scriptures since they were newly composed and regarded with suspicion in most cases. However, since they were preserved in written form, they could survive and thrive later. Also, it is widely agreed that the Mahāyāna movement was begun in the first century BCE, which was the time when writing came into use.94 This is not simply an accident of history—the rise of Mahāyāna Buddhism is directly related to this context.

Donald S. Lopez also believes that Mahāyāna scriptures were composed in written form.95 Even though, superficially, Mahāyāna scriptures are stylistically similar to early Buddhist texts, “the Mahāyāna sūtras differ from the earlier works, however, in their self-consciousness and often exaltation of their own status as texts, as physical objects, with many works being devoted almost entirely to descriptions of benefits to be gained by reciting, copying, and worshipping them.”96 In other words, Mahāyāna scriptures are always aware of their status as texts that were composed in written form, whereas the early Buddhist texts were merely the words of the Buddha, without exalting themselves as texts.

David McMahan also puts forward that Mahāyāna scriptures are the result of literary composition.97 He mentions the same reason that Gombrich had several years prior, that

Mahāyāna Buddhism arose at the same time as the development of a writing system in India. The newly emergent Indian literary culture disrupted the long oral tradition of Buddhism that had been in existence since the death of the Buddha. Writing transformed the oral tradition to the

94 The first examples of writing in India are the inscriptions of king Aśoka of the Mauryan dynasty, dated roughly to the 3rd century BCE. We do not have any evidence of Indian writing earlier than this period (Salomon 1995, 272). However, we cannot be certain whether or not writing was used in the composition and transmission of religious texts immediately after its introduction in Indian culture. The earliest record of the use of writing for the preservation of Buddhist texts is in the Dīpavaṃsa and the Mahāvaṃsa. Both texts have the same statement saying that the writing down of the Pāli canon took place in the reign of king Vaṭṭagāmaṇī of Sri Lanka in the first century BCE. The purpose of this action is to avoid the loss or the decay of the Buddha’s teachings (Collins [1990] 2005, 77-78). 95 Lopez 1995, 21-47. 96 Lopez 1995, 41. 97 McMahan 1998, 251. 50

world of visual knowledge where people could read from manuscripts. McMahan gives three crucial reasons why writing was important for Mahāyāna Buddhism:

First, written texts were essential to the survival of this heterodox tradition; second, they provided a basis for one of the most important aspects of early Mahāyāna practice, that is, the worship of written sūtras themselves; and third, writing contributed to a restructuring of knowledge in such a way that vision, rather than hearing, became a significant mode of success to knowledge.98

His last point is questionable. He claims that Mahāyāna sūtras show the disruption of literacy in the continuity of the oral/aural tradition. He argues that Pre-Mahāyāna Buddhism used the oral tradition to claim legitimacy and authority over the words of the Buddha. However, Mahāyāna sūtras which were composed after the death of the Buddha did not have such institutional maintenance and legitimacy. To legitimate them, Mahāyāna authors established the “higher visionary worlds” which were revealed only to Mahāyāna followers. This emphasis on vision is the result of the shift of the source of knowledge “from a primary oral and auditory mode to a primary visual mode.”99 My contention, however, is that oral recitation and aural comprehension were still the central focus of Mahāyāna Buddhism, even though writing was introduced and employed in the composition and transmission of Mahāyāna sūtras. The dominant oral stylistic features used in the composition of the KP which will be discussed later in this study will make my point clearer.

Paul Harrison also maintains that Mahāyāna sūtras were composed in written form.100 He goes further in suggesting that the authors might be forest monks who were engaged in ascetic practice, since many early Mahāyāna sūtras mentioned samādhi and the experience of

98 McMahan 1998, 254. 99 McMahan 1998, 260. 100 Harrison 2003, 133. 51

meditation. He contends that not only town-dwelling monks could be dharma preachers and engaged in textual preservation, but forest monks too had to deal with the business of preserving, transmitting and studying scriptures. Nevertheless, some characteristics of these scriptures differ significantly from the early Buddhist texts such as in the complexity of narrative structure and the length. Therefore, he concludes that early Mahāyāna monks might have used writing as their medium to compose scriptures, as he states:

Although we can hardly prove that texts too were not produced orally, the frequent references in them in book-form strongly suggest that they began life in written form. Many of them certainly have features which we tend to associate more with literacy than orality, such as structural complexity and length, for example.”101

Alan Cole is another scholar who asserts that Mahāyāna scriptures are originally written texts.102 He writes at the beginning of his book Text as Father: Paternal Seductions in Early

Mahāyāna Buddhist Literature that Mahāyāna texts were written as texts.103 With this assumption, he gives seven types of evidence that are quite convincing. First of all, Mahāyāna sūtras refer to themselves as physical texts which should be copied and passed on. Second, the style of the representation of Buddha’s voice in Mahāyāna scriptures is different from the early

Buddhist texts. Third, Mahāyāna sūtras speak of themselves as texts in their narratives. Fourth,

Mahāyāna texts are aware of their status and insecure about their authenticity. Fifth, the narratives in Mahāyāna sūtras are usually told through the point of of omniscient narrators, which is fundamentally impossible. Many stories could not be told by any character in the sūtras, who would exist within a limited geographical and temporal zone. It is characteristic of a literary

101 Harrison 2003, 134. 102 Cole 2005, 13-18. 103 Cole 2005, 14. 52

product to be able to produce the world as the authors want it to be. Sixth, Mahāyāna sūtras try to overcome traditional Buddhism, even by dating their texts prior to the traditional ones. And lastly, Mahāyāna sūtras are more flexible with regard to time and space, which is considered to be a literary technique. With all of this evidence, Cole then concludes that “many Mahāyāna sūtras were first produced as literary compositions.”104

Among the scholars who posit that Mahāyāna sūtras were composed in written form, there remain two different opinions on the beginning of the use of writing in the composition of

Mahāyāna sūtras. Some scholars such as Schopen105 and Apple106 attempt to show that there was a shift from a purely oral tradition at the beginning of the Mahāyāna movement to a written tradition later. However, some other scholars such as McMahan, Harrison and Cole posit that writing might have been employed from the very beginning of the composition of Mahāyāna sūtras. Although dwelling on this topic is beyond the scope of this study, it should be noted that the self-awareness of Mahāyāna texts as physical texts is found even in the earliest Chinese translations by Lokakṣema including the Aṣṭasāhasrikā Prajñāpāramitā and the Pratyutpanna

Samādhi Sūtra. Any attempt to prove the chronological change from pure orality to literacy in

104 Cole 2005, 17. 105 Schopen 1975, 159 traces this change from three texts: the Diamond Sūtra, the Aṣṭasāhasrikā Prajñāpāramitā, and the Kāśyapaparivarta. He only focuses on passages that have the phrase sa pṛthivīpradeśaś caityabhūto bhavet. He first proposes that the passages concerning the cult of the book in the Diamond Sūtra and the Kāśyapaparivarta are different from the Aṣṭasāhasrikā Prajñāpāramitā and might even predate it. He observes that both the Diamond Sūtra and the Kāśyapaparivarta point to the fact that their primary concern is still the oral tradition. “This is the fact that in both texts, immediately after the articulation of the sa pṛthivīpradeśaś phrase, the emphasis suddenly shifts away from the ‘spot of earth’ and falls on the person who recites, etc. the dharmaparyāya. (…) As regards the KP, then, the final impression is that, in spite of its references to writing, its primary orientation is still to the oral tradition.” 106 Apple 2014, 35 studies the occurrences of the phrase dharmaparyāyo hastagato in Sanskrit texts and Chinese and Tibetan translations to understand the implication of this phrase. From his investigation, he argues that “a Mahāyāna-based Dharma-discourse was in the form of a book and that it was held as a textual object infused with the presence of a Tathāgata.” He also adds that “these texts may have initially existed as oral discourses before being written down.” 53

Mahāyāna texts would require earlier textual evidence, which could happen in the future with the help of Gāndhārī manuscripts—that is, if the change ever existed.

To come back to my argument in this study, the significance of the oral tradition cannot be discarded from even Mahāyāna literary production. Even though several scholars seem to agree that Mahāyāna sūtras were composed in written form and that writing played an important role in the rise and survival of Mahāyāna Buddhism, they never ignore the importance of the oral tradition in Mahāyāna sūtras. One significant point which is usually referred to as evidence for the emphasis on oral tradition in Mahāyāna sūtras is the crucial role of Dharma preachers (Skt. dharmabhāṇaka). The word dharmabhāṇaka in Mahāyāna sūtras refers to one who could read or memorize, recite, and preach Mahāyāna sūtras, usually in public.107 Several Mahāyāna sūtras mention that dharmabhāṇakas would either hold manuscripts in hand while preaching or recite them from memory.108 In the KP, the dharmabhāṇaka seems to be revered by the audience as if he was the Tathāgata.109 The extensive number of references to Dharma preachers in Mahāyāna sūtras and their significant role in dharma-preaching activity confirms that oral-centered tradition was still practiced in Mahāyāna Buddhism.110

107 Drewes 2011, 338; 340. 108 Apple 2014, 29 suggests that the phrase dharmaparyāyo hastagato “holding Dharma-discourse in one’s hand” in Mahāyāna sūtras can be understood figuratively and literally. The dharmabhāṇaka could hold actual manuscripts in hand while reciting or recite from memory. He says “embodying the dharmaparyāya in oneself through recitation and memorization or through holding it in book form is attested from the times of the earliest extant exemplars of preserved Mahāyāna sūtras.” 109 imaṃ dharmaparyāyaṃ śṛṇuyād vā udgṛhṇīyād vā likhed vā paryāpnuyād vā / tasya dharmabhāṇakasyāntike evaṃrūpā gauravā-m-utpādayitavyaḥ tadyathāpi nāma kāśyapa tathāgatasya (von Stäel-Holstein 1926, §160). One should hear or take up or copy or comprehend this discourse on the doctrine from a certain dharma- preacher, then he must generate such a respectful attitude toward that dharma-preacher just like, Kāśyapa, toward the Tathāgata. 110 See Drewes 2011, 331-372; Gummer 2012, 137-160. Gummer studies the role of Dharma preachers in Suvarṇa- (pra)bhāsottama Sūtra. She argues that “While Buddhists in various times and places have certainly revered the material text, the central role many such sūtras accord to the dharmabhāṇaka (the Buddhist preacher) demands that greater attention be paid to the role of oral performance in the actualization of the sūtra’s self-proclaimed transformative potential” (2012, 137). Drewes goes further and posits that dharmabhāṇakas were the “primary

54

Besides this account of the Dharma preachers in Mahāyāna sūtras which confirms the significance of oral practice in Mahāyāna Buddhism, I propose that the stylistic features of the composition of Mahāyāna sūtras can also prove the emphasis of orality in Mahāyāna literary composition. As already pointed out by earlier scholars, Mahāyāna sūtras make use of mnemonic features which are associated with oral tradition such as repetitions, formulaic phrases, and numerical lists. I will here examine the stylistic features of the KP to determine to what extent mnemonic and other oral practices appear in this text. My contention is that, even though we tend to associate Mahāyāna sūtras with literary composition, the main concern of the authors of the KP is to facilitate oral recitation and aural comprehension. This fact can be proven on the basis the stylistic features found in the KP.

2. Stylistic Features in the Kāśyapaparivarta

Mahāyāna sūtras have many compositional features similar to early Buddhist texts.

These features are considered mnemonic characteristics of oral composition and transmission and include numerical lists, repetition, and formulaic phrases. The KP is one of many Mahāyāna sūtras that have these oral-formulaic characteristics such as the use of similes and numerical lists of which the model is the early Buddhist texts. It also has repetitions of sounds, words, and phrases. The sample phrases and words in this study are selected from the whole text to illustrate the dominant oral-stylistic features in this text.

2.1 Formulaic expressions

agents” of the Mahāyāna movement. “Dharmabhāṇakas were the central figures of this movement. They composed Mahāyāna sūtras and claimed that they had been entrusted to them by the Buddha himself. They typically presented themselves as exalted bodhisattvas with knowledge and power nearly equal to that of Buddhas” (2011, 362-363). 55

Formulaic expressions or standardized phrases can be found all over Buddhist texts not limited to the Pāli canon. It is considered a significant feature of oral literature. In the study of formulas found in the Dīghanikāya, Allon focuses on the formulas that are used to depict the situation when someone approaches the Buddha, which he claims are found very often in his materials—the Dīghanikāya. Even though there are some variations regarding persons, places, and means, the approaching action is depicted in standardized formulas. This formula is

“predictable in the majority of cases.”111

The authors of the KP also used formulaic expressions in the composition of this text. I can identify two major formulas that are used in the KP: the formulas of numerical lists and the formulas of similes. It is noticeable that the authors of the KP used formulaic expressions when they listed the instructions as well as when they mentioned similes to explain any new concept.

2.1.1 Formula of numerical lists

The KP starts with lists of suggestions and restrictions for bodhisattvas. They are listed numerically like those in Cattukanipāta of the Aṅguttaranikāya in the Pāli canon, which is presented in form of “a stereotyped formulaic presentation.”112 The Aṇguttaranikāya starts from sūtras that have only a single aspect, increasing by one up to sūtras that have eleven aspects, which is the highest number of all sūtras in this nikāya.113 But the KP has only lists of four and two teachings. They are arranged in pairs of four prescribed behaviors and four proscribed behaviors; however, the pairs are not always on the same topic. The authors sometimes discuss

111 Allon 1997a, 111. 112 Norman 1983, 56. 113 See Norman 1983, 54-57; von Hinüber 1996, 38-41 for more information on the characteristics of the Aṅguttaranikāya. 56

one set of teachings and then switch to another topic for the restriction. For instance, section 11 mentions the four stumblings of bodhisattvas but section 12 explains the four paths of bodhisattvas. There are 22 numerical lists of four at the beginning of the sūtra and 10 numerical lists of two in §111-120.

The formula of numerical lists of four

The first 22 numerical lists always start with the number four (Skt. catur) in its proper declension. If it is in the nominative, it is followed by ime before the vocative of kāśyapa. If it is in the instrumental, it is immediately followed by kāśyapa. Then the authors explain the topic of each list before they use the phrases katame catvāraḥ yaduta or katamaiś caturbhiḥ yaduta depending on the grammatical case of the first part. Then it is followed by a list of four instructions before the authors repeat the first phrase and conclude with a verse.

1) Nominative formula à catvāra ime kāśyapa + (topic) / katame catvāraḥ yaduta +

(list of four instructions) / ime kāśyapa catvāraḥ + (repetition of the topic) / tatredam ucyate +

(verse)

For example,

catvāra ime kāśyapa dharmā bodhisatvasya prajñāpārihāṇāya saṃvartante / katame catvāraḥ yaduta agauravo bhavati dharme ca dharmabhāṇake ca / dharmamātsaraḥ ca bhavati / dharmācāryamuṣṭiñ ca karoti dharmakāmānañ ca pudgalānāṃ dharmāntarāyaṃ karoti / vicchandayati vikṣipati / na deśayati / praticchādayati / ābhimānikaś ca bhavaty ātmotkarṣī parapaṃsakaḥ ime kāśyapa catvāro dharmā bodhisatvasya prajñāpārihāṇāya saṃvartate / tatredam ucyate // agauravo bhavati ca dharmabhāṇake dharmeṣu mātsaryarato ca bhoti / ācāryamuṣṭiṃ ca karoti dharme dharmārthikānā ca karoti vighnam vicchandayanto vividhaṃ kṣipanto dharmaṃ na deśayati jinapraśastān so ātma utkarṣaṇi nityayukto parapaṃsane cābhirataḥ kusīdo / caturo ime dharmā jinena proktā prajñāprahāṇāya jinorasānām 57

etāṃ hi catvāri jahitvā dharmāś caturo parāṃ dharmajinokta bhāvayet //114

2) Instrumental formula à caturbhiḥ kāśyapa + (topic) / katamaiś caturbhiḥ yaduta +

(list of four instructions) / ebhiḥ kāśyapa caturbhiḥ + (repetition of the topic) / tatredam ucyate

+ (verse)

For example,

caturbhiḥ kāśyapa dharmaiḥ samanvāgatasya bodhisatvasyotpannotpannā kuśalā dharmāḥ paryādīyante yair na vivardhati kuśalair dharmaiḥ katamaiś caturbhiḥ yaduta abhimānikasya lokāyatanamantraparyeṣṭyā / lābhasatkārādhyavasitasya kulapratyavalokanena / bodhisatvavidveṣābhyākhyānena / aśrutānām anuddiṣṭānāṃ ca sūtrāntānāṃ pratikṣepeṇa ebhiḥ kāśyapa caturbhir dharmaiḥ samanvāgatasya bodhisatvasyotpannotpannā kuśalā dharmāḥ paryādīyante yair na vivardhate kuśalair dharmaiḥ tatredam ucyate // lokāyikam eṣati ābhimāniko kulāni ca yojayati lābhārthato buddhaurasā dviṣate ca bodhisatvāṃs teṣām avarṇaṃ bhaṇate samantāt noddiṣṭa no cāpi śrutāsūtrāntā hitena kṣipīta imi jinena proktān tam ehi dharmehi samanvitasya kuśaleṣu dharmeṣu na vṛddhir asti / tasmād dhi paṇḍita bodhisatvo dūrān vijahyāc caturo pi dharmān imā niṣevanta sudūri bodhaye nabhaṃ va bhūmīya sudūradūre //115

The formula of numerical lists of two

There are 10 numerical lists of two in §111-120. In this part, the authors refer to renunciants (pravrajita) in general. It starts with the number two dvāv imau in the nominative case, followed by the vocative of kāśyapa and pravrajitasya in the genitive, which precede the topic of the list. Before the explanation of the two instructions, there is a rhetorical question katamau dvau yaduta. After the explanation, like the formula of list of four, the topic is repeated and is followed by a verse.

114 von Stäel-Holstein 1926, §1. 115 von Stäel-Holstein 1926, §5. 58

dvāv imau kāśyapa pravrajitasya + (topic) / katamau dvau yaduta + (instruction 1) ca

(instruction 2) cetīme kāśyapa dvau pravrajitasya + (repetition of the topic) / tatredam ucyate

+(verse)

For example,

dvāv imau kāśyapa pravrajitasya gāḍhabandhano / katamau dvau / yadutātmadṛṣṭikṛtabandhanaṃ ca lābhasatkāraślokabandhanaṃ cetīme kāśyapa dvau pravrajitasya gāḍhabandhanaṃ / tatredam ucyate // dve bandhane pravrajitasya gāḍhe dṛṣṭīkṛtaṃ bandhanam uktam āryaiḥ satkāralābhayaśabandhanaṃ ca te sarvadā pravrajitenatyajye //116

2.1.2 formula of similes

The use of similes in the KP is extensive. The authors considered similetic language to be a significant means to express their concepts and instructions. These similes are not stated randomly but were arranged in the same stylistic pattern. The authors introduce similes with the common tadyathāpi followed by an emphatical particle nāma and kāśyapa in the vocative case.

Then they explain the subject of comparison before they introduce the object of comparison by using the word evemeva.

tadyathāpi nāma kāśyapa + (subject of comparison) / evem eva + (object of comparison)

For example,

tadyathāpi nāma kāśyapa kuṃbhakārasya bālabhājaneṣūdārāgnidānam / evam eva kāśyapa bālaprajñeṣu bodhisatvasyodāradharmadeśanā veditavyaḥ117

None of these sample formulas found in the KP were newly employed in the Mahāyāna literature. They were regularly used in the early Buddhist texts. The reason that they were also

116 von Stäel-Holstein 1926, §112. 117 von Stäel-Holstein 1926, §51. 59

used in the KP might be that they were regarded as effective in the early Buddhist texts for the arrangement of the texts. The authors of the KP as well as other Mahāyāna sūtras who were familiar with the early Buddhist texts and their stylistic features might have memorized these formulas from the early Buddhist texts and used them in their new sūtra composition. Not only do they help in the process of the proliferation of texts as Allon suggests,118 but the adoption of classical Buddhist idiom also helps the authors of Mahāyāna sūtras to disguise their roles in the production of these texts.

However, the use of oral formulas does not confirm that the texts were composed and transmitted orally. First of all, the authors could make use of the oral-formulaic passages with which they are familiar, since they will help them to structure the text and could be pulled from their memory anytime. In other words, Mahāyāna authors could adopt an already extensive repertoire of Buddhist phrases and stock expressions in their literary composition. Secondly, the use of formulaic passages in the KP is not consistent. Instead of paying attention to the similar pattern of composition for the sake of memorization, the author of the KP rather switches between different formulas such as nominative formulas and instrumental formulas, so that the sentences are grammatically correct. They focus on the content of the text more than the formulaic language, while in the early Buddhist texts the focus is placed on the pattern of textual arrangement. The content usually comes after.119 If the authors of the KP had intended to

118 Allon 1997a, 162. 119 One example is mentioned by Anālayo 2007, 12. The application of a stock passage or pericope can be inconsistent. The pericope that describes the Buddha or a monk getting ready for alms begging in the morning is used in the Vinaya and the Udāna without adjustment even though it does not fit the context. He states that “Even though the circumstances make it clear that the invitation to come to the local hall must have taken place in the late afternoon or evening, the Vinaya and the Udāna nevertheless report that it was “in the morning.” Another example which shows that the early Buddhist texts prioritize the formulaic pattern of the text over the content is the arrangement of the Aṅguttaranikāya. The text is arranged from number one to number eleven. K.R. Norman states that “In some of the higher-numbered nipātas, i.e. from the chakka-nipāta onwards, combinations of smaller numbers are sometimes resorted to to get the higher designations” (1983, 55). Since there are not many

60

facilitate the oral memorization of the text, they could have changed or adjusted the content in order to minimize the grammatical differences of some passages so that they were all similar and easy to memorize.

2.2 Repetitions

The KP also has some repetitive forms similar to early Buddhist texts, such as repetitions of sounds, words and phrases. Allon studies the repetitive style of sūtras in the Dīghanikāya in detail.120 He investigates the methods utilized in the oral composition of early Buddhist texts by examining the repetition of sentences, passages, whole sections of the text and the repetition of set structures of Udumbalikasīhanādasutta in the Dīghanikāya. He finds that there are many types of repetitions in this sūtra including verbatim repetition, repetition with minor modifications, repetition with important modifications and repetition of compositional structures.

He then calculates the total repetition in Udumbalikasīhanādasutta and finds that eighty-seven percent of the text has some kind of repetition. He concludes that this number is notably high throughout the whole composition of the sūtra. This repetition is “undoubtedly a mnemonic device.”121

This stylistic pattern can also be found in the KP. For example, the aforementioned numerical formulas are used 32 times (22 times for the list of four and 10 times for the list of two) and the formula of similes is used more than 50 times. Besides these formulaic expressions,

Buddhist teachings that have a higher number than six, the combination of two or three sets of teachings is unavoidable. The content of the text is not as important as the pattern of the arrangement, as K.R. Norman also says that “Sometimes the complements have no obvious connection. Sometimes some connection can be deduced (…). Sometimes the component parts are opposites (…)” (1983, 55). 120 Allon 1997b, 51. 121 Allon 1997b, 52. 61

we can also find other types of repetitions, for example repetitions of sounds, words, and phrases:

catvāra ime kāśyapa bodhisatvamārgāḥ / katame catvāraḥ samacittatā sarvasatveṣu / buddhajñānasamādāpanatā sarvasatveṣu samadharmadeśanā sarvasatveṣu samyakprayogatā sarvasatveṣu.122

This passage has the sound sama- that is repeated in the beginning of each phrase. samacittatā, samādāpanatā, and samadharmadeśanā. The last word samyakprayogatā could be counted as a part of this sound similarity if we assume that the original version of this text was composed in Prakrit ( à sammā/sama). The sound -tā repeated at the end of three words is also remarkable. Moreover, this passage also repeats that word sarvasatveṣu four times following four suggestions.

lābhasatkārārthiko bhavati na dharmārthikaḥ / kīrtiśabdaślokārthiko bhavati na guṇārthikaḥ / ātmasukhārthiko bhavati na satvaduḥkhāpanayanārthikaḥ / parṣadgaṇārthiko bhavati na vivekārthikaḥ123

This passage repeats the word arthika eight times. Each individual sentence uses the word arthika two times. The first use of the word arthika in each sentence refers to what a bodhisattva is (…arthiko bhavati), and the second is what a bodhisattva is not (na…arthikaḥ).

This pattern is repeated four times in one passage.

catvāra ime kāśyapa bodhisatvasya bhūtakalyāṇamitrāṇi / katamāni catvāri / yācanako bodhisatvasya bhūtakalyāṇamitraṃ bodhimārgopastaṃbhāya saṃvartate / dharmabhāṇako bodhisatvasya bhūtakalyāṇamitraṃ śrutaprajñopastaṃbhāya saṃvartate / pravrajyāsamādapako bodhisatvasya bhūtakalyāṇamitra sarvakuśalamūlopastaṃbhāya saṃvartate / buddhā bhagavanto bodhisatvasya

122 von Stäel-Holstein 1926, §12. 123 von Stäel-Holstein 1926, §15. 62

bhūtakalyāṇamitra sarvabuddhadharmopastaṃbhāya saṃvartate / ime kāśyapa bodhisatvasya bhūtakalyāṇamitrāṇi124

This passage repeats the phrase bodhisatvasya bhūtakalyāṇamitraṃ upastaṃbhāya saṃvartate in each aspect of the list. Since there are four aspects in one list, this phrase is repeated four times.

Even though repetitions are found in some parts of the KP, these repetitions do not seem consistent enough to signify oral composition. From Allon’s study, it is clear that repetition is an important feature of oral transmission because it is supposed to assist in the oral memorization of the sūtras. However, the repetitive style of the KP is different from that in the early Buddhist texts. The following paragraphs are sample passages from the Aṅguttaranikāya of the Pāli canon that show what repetition in early Buddhist text looks like.

Catuhi, bhikkhave, dhammehi samannāgato bālo avyatto asappuriso khataṃ upahataṃ attānaṃ pariharati sāvajjo ca hoti sānuvajjo viññūnaṃ bahuñ ca apuññaṃ pasavati. Katamehi catuhi? Ananuvicca apariyogāhetvā avaṇṇārahassa vaṇṇaṃ bhāsati, ananuvicca apariyogāhetvā vaṇṇārahassa avaṇṇaṃ bhāsati, ananuvicca apariyogāhetvā appasādanīye ṭhāne pasādaṃ upadaṃseti, ananuvicca apariyogāhetvā pasādanīye ṭhāne appasādaṃ upadaṃseti—imehi kho, bhikkhave, catuhi dhammehi samannāgato bālo avyatto asappuriso khataṃ upahataṃ attānaṃ pariharati, sāvajjo ca hoti sānuvajjo ca viññūnaṃ, bahuñca apuññaṃ pasavati. Catuhi, bhikkhave, dhammehi samannāgato paṇḍito vyatto sappuriso akkhataṃ anupahataṃ attānaṃ pariharati, anavajjo ca hoti ananuvajjo ca viññūnaṃ, bahuñ ca puññaṃ pasavati. Katamehi catuhi? Anuvicca pariyogāhetvā avaṇṇārahassa avaṇṇaṃ bhāsati, anuvicca pariyogāhetvā vaṇṇārahassa vaṇṇaṃ bhāsati, anuvicca pariyogāhetvā appasādanīye ṭhāne appasādaṃ upadaṃseti, anuvicca pariyogāhetvā pasādanīye ṭhāne pasādaṃ upadaṃseti—imehi kho, bhikkhave, catuhi dhammehi samannāgato paṇḍito viyatto sappuriso akkhataṃ anupahataṃ attānaṃ pariharati, anavajjo ca hoti ananuvajjo ca viññūnaṃ, bahuñca puññaṃ pasavatīti.125

124 von Stäel-Holstein 1926, §14. 125 Morris [1888] 1976, 3-4. 63

, possessing four qualities, the foolish, incompetent, bad person maintains himself in a maimed and injured condition; he is blameworthy and subject to reproach by the wise; and he generates much demerit. What four? (1) “Without investigating and scrutinizing, he speaks praise of one who deserves dispraise. (2) Without investigating and scrutinizing, he speaks dispraise of one who deserves praise. (3) Without investigating and scrutinizing, he believes a matter that merits suspicion. (4) Without investigating and scrutinizing, he is suspicious about a matter that merits belief. Possessing these four qualities, the foolish, incompetent, bad person maintains himself in a maimed and injured condition; he is blameworthy and subject to reproach by the wise; and he generates much demerit. “Bhikkhus, possessing four qualities, the wise, competent, good person preserves himself unmaimed and uninjured; he is blameless and beyond reproach by the wise; and he generates much merit. What four? (1) “Having investigated and scrutinized, he speaks dispraise of one who deserves dispraise. (2) Having investigated and scrutinized, he speaks praise of one who deserves praise. (3) Having investigated and scrutinized, he is suspicious about a matter that merits suspicion. (4) Having investigated and scrutinized, he believes a matter that merits belief. Possessing these four qualities, the wise, competent, good person preserves himself unmaimed and uninjured; he is blameless and beyond reproach by the wise; and he generates much merit.”126

These two paragraphs are from a sūtra in Cattukanipāta of the Aṅguttaranikāya. They are stylistically similar to the first part— numerical lists— of the KP which is presented in pairs of numerical lists of four. They mention two types of persons: the fool and the wiseman. The bold italic parts are words and phrases that are repeated in the same paragraph, and the italic parts are words that are not repeated in the same paragraph, but which are repeated in the second paragraph that comes later. From the two paragraphs above, we can see that almost all words are repeated in the same paragraph. Only a few words: katamehi catūhi and imehi kho are not repeated in the same paragraph. They are merely connecting words which can be used only once

126 Translation by Bodhi 2012, 388-389. 64

in one paragraph. When the second paragraph is compared, the whole passage is repeated in the same form with minor changes (affirmative to negative). The two paragraphs point out the differences between the fool and the wiseman by changing only a few words and using the same patterns. This is what we mean by mnemonic devices, which are the features that were created to help memorization in the early Buddhist texts.

Superficially, the numerical lists in the KP seem similar to these two aforementioned paragraphs:

catvāra ime kāśyapa dharmā bodhisatvasya prajñāpārihāṇāya saṃvartante katame catvāraḥ yaduta agauravo bhavati dharme ca dharmabhāṇake ca/ dharmamātsaraḥ ca bhavati dharmācāryamuṣṭiñ ca karoti / dharmakāmānañ ca pudgalānāṃ dharmāntarāyaṃ karoti / vicchandayati vikṣipati na deśayati praticchādayati / ābhimānikaś ca bhavaty ātmotkarṣī parapaṃsakaḥ / ime kāśyapa catvāro dharmā bodhisatvasya prajñāpārihāṇāya saṃvartate.127 catvāra ime kāśyapa dharmā bodhisatvasya mahāprajñātāyai saṃvartante katame catvāraḥ yaduta sagauravo bhavati dharme ca dharmabhāṇake ca / yathāśrutāṃś ca dharmān yathāparyāptān parebhyo vistareṇa saṃprakāśayati nirāmiṣeṇa cittena na pratikāṃkṣayati śravaṇaṃ lābhasatkāraślokaṃ / bāhuśrutyena ca prajñāgamaṃ viditvā ādiptaśiraścailopamaḥ śrutaṃ paryeṣate śrutāś ca dharmān dhārayati / pratipattisāraś ca bhavati na vyāhārapadavākyaparamaḥ ime kāśyapa catvāro dharmāḥ bodhisatvasya mahāprajñatāyai saṃvartante.128

These are four qualities, Kāśyapa, that lead to the bodhisattva's loss of wisdom. Which four? Namely: he is not respectful with regard to Dharma and Dharma-preacher; and he is stingy with the teaching and he makes the teacher's fist; and he makes an impediment of Dharma for those people who desire Dharma, he discourages, refuses, not teach, but conceals it; and he is very proud, thinks highly of himself, he is the one who disparages others. These four qualities, Kāśyapa, lead to the bodhisattva's loss of wisdom. These four qualities, Kāśyapa, lead to the bodhisattva's great wisdom. Which four? Namely: he is respectful with regard to Dharma and Dharma- preacher; he explains Dharma extensively to others in the way he learned it and the way he understood it with a mind free from worldliness, he does not desire profit, reverence or fame; and having known that the coming of wisdom is due to

127 von Stäel-Holstein 1926, §1. 128 von Stäel-Holstein 1926, §2. 65

learning much, he strives after learning like one whose head or clothes are lit; and being learned, he upholds the teachings; and the practice is the essence, words and verbal expressions are not foremost [for him]. Kāśyapa, these four qualities lead to the bodhisattva's great wisdom.”

The two paragraphs (§1-§2) above from the KP exemplify some of the repetitions in

Mahāyāna sūtras. In terms of the content, they agree with each other because the first one discusses four practices that lead to the destruction of bodhisattvas’ wisdom (prajñāpārihāṇāya saṃvartante) and the second paragraph talks about the four practices that lead to the increase of bodhisattva’s wisdom (mahāprajñātāyai saṃvartante). The arrangement is similar to the earlier

Pāli passages where the fool and the wiseman are mentioned side-by-side. The formula of the two paragraphs is also parallel, and the first practice is also repeated in both paragraphs: §1 agauravo bhavati dharme ca dharmabhāṇake ca and §2 sagauravo bhavati dharme ca dharmabhāṇake ca. Superficially, this pattern seems to follow the traditional oral features, which is very similar to the early Buddhist texts.

However, nothing is repeated after the first practice in the two sections. While the first section talks about the stinginess of bad bodhisattvas, the second section contrasts it by mentioning the one who explains teaching in detail to other people. The content seems to be parallel, but the wording is completely dissimilar. The nikāyas in the Pāli canon, when talking about two opposite things, will use negative formulas to negate the first thing using same wording. For instance, in the passages above when the first paragraph uses bālo avyatto asappuriso khataṃ upahataṃ, the second one contrasts it by using paṇḍito vyatto sappuriso akkhataṃ anupahataṃ, which is comprised of merely antonymous and negative forms of the first

66

passage.129 But in the KP, the authors are more willing to innovate or depart from parallel structure. While the first paragraph says dharmamātsaraḥ ca bhavati dharmācāryamuṣṭiñ ca karoti “and he is stingy with the teaching and he makes the teacher’s fist,” the second paragraph says yathāśrutāṃś ca dharmān yathāparyāptān parebhyo vistareṇa saṃprakāśayati nirāmiṣeṇa cittena na pratikāṃkṣayati śravaṇaṃ lābhasatkāraślokaṃ “he explains Dharma extensively to others in the way he learned it and the way he understood it with a mind free from worldliness, he does not desire profit, reverence or fame.” Even though the notion of the second paragraph is merely the opposite of the first one, instead of using a negative formula like early Buddhist texts, the authors seem to be more innovative and use a different phrase. This suggests that, even though repetitions are used in the composition of the KP, they are not so consistent that we can relate them to the original method of oral composition.

2.3 Other Stylistic Features

2.3.1 The arrangement of lexical strings in the waxing syllable principle

One of the most common characteristics of the early Buddhist texts is the production of a series of words or phrases that might be synonymous or grammatically related. According to

Allon, we find all types of lexical sequences in the Pāli canon: verb, adjective, noun, and adverb arranged by the length of words, or, as it is called, the waxing syllable principle.130 The waxing

129 Anālayo 2007, 16 notes that one of the oral features of the early Buddhist texts is the occurrence of repetition in positive and negative aspects. He states that “When treating a particular topic in its positive and negative manifestations, for example, it is a standard procedure in the discourses that the same passage is repeated with precisely the same words and formulations used for the positive case, making only the minimal changes required to adjust these to the negative case”. He also adds that “The same procedure becomes even more prominent when a series of different perspectives on a particular topic are explored. Thus, an examination of four types of person or modes of acting, for example, can use four times nearly the same text in order to achieve its aim”. 130 Allon 1997a, 191. 67

syllable principle entails the arranging of words from the words with the fewest syllables to the words with the most syllables, for instance rāja-kathaṃ cora-kathaṃ mahāmatta-kathaṃ

(4+4+6) or anna-kathaṃ pāna-kathaṃ vattha-kathaṃ sayana-kathaṃ (4+4+4+5).131 Each word in the string can also have sound similarities which help with the memorization of the whole string. Anālayo comments that the arrangement of the series of several words that are synonymous or related in the same phrase will preserve the whole series, since a single word is more easily subject to loss than a group of words.132

We can find a number of lexical strings in the KP that are usually arranged according to the waxing syllable principle. In section 7, we find three sets of lexical strings arranged this way:

catvāra ime kāśyapa kuṭilāś cittotpādās tena bodhisatvena parivarjitavyāḥ katame catvāra yaduta kāṃkṣā vimatir vicikitsā sarvabuddhadharmeṣu / mānamadamrakṣakrodhavyāpādāḥ sarvasatveṣu irṣyāmātsaryaṃ paralābheṣu avarṇāyaśokīrtiśabdaślokaniścāraṇatayā bodhisatveṣu ime kāśyapa catvāraḥ kuṭilāś cittotpādās tena bodhisatvena parivarjitavyāḥ133

The three strings are as follows:

kāṃkṣā / vimatir / vicikitsā à 2+3+4

māna / mada / mrakṣa / krodha / vyāpādāḥ à 2+2+2+2+3

a / varṇā / yaśo / kīrtiśabda / ślokaniścāraṇatayā à (1)+2+2+4+8

Not only do these three sets follow the waxing syllable principle, but they also show some sound similarities such as (1) vimatir / vicikitsā, (2) māna / mada / mrakṣa, (3) krodha / vyāpādāḥ.

From this example, we can see that the authors and transmitters of the KP were paying attention

131 Allon 1997a, 197. 132 Anālayo 2007, 6. 133 von Stäel-Holstein 1926, §7.

68

to the length of words when they composed and transmitted the text. To confirm this phenomenon, I will show in detail all lexical strings in the KP. I separate them into 2 groups: strings with connecting words and strings without connecting words.

Strings with connecting words

This type of sequence of words—adjective, adverb, or noun—has connecting words such as vā and na in between each word.

vā group

There are 6 sets of strings of words that have vā in between. In these six sets, four sets are considered to be single strings and the three other sets are considered to be two sets in one string.

One string can comprise from three words (the most common) up to as many as eight words in

§142.

one set per one string

§8 rājyapārihāṇir vā / dhanapārihāṇir vā / kāyajīvitāntarāyo à 6+6+8

§71 gṛhe vā / layane vā / avavarake vā à 2+3+5

§107 vākruṣṭo vā / tāḍito vā / paṃsito vā / bhaṇḍito vā / roṣito vā à3+3+3+3+3

§160 bhāṣyate vā / deśyate vā / likhyate vā / likhito vā / pustagataṃ vā à 3+3+3+3+4

two sets per one string

§97 rajyati vā / duṣyati vā / muhyati vā / atītaṃ vā / anāgataṃ vā / pratyutpannaṃ vā à

3+3+3, 3+4+4

69

§142 ātmā vā / satvo vā / jīvo vā / jantur vā / poṣo vā / pudgalo vā / manujo vā / mānavo vā / kartā vā / kārako vā / vedako vā / jānako vā / saṃjānako vā / utthāpako vā / samutthāpako vā à 2+2+2+2+2+3+3+3, 2+3+3+3+4+4+5

In section §97, even though it should be considered two sets since the meaning and grammatical category of the first set and the second sets are different, the entire string also follows the waxing syllable principle (3+3+3+3+4+5). In section §142, it is clearly composed of two separate strings based on the waxing syllable principle and the meaning of each set. The first set concerns living-beings in general, while the second set concerns the actors of verbs. In addition, the second set demonstrates sound similarity by repeating word-final -ko, which indicates the doer.

na group

There are four sets of strings of words that have na in between. Sometimes na is attached to the words if the word begins with a vowel or is a compound word. One string of words can comprise from two up to eight words.

§52 nasatva / najīva / napoṣa / napudgala / namanuja / namānavā à 3+3+3+4+4+4

§124 na nirvedāya / na virāgāya / na nirodhāya / no-paśamāya / nā saṃbodhaye / na

śrāmaṇyāya / na brāhmaṇyāya / na nirvāṇāya à 4+4+4+4+4+4+4+4

§139 nāvataranti / nādhimucyanti / nāvagāhanti à 5+5+5

§135 na śīlamanyanā / na śīlakalpanā / na vikalpanā / na saṃkalpanā /na parikalpanā

à 5+5, 4+4+5

70

The string in section §135 is a special one. If we consider it to be only one set, it does not follow the waxing syllable principle. However, when we consider the meaning of each word, it seems that there are two separate sets: the first set has two words starting with śīla- (śīlamanyanā

/ śīlakalpanā) and the second set has three words. As a result, we have here two sets of words that follow the waxing syllable principle.

No connecting word

Another type of lexical string found in the KP is those without a connecting word. These words can be adjectives, nouns or verbs, and the string can comprise from two words up to five words.

§8 ākrośa / paribhāṣaṇa / kutsana / paṃsana / tāḍana / tarjana / vadhabandhanā à

3+5,3+3+3+3+(2+3)

§10 tannimnaḥ / tatpravaṇaḥ / tatprāgbhāraḥ / tadguṇapratikāṃkṣī à 3+4+4+7

§14 yācanako / dharmabhāṇako / pravrajyāsamādapako / buddhā bhagavanto à 4+5+8,

6

§23 dharmacaryayā / samacaryayā / kuśalacaryayā à 5+5+6

§49 ikṣukṣetreṣu / śālikṣetreṣu / mṛdvīkākṣetreṣu à 5+5+6

§134 uttrasati / saṃtrasati / saṃtrāsamāpadyate à 4+4+7

§152 ucchada / snapana / parimardana / bhedana / vikiraṇa / vidhvaṃsana + (dharmaḥ)

à 3+3+5, 3+4+4

71

There are some important notes on this group. First of all, section §8 has two sets of words: the first one has ākrośa / paribhāṣaṇa (3+5) and the second one has kutsana / paṃsana / tāḍana / tarjana / vadhabandhanā (3+3+3+3+{2+3}). One might argue that this string should be regarded as one set which does not follow the waxing syllable principle. However, the meaning of the first set concerning verbal abuse and the second set concerning physical abuse demonstrates that it can be separated into two sets of lexical strings. Secondly, the last word vadhabandhanā can be considered as two words: vadha and bandhanā (killing and binding). But because of the tendency of the waxing syllable principle I consider it as a compound word vadhabandhanā (binding to kill) which has five syllables.

The set of words in section §14 is a special set which is taken from head words of the four sentences in one passage.134 In other words, they are not placed close to each other in original text but are each the head word of each sentence. This example shows that even though they are not a string of words that are next to each other, the authors of the KP also employ the waxing syllable principle when they make a list of teachings that are arranged under the same topic. However, the use of the waxing syllable principle in this case is not complete since it only works for the first three words but not the last one.

In addition to section §14 where the entire string does not follow the waxing syllable principle, we also find this inconsistency in other strings, too. For example, section §160 has a set of words as follows: śṛṇuyād vā / udgṛhṇīyād vā / likhed vā / paryāpnuyād vā. This set has a syllable structure of 3+4+2+4, meaning that it is not arranged from the least to the most

134 yācanako bodhisatvasya bhūtakalyāṇamitraṃ bodhimārgopastaṃbhāya saṃvartate / dharmabhāṇako bodhisatvasya bhūtakalyāṇamitraṃ śrutaprajñopastaṃbhāya saṃvartate / pravrajyāsamādapako bodhisatvasya bhūtakalyāṇamitra sarvakuśalamūlopastaṃbhāya saṃvartate / buddhā bhagavanto bodhisatvasya bhūtakalyāṇamitra sarvabuddhadharmopastaṃbhāya saṃvartate (von Stäel-Holstein 1926, §14). 72

syllables. I assume that because this string is unique in Mahāyāna literature, it might not have gone through the arrangement according to the waxing syllable principle that is meant to facilitate oral memorization or comprehension. This set of words could have been recorded in written form in the manuscript at the beginning of the composition and was therefore not arranged for the sake of memorization. However, with only one example, it is still too early to come to this conclusion.

Another important point to note about the string of words is that many members of each lexical sequence usually contain some sound similarities. These sound similarities can be found at the beginning, the middle, or the end of words. They can be similar consonants or vowels. For example, section §107 has sound similarity (-ito/-ṭo) at the end of each word: vākruṣṭo vā / tāḍito vā / paṃsito vā / bhaṇḍito vā / roṣito vā. In section §160 we also find sound similarity at the end of each word caused by the verb conjugation: bhāṣyate vā / deśyate vā / likhyate vā / likhito vā / pustagataṃ vā. Even though the last two words are not conjugated verbs, they also have the same t sound at the end: likhito vā / pustagataṃ vā. In section §97, we find two sets of sound similarities: the first one is similar to section §160 in that the verb conjugation creates the similarity of sound: rajyati vā / duṣyati vā / muhyati vā. The second set has -aṃ at the end: atītaṃ vā / anāgataṃ vā / pratyutpannaṃ vā. Beside verb conjugation, noun declensions also create sound similarities such as (§124) na nirvedāya / na virāgāya / na nirodhāya / no- paśamāya / nā saṃbodhaye / na śrāmaṇyāya / na brāhmaṇyāya / na nirvāṇāya. The two words

śramaṇa and brahmaṇa also sound almost identical except for the beginning consonants: na

śrāmaṇyāya / na brāhmaṇyāya. Sections §8, §135, §142, and §152 also have sound similarity at the end of each word: (§8) paribhāṣaṇa / kutsana / paṃsana / tāḍana / tarjana / vadhabandhanā, (§135) na śīlamanyanā / na śīlakalpanā / na vikalpanā / na saṃkalpanā /na

73

parikalpanā, (§142) kārako vā / vedako vā / jānako vā / saṃjānako vā / utthāpako vā / samutthāpako vā, (§152) snapana / parimardana / bhedana / vikiraṇa / vidhvaṃsana. In section

§10, we see a special case where the sound similarity occurs at the beginning of each word: tannimnaḥ / tatpravaṇaḥ / tatprāgbhāraḥ / tadguṇapratikāṃkṣī. This similarity of sound is created by compounds beginning with the same first member tat. Moreover, this set also has another set of sound similarity internally: tatpravaṇaḥ / tatprāgbhāraḥ / tadguṇapratikāṃkṣī.

2.3.2 Sets of related subjects

Another stylistic feature that is employed by the authors of the KP is to arrange the content by grouping related subjects together. This type of arrangement is prominent in

Saṃyuttanikāya (Skt. Saṃyuktāgama). According to K.R. Norman, the arrangement of

Saṃyuttanikāya is made by three main categories: subject of doctrine; class of , demon, or man; and some prominent person as speaker or hero. Norman also states that the arrangement of

Saṃyuttanikāya is “an obvious editorial practice”.135 In other word, the Saṃyuttanikāya is the result of the compilation of the Buddhist teachings that might have been passed down without systemic arrangement in the early period either in the time of the Buddha or after his death. To memorize and transmit these teachings, Buddhist monks arranged them by grouping the teachings that had the same subject-matter together. This practice was to facilitate memorization and oral recitation.

The KP is not as long as the Saṃyuttanikāya and its arrangement is not as well ordered.

However, in some parts of the text, especially the simile part, the text shows that the authors attempted to group the same subjects together. For example, when they mention the earth (§29),

135 Norman 1983: 50 74

it is followed by fire (§30), water (§31), and wind (§32) as a set of elements; or when they mention the moon (§34), it is followed by the sun (§35). Sections §129-130 mention doctor and patient and sections §131-132 concern precious objects such as jewels and gold. In addition, sometimes a negative condition and a positive condition follow each other. For example, section

§72 explains that seeds cannot grow in the air; then section §73 says that seeds can grow in soil.

Likewise, section §74 says that lotus cannot be planted in the forest, but it can grow in the mud

(§75).

This type of textual arrangement does not only facilitate the memorization of the text, but also aural comprehension so that the audience can follow the content of the text without confusion. In some parts, the authors attempt to show contrasting ideas of different related subjects such as the planting of lotus in the forest and in the mud. Although the early Buddhist texts employ this type of arrangement to help with oral transmission, it is not only exclusive to the oral tradition. Such arrangement is useful in any type of text.

2.3.3 Numerical order

Numerical arrangement is another method that can aid memorization. One of the most prominent collections that is arranged numerically is the Aṅguttaranikāya. Besides this nikāya, there are two sūtras in the Dīghanikāya that are also arranged in numerical order: Saṃgītisutta and Dasuttarasutta. Gethin has commented on the significance of the numerical list: “A list immediately imparts to the discourse a structure that makes it more easily remembered by the one giving the talk. At the same time a talk based on lists is easier to follow and remember for those listening.”136 We can imagine that when one who gives a talk has a list in one’s hand, it

136 Gethin 1992, 150. 75

will “safeguard against losing one’s way in a talk or forgetting a section of it.”137 The numerical arrangement of the early Buddhist texts could have been done later in the editorial process for the sake of memorization. For this reason, such texts that are arranged numerically can be considered

“late one[s].”138

Numerical lists are used frequently in the KP, as mentioned previously. Besides the aforementioned numerical lists of four and two which are prominent in this text, one passage in particular also emphasizes the significance of lists in numerical order:

catvāri samyakprahāṇāni sarvākuśaladharmaprahāṇāya cikitsā / sarvakuśaladharmapāripūryai saṃvartante / catvāro-d-ṛddhipādāḥ kāyacittapiṇḍagrāhotsargāya saṃvartante / cikitsā : paṃcendriyāṇi paṃca balāni / aśrāddhyakausīdyamuṣitasmṛticittavikṣepa- asaṃprajanyatāduṣprajñātānāṃ cikitsā : sapta bodhyaṃgāni dharmasamūhājñānasya cikitsā : āryāṣṭāṃgo mārga dauṣprajñāsarvaparapravādināṃ kumārgapratipannānāṃ cikitsā139

The four perfect abandonments, which are the treatment for the abandonment of all bad qualities, lead to the fulfilment of all good qualities. The four elements of supernatural power, which are the treatment, lead to the renunciation of the grasping of body and mind as unit. The five faculties are five powers, the treatments that lead to the disbelief, sloth, stolen , mental distraction, no-intelligence, and wrong knowledge. The seven members of enlightenment are the treatment of the heap of ignorance with regard to the teaching. The eight noble paths are the treatment of those who practice wrong paths, who are ignorant promoters of all false [views].

This passage shows the numerical order from four to eight. The arrangement of this passage not only facilitates memorization, but it also prevents confusion in the audience. Tt would be apparent if one were to switch the order, such as the seven members of enlightenment to before the five faculties since it is not in numerical order.

137 Gethin 1992, 150. 138 Norman 1983, 43. 139 von Stäel-Holstein 1926, §95. 76

3. Discussion

The investigation of stylistic features used in the composition of the KP confirms that even though there was a shift of media used in composition and transmission from an oral tradition in the early Buddhist texts to the literary or written tradition in Mahāyāna scriptures, the compositional style of Buddhist sūtras did not drastically change. Mahāyāna authors were influenced by the language of oral literature and they presented their texts as oral in origin. Even though they were composed in writing, the majority of their recipients are likely to have encountered these texts in aural form. We find many stylistic features in the KP which confirm the emphasis of the oral/aural tradition: formulaic expressions, repetitions, and other miscellaneous styles that are related to oral practice. All of these stylistic features are also found dominantly and consistently used in the early Buddhist texts.

Because of many oral stylistic features found in Mahāyāna sūtras, some scholars, especially Drewes, go further and argue that Mahāyāna scriptures were composed and transmitted orally. Drewes points out that Mahāyāna sūtras frequently make use of mnemonic and oral practices. It is unusual for written scriptures to have these dominant features of orality.

The mnemonic devices are used in Mahāyāna scriptures in order to facilitate memorization. He also adds that writing was not obligatory for the preservation of Mahāyāna scriptures in the early phase of the movement. There is no reason to believe that the textual practices of Mahāyāna

Buddhism were different from mainstream Buddhists.140

In addition to this, Drewes has recently written an article, entitled “Oral Texts in Indian

Mahāyāna”, providing more evidence to support his argument on the oral origin of Mahāyāna

140 Drewes 2010, 60; 2015, 133-135. 77

texts. In this article, he dedicates most of the content to the investigation of three Sanskrit verbs which are the key words in Mahāyāna textual practices: the three are ud√grah, √dhṛ, and pari ava√āp (or pari√āp). He argues that these three words are primarily related to the oral tradition, not the written tradition, as he claims that “all three words seem to refer, or be understood to refer, to memorization consistently through the whole history of the tradition.”141 He then rejects

Gombrich’s argument on the significance of the writing system that “Mahāyānists surely could have preserved their texts without writing…Writing was not necessary for the Mahāyāna to emerge.” Finally, he concludes that “… it is no longer clear that Mahāyāna sūtras were written compositions in the first place. Since oral and mnemonic practices seem to have remained central in Mahāyāna, there is significant likelihood that Mahāyāna sūtras were typically composed orally.” 142

Even though I agree that the oral tradition had a great influence on Mahāyāna sūtras, the assumption that they were composed orally seems unconvincing. Mahāyāna sūtras usually refer to themselves as physical books that should be recited, copied, and preserved. This is a crucial point which differs from their predecessors—the early Buddhist texts—that never refer to themselves as texts, but merely the words of the Buddha. Moreover, Cole has already pointed out that Mahāyāna sūtras “often present the Buddha speaking and discussing the text itself in ways to promote and legitimize it.” This shows that “Mahāyāna sūtras are self-aware in the way that earlier Buddhist discourse is not.” The complex narrative flow in Mahāyāna sūtras illustrates that

“once the language was visible in orthography and book form, it was a whole lot easier to write narratives in which the narrative, as object, figures in that flow of narrative events.”143

141 Drewes 2015, 219. 142 Drewes 2015, 132-133. 143 Cole 2005, 15. 78

Furthermore, the oral stylistic features in the KP alone might not be adequate to confirm that this text was originally composed orally. Allon suggests that “stylistic features do not prove that this literature was essentially an oral one, for written texts can utilise or mimic characteristics of an earlier oral tradition.”144 In the case of the KP, we do not find consistency in the usage of the oral stylistic features of this sūtra. Sometimes the repetitions are found in one passage but not another. Sometimes formulaic expressions are changed in accordance with the context which creates inharmonic stylistic patterns. This is a great difference from the nikāyas in the Pāli canon where most of the oral stylistic patterns are used consistently in order to facilitate their oral transmission. Moreover, we need to note that the good rhythmic sound in Indian literature, which originated from oral stylistic features, became the artistic style in Indian poetic composition. Rocher has emphasized the significance of “sound” in Indian literary culture. He contends that not only is sound important for the liturgies, but the beautiful sound can also be “a source of pleasure and bliss.”145 This is confirmed in Alaṃkāra-śastra, where some oral- originated features, particularly repetition, are considered to be a type of esthetic style of composition, such as . The use of yamaka can be found in the earliest Indian literature— the Vedas, and even in any later kāvya.146 It shows that oral/aural tradition was never discarded from the Indian literary culture even though writing had been employed more than two millennia prior.147

The oral characteristics found in Buddhist texts do not only serve as a mnemonic device, but they also have other functions in the Buddhist tradition. First of all, the oral stylistic features

144 Allon 1997b, 54. 145 Rocher 1994, 11-12. 146 See Söhnen 1995, 495-520 for the discussion on the concept of yamaka in Alaṃkāra-śastra. 147 Pollock 2006, 83. “The Daṇḍin and Bhartṛhari who celebrated the spoken lived in a world where the written had permeated both kāvya and śāstra for centuries. Many of the foundational texts of the Sanskrit intellectual tradition were composed in a literate environment even as they bear the shadow of the oral.” 79

such as repetition and formulaic expressions are beneficial to the proliferation of the composition. When the same type of expression needs to be repeated, the authors can instantly recall it from their memory because “the wording is already available.”148 Allon suggests that

“these formulas acted as prefabricated building-blocks.”149 As a result, the authors can generate new compositions with the existing wording patterns. For example, sections 21-22 in the KP were clearly added later because they did not have summary verses and they did not exist in the earlier versions. However, these two later sections were presented using the same formula of the list of four. This shows that the formulaic expression can help proliferate the text even in later period. Moreover, Gethin has suggested that the use of numerical lists in the composition of

Buddhist texts “constituted a meditation exercise in itself that cultivated insight, wisdom, and mindfulness and inspired faith in the teaching of the Buddha.”150 Anālayo also points out how important oral recitation is in Buddhist tradition. He finds that in several early discourses the five occasions for reaching liberation are stated. One of these five occasions emphasizes oral recitation which can develop one’s mind toward liberation.151 We may not be able to ensure whether or not oral recitation has a real impact on the liberation of mind. However, if the tradition regards oral recitation as one of the ways to liberation, it is not surprising that they still follow this practice nowadays.

Besides the benefits to composition and meditation, oral recitation in Mahāyāna sūtras is said to be one of the great ways of merit-making. The significance of oral recitation was shifted from being the only method of textual transmission and preservation in the early Buddhist texts to serving as the way to gain merit. Mahāyāna followers were encouraged to recite Mahāyāna

148 Allon 1997a, 162. 149 Allon 1997a, 162. 150 Gethin 1992, 167. 151 Anālayo 2007, 16. 80

texts not to preserve them but to gain merit from reciting them. Merit making from reciting texts, to my knowledge, is not stated in the early Buddhist texts, presumably because oral recitation used to be a normal way to transmit texts. Anyone who wished to learn Buddhist texts would have had to recite the texts orally. In other words, oral recitation was the only way of learning the

Buddha’s words that monks and nuns would have had to practice in their daily life without anyone encouraging them to do so. Mahāyāna monks modeled their texts and practices on early

Buddhism. In order to claim the authority and authenticity of their new texts and practices, they would have had to encourage their followers to recite their texts in the way that the Mainstream

Buddhists did with the early texts. By doing so, reciters would gain great merit even if they could recite only one stanza.

Finally, oral texts such as those found in the Pāli canon usually preserve their original oral features, even though they were written down later after writing was introduced to their society. In fact, the changes in oral transmission might be intended to help memorization. In other words, the transmitters would change some inconsistent features in their texts to harmonize the entire text so that it is easier to memorize. They would add more standardized phrases in their texts or rearrange the disorganized content. It would be quite wrong if we assumed that the

Buddha and his disciples talked stylistically exactly like what appears in the Pāli canon that we have today. In fact, the composition of the Pāli canon has been adjusted and changed throughout its history before it was fixed. That adjustment occurred to help standardize the entire canon so that it assisted people who memorized it. Allon claims that the original basis of the Pāli canon might have been the discourses of the Buddha that had been improvisatorially given by monks and nuns after the death of the Buddha from their memory. These discourses which were

“fundamentally different from the essentially fixed, memorised texts transmitted by the

81

community”152 later became the origin of fixed texts. The transmitters of early Buddhist texts tried to “minimise differences and maximise similarities” by “using a standardised diction”,

“proliferating similar word elements”, and “pursuing repetition on truly large scale.”153 This, I believe, did not happen all at once but rather took centuries to accomplish. Therefore, if the KP had been primarily transmitted orally like the Pāli canon, it would have been harmonized more in the course of a millennium. The fact that the KP preserves only some oral features and they are used inconsistently might suggest that it was written down from the beginning so that it did not go through the process of “minimizing differences” and “maximizing similarities”.

In sum, the oral features found in the KP is not sufficient to support the assumption that it was composed and transmitted orally since the other internal and external evidence seems to support that it was composed and transmitted in written form. However, the examination of stylistic features in this text helps to confirm on solid grounds that the oral tradition had crucial influence on the composition of this text. Oral tradition, which is always prioritized in Buddhist practice of any tradition, is still performed nowadays. Anālayo has written that “oral recitation continued for a considerable time even after the writing down of the discourses and is to some degree still practised today, which shows that it serves a greater purpose than preservation of the text.”154 In addition to this, Pollock also suggests that “Orality in India sometimes seems as much ideology as a fact of practice, for the oral ideal persisted long after writing had become fundamental to the Sanskrit tradition itself.”155 This seems to be true with the composition of the

152 Allon 1997b, 54. 153 Allon 1997b, 53. See also Gethin 2007, 383. “[A]lthough over time these repetition sections have become more or less fixed, they originally seem to have been composed in a manner that invites addition and expansion—within certain parameters.” 154 Anālayo 2007, 17. 155 Pollock 2006, 83 argues that the kāvya genre literature in India is a new type of literature which differs from the Vedas. “[t]he form of language usage called kāvya nonetheless represents something profoundly new in Indian

82

KP and many other Mahāyāna sūtras since their authors still employed an oral style in their composition. However, since writing later became the main method of textual composition and transmission, the use of oral features that used to be a mnemonic device was not employed consistently, or at least not as consistently as the early Buddhist texts that relied exclusively on oral transmission. While Allon demonstrated that eighty-seven percent of wording in the

Udumbalikasīhanādasutta is repetition in some way or another, there is no significant percentage of repetition in the KP. There are, indeed, some oral features in this text as I show above but it is not sufficient to confirm that they are significant for memory aid. This, at least, suggests that the

KP was chiefly circulated in written form without relying heavily and primarily upon oral transmission.

cultural history” (76). The innovations in the kāvya, he states, “include new specific norms, both formal and substantive, of new genre categories; and the application of a new storage technology, namely, writing” (77). In the Rāmāyaṇa, the first poem, the sage Vālmīki shows how to compose his poem and transmit it orally to his disciples. Pollock contends that this is the fiction of written culture “[f]or it clearly recognizes orality as such from outside orality, so to speak, and in a way impossible to do in a world ignorant of any alternative—ignorant, that is, of writing” (78). Orality in Indian culture seems to be more ideology than practice. Even in the time when writing was widely used, Indian authors such as Daṇḍin and Bhartṛhari still celebrated the orality of their works. 83

CONCLUSION

This study provides more evidence to confirm that oral tradition and literary tradition intertwined and could not be separated from each other in the process of the composition and transmission of Mahāyāna sūtras.156 Most scholars seem to agree that Mahāyāna sūtras, albeit chiefly influenced by oral/aural tradition, were written down from the start and did not go through the process of oral transmission. Even though our historical and textual evidence seems to support that Mahāyāna texts were composed and transmitted in written form, the nature of the texts demonstrates substantial borrowing from the oral/aural tradition. In other words, they were composed for oral recitation, not for private reading.

In the first part of this study, I investigate the historical context of the KP by examining each of the versions that came down to us in Chinese and Tibetan translations as well as the

Sanskrit manuscripts. I propose that the process of the development of the KP comprises three main stages: the early stage, the middle stage, and the last stage. The early stage is the period when the text might have been only the teachings of Mahāyāna teachers. In this stage, it is possible that the language used in the text might have been some , not Sanskrit.

Unfortunately, we do not have any record of this text nor of any other Mahāyāna sūtras in this early period because the beginning of Mahāyāna sūtras happened off-camera. The nature of the beginning phase of Mahāyāna sūtras is still quite unclear.

The middle stage is the period when most of our extent texts were developed differently in allegedly different Mahāyāna societies that owned the text. These different versions were

156 See also Strauch 2014, 789. In his study on the manuscript culture of early Buddhism, he also proposes that the writing down of early Buddhist texts and the oral tradition “influenced each other.” However, he suggests that this circumstance occurred because the process of writing down the early Buddhist texts did not take place at once but “it rather took place as a series of consecutive events” while oral tradition was still going on. However, he notes that “the mechanisms of these complex processes are largely unknown.” 84

translated into Chinese, and some are still preserved in manuscript fragments in Sanskrit. The most important feature of the versions of the KP in this stage is that they did not have the summary verses at the end of each section. However, each version had slightly different wording. The differences did not have crucial effect on the overall meaning of the text, but they were different in terms of word choice and order. In addition, there are some additional parts at the end of the sūtra that are clearly different in each version. I propose that this ending part of the text is evidence of the relationship between each version. Lokakṣema’s version, which is the earliest version, does not have this part, while all later versions do. The Jin and the Qin versions, despite their similarity, are different at the end. Thus, we can assume that they derived from different ancestors. The Ratnamegha version is more likely to derive from the Qin’s ancestor but was also interpolated from the Jin’s ancestral line.

The last stage is the stage when there was an attempt to harmonize the text. The three versions in this stage—the Song version, the Sanskrit version, and the Tibetan version—all have summary verses at the end of each section. In addition, these three versions in the last stage conflated and harmonized the last section of the sūtra which used to be different between the Jin and the Qin versions. This harmonized version became a new version which was later translated into Chinese and Tibetan. Although the three texts are usually similar, there are still some slightly different wordings. This leads to the conclusion that the development of Mahāyāna texts was occurring continuously and it is doubtful that the concept of a standard version ever existed.

At one point, a Mahāyāna community might consider one version of the text as a standard version, but this standard version might later become just another variant when time passed or when comparing the same text of different Mahāyāna communities.

85

In examining the variants of the KP, we can confirm that there was connection between different Mahāyāna societies which transmitted different versions of the same text. They might even have had access to different versions so that they could consult and combine different versions. However, among all the versions that we have, there is no single version that is completely identical to another version. These variants, I propose, might be the result of the use of writing from the beginning of the Mahāyāna movement. While the early Buddhist texts relied primarily on verbatim transmission, Mahāyāna texts were composed and transmitted in written form. As a result, changes could happen easily between each version without correction through communal recitation as in the early Buddhist tradition. Although this claim might seem counterintuitive since we usually associate the change of the text with the oral tradition, in ancient Indian culture oral transmission was practiced and passed on for centuries or millennia before the introduction of writing. Oral transmission in India was quite accurate as we can see from the transmission of Vedic texts. Buddhist monks modelled their practice on this ancient

Vedic culture. It is therefore through unaudited writing by individuals that changes are most likely to have occurred.

The second part of this study focuses on the stylistic features of the KP that elucidate more about the composition and transmission of this sūtra. I investigated the theory proposed by several scholars that Mahāyāna sūtras were composed and transmitted in written form. The dominant characteristics of Mahāyāna literature such as self-reference as a physical book that should be copied, preserved and worshipped; the change of narration style from the early

Buddhist texts; and the length and complexity of the narrative indicate the literariness of

Mahāyāna sūtras. However, even at first glance it can be noticed that stylistic features of

Mahāyāna sūtras are not completely different from those in early Buddhism. I find that there are

86

many oral stylistic features used in the composition of the KP, for instance formulaic expressions, repetition, and some miscellaneous features such as the waxing syllable principle, the arrangement of related subjects, and numerical order. These stylistic features are considered to be dominant features of oral literature which function as mnemonic devices.

If Mahāyāna sūtras were not composed and transmitted orally like their predecessors, why can we still find dominant oral stylistic features in this sūtra? The best explanation for this circumstance is that the oral/aural tradition was never discarded from Indian literary culture.

Collins has emphasized the complex interaction of the oral/aural tradition and the use of writing in the transmission of Buddhist texts. He argues that even though the Buddhist tradition accepted the use of writing fairly early as an important method to preserve their texts, the significance of the oral/aural tradition, as shown in the Pāli canon, was still emphasized continuously. This same social value would have been shared among all Buddhist societies.157 We then can assume, then, that although the KP might have been composed in written form, it was intended to be recited and read aloud for public preaching. Mahāyāna sūtras were never intended to be read privately as is done with literature nowadays. To borrow Pollock’s words, “[T]here is no category for literature as something read.”158 Manuscripts might have been used when they were preached in public and later became sacred objects for worshipping. The same action is still practiced even in the twenty-first century in Thailand where monks hold manuscripts in hand and read them aloud to the public. Buddhists often receive texts by listening, not reading individually. More importantly, oral recitation has authentication value. In Mahāyāna sūtras, it is regarded as a great way of merit-making. One would gain much merit from reciting even one stanza of a Mahāyāna

157 Collins 1992, 129. 158 Pollock 2006, 84. 87

sūtra. This indicates that oral recitation was still significantly valued in early Mahāyāna

Buddhism. To present their new tradition as an old and authentic one, Mahāyāna authors had to encourage their followers to recite their new texts orally so that they could claim the authority and authenticity of their texts.

Moreover, oral tradition in Indian culture was still very important even though we might think that writing was more efficient for textual preservation. Goody has suggested three advantages of the oral tradition in the early Indian culture. The first advantage concerns the scarcity of written texts. He states that “in a manuscript culture, copies of books are rare and expensive.” For this reason, “access is often more simply gained by listening rather than by reading.”159 The second reason concerns the process and the efficiency of oral recitation. He posits that “the process of reading aloud means that the pupil can ask questions and hence improve his opportunities for learning.”160 In other words, oral recitation means that students can interact with their teacher immediately if they do not understand something. The final reason which is the most important one concerns the security of the teacher’s status. He emphasizes that

“by retaining control over the process of transmission, we render our jobs more secure.”161 Even though Goody’s statements are not intended directly for Buddhist textual culture, I think that they can also be applied to explain the phenomena in Mahāyāna Buddhism. It can be noticed in many Mahāyāna sūtras that the emphasis is usually placed on Dharma-preachers (Skt. dharmabhāṇaka), including in the KP.162 I would argue that this might suggest that the Dharma-

159 Goody [1987] 1993, 118. 160 Goody [1987] 1993, 118. 161 Goody [1987] 1993, 119. 162 In section 160 of the KP, the significance of the Dharma-preacher is emphasized as follows: “When this discourse on the doctrine, the Ratnakūṭa, is expounded or taught or written or copied or exists in book form on a certain spot of earth, then that spot of earth becomes a shrine for the world together with its gods. And should one hear or take up or copy or comprehend this discourse on the doctrine from a certain preacher of the

88

preachers, who composed, preserved, and recited Mahāyāna sūtras, placed themselves at the center of Mahāyāna Buddhism. They, as Goody proposes, did not want to lose their job and their status as the main characters in Mahāyāna Buddhism.

Finally, the question about the authors of Mahāyāna sūtras can also be answered by considering the stylistic features of the composition of Mahāyāna sūtras. First of all, we can be quite sure that the radical change of stylistic features from early Buddhism to Mahāyāna

Buddhism might never have taken place. Mahāyāna authors preserved the archaic compositional styles which indicates that they were quite familiar with and conservative about the early

Buddhist tradition. They maintained what they learned and recited from early Buddhist texts.

The KP as well as other Mahāyāna texts shows a high level of familiarity on the part of its authors and audience with early Buddhist literature. They were monks who were educated in the traditional ordination lineage. Silk has commented that “it is nearly impossible to image

Mahāyāna sūtras to have been written by anyone other than such monks, rather more likely, communities of such monks.”163 In brief, the stylistic features of the KP seem to not support the theory that the Mahāyāna movement was a lay movement. It is more likely that authors of this sūtra would have been monks who were fairly familiar with the early Buddhist texts.

A final question still remains. Why did Buddhist monks accept and adopt the technology of writing earlier than any other religions in India? Would they not be more conservative in favor of received tradition? The belief that the orthodox Buddhist monastic should be conservative and should reject any type of new technology seems incorrect, since the monastic has always made use of technological advancement in the dissemination of their teaching. In my experience,

teaching, then he must generate such a respectful attitude toward that preacher of the teachings, just like, Kāśyapa, toward the Tathāgata. One who will reverence, respect, honor and worship a preacher of the teaching will, at the moment of his death, be able to see the Tathāgata.” 163 Silk 1994, 33. 89

Buddhist monks claim that as long as the Vinaya does not prohibit them from using technology, they have the right to use it, especially when it is useful for the spreading of their teaching. The application of technology to the dissemination of Buddhist scriptures can also be seen from the case of the invention of printing. The first book that was printed in China was not a royal decree nor a commercial record, but it was the Diamond Sūtra. Technology was first employed for the sake of the religion. This is also true in Western history as the first book that was published by

Johannes Gutenberg was the Bible. Religious societies do not refuse technology. They accept any approach that might facilitate their quest of the dissemination of their teachings. Likewise, when writing was invented and used in India as a means to help to preserve texts, Mahāyāna authors could have embraced it quickly as they might have anticipated that this would be the way to preserve their novel scriptures. Therefore, the composition and transmission of Mahāyāna texts could have relied mostly on the writing of texts.

However, there is not an absolute way to prove that the early Buddhist texts were composed and transmitted orally while Mahāyāna sūtras were composed and transmitted in written form. The study of the stylistic features of Mahāyāna sūtras is one of many factors to consider when speculating about what happened approximately two thousand years ago.

Correspondingly, the internal evidence in Mahāyāna texts that points in the same direction seems to support this. This study relies on only one particular text which, of course, cannot represent the history of the entirety of Mahāyāna literature. More research on stylistic features of other

Mahāyāna sūtras will shed brighter light on the composition and transmission of Mahāyāna

Buddhism.

90

BIBLIOGRAPHY

Allon, Mark. 1997a. Style and Function: A Study of the Dominant Stylistic Features of the Prose Portions of Pāli Canonical Sutta Texts and Their Mnemonic Function. Tokyo: The International Institute for Buddhist Studies.

——— 1997b. “The Oral Composition and Transmission of Early Buddhist Texts”. In P. Connolly and S. Hamilton, ed., Indian Insights: Buddhism, Brahmanism and Bhakti Papers from the Annual Spalding Symposium on Indian Religions (1989-1994). pp. 39- 61. London: Luzac Oriental.

Anālayo. 2007. “Oral Dimensions of Pāli Discourses: Pericopes, other Mnemonic Techniques and the Oral Performance Context.” In Canadian Journal of Buddhist Studies, 3: 5-34.

——— 2011. A Comparative Study of the Majjhima-Nikāya. 2 vols. Taipei: Dharma Drum.

——— 2014. “The Brahmajāla and the Early Buddhist Oral Tradition.” In Annual Report of the International Research Institute for Advanced Buddhology at Soka University for the Academeic Year 2013: 41-60.

——— 2016. “Āgama and Aṅga in the Early Buddhist Oral Tradition.” In Singaporean Journal of Buddhist Studies 3: 9-37.

Apple, James B. 2014. “The Phrase Dharmaparyāyo Hastagato in Mahāyāna Buddhist Literature: Rethinking the Cult of the Book in Middle Period Indian Mahāyāna Buddhism.” In Journal of the American Oriental Society 34.1: 25–50.

——— 2017. “The Old Tibetan Version of Kāśyapaparivarta Preserved in Fragments from Dunhuang (1).” In Annual Report of the International Research Institute for Advanced Buddhology at Soka University for the Academeic Year 2016: 205-230.

Bendall, Cecil, ed. 1902 (1897). Śikṣāsamuccaya: A Compendium of Buddhistic Teaching Compiled by Śāntideva Chiefly from Earlier Mahaāyaāna-Suūtras. Bibliotheca Buddhica 1. St. Petersburg. Rpt. Osnabrück: Biblio Verlag.

Bodhi, Bhikkhu, trans. 2012. The Numerical Discourses of the Buddha: A translation of the Aṅgutara Nikāya. Boston: Wisdom Publications.

Boucher, Daniel. 2001. “The Textual History of the Rāṣṭapālaparipṛcchā: Note on its Third- Century Chinese Translation.” In Annual Report of the International Research Institute for Advanced Buddhology at Soka University for the Academeic Year 2000: 93-116.

——— 2008. Bodhisattvas of the Forest and the Formation of the Mahāyāna: A Study and Translation of the Rāṣṭrapālaparipṛcchā-sūtrā. Honolulu: University of Hawaii Press.

91

Cole, Alan. 2005. Text as Father: Paternal Seduction in Early Mahāyāna Buddhist Literature. California: University of California Press.

Collins, Steven. 2005 (1990). “On the Very Idea of the Canon.” In Journal of 15: 89-126. Reprinted in Paul Williams, ed. 2005. Buddhism: Critical Concepts in Religious Studies volume I Buddhist Origins and the Early in South and Southeast Asia. pp. 72-95. New York: Routledge

——— 1992. “Notes on some Oral Aspect of Pāli Literature.” In Indo-Iranian Journal 35: 121- 135.

Conze, Edward. 1952. “The Composition of the Aṣṭasāhasrikā Prajñapāramitā.” In Bulletin of the School of Oriental and African Studies 14.2: 251-262.

Drewes, David. 2007. “Revisiting the Phrase “Sa Pṛthivīpradeśaś Caityabhūto Bhavet” and the Mahāyāna Cult of the Book.” In Indo-Iranian Journal 50.2: 101–143.

——— 2010. “Early Indian Mahāyāna Buddhism I: Recent Scholarship”. In Religion Compass 4.2: 55-65.

——— 2011. “Dharmabhāṇakas in Early Mahāyāna”. In Indo-Iranian Journal 54: 331-372.

——— 2015. “Oral Texts in Indian Mahāyāna”. In Indo-Iranian Journal 58: 117-141.

Falk, Harry and Seishi Karashima. 2012. “A First-Century Prajñāpāramitā Manuscript from Gandhāra—Parivarta 1 (Text from the Split Collection 1).” In Annual Report of the International Research Institute for Advanced Buddhology at Soka University of the Academic Year 2011, 19-61 + plates 5~7.

——— 2013. “A First-Century Prajñāpāramitā Manuscript from Gandhāra—Parivarta 5 (Texts from the Split Collection 2).” In Annual Report of the International Research Institute for Advanced Buddhology at Soka University of the Academic Year 2012: 97–169.

Forte, Antonino. 2002. “The South Indian Monk Bodhiruci (D. 727): Biographical Evidence.” In Antonino Forte and Federico Masini, eds., A Life Journey to the East: Sinological Studies in Memory of Giuliano Bertuccioli (1923–2001). pp. 77–116.Kyoto: Scuola Italiana di Studia sull’Asia Orientale.

Gethin, Rupert. 1992. “The Mātikās: Memorization, Mindfulness, and the List.” In Janet Gyatso, ed., In the Mirror of Memory: Reflections on Mindfulness and Remembrance in Indian and . pp. 149-172. Albany: State University of New York Press.

——— 2007. “What’s in a Repetition?: On Counting the Suttas of Saṃyutta-nikaya.” In The Journal of the Pali Text Society 29: 365-87.

92

Gombrich, Richard. 2005 (1988). “How the Mahāyāna Began.” In Journal of Pali and Buddhist Studies. pp. 29-46. Reprinted in Paul Williams, ed., 2005. Buddhism: Critical Concepts in Religious Studies Volume 3 The Origins and Nature of Mahāyāna Buddhism; Some Mahāyāna Religious Topics. New York: Routledge.

Goody, Jack. 1993. The Interface between the Written and the Oral. Cambridge: Cambridge University Press.

Gummer, Natalie D. 2012. “Listening to the Dharmabhāṇaka: The Buddhist Preacher in and of the Sūtra of Utmost Golden Radiance”. In Journal of the American Academy of Religion 80.1: 137-160.

Harrison, Paul. 1993. “The Earliest Chinese Translations of the Mahāyāna Buddhist Sūtras: Some Notes on the Works of Lokakṣema.” In Buddhist Studies Review 10.2: 135–177.

——— 1995. “Searching for the Origins of the Mahāyāna: What Are We Looking For?” In The Eastern Buddhist New Series 28.1: 48-69.

——— 2003. “Mediums and Messages: Reflections on the Production of Mahāyāna Sūtras.” In Eastern Buddhist 35.1/2: 115–151.

Hartmann, Jens-Uwe. 2009. “From Words to Books: Indian Buddhist Manuscripts in the First Millenium CE”. In Stephen C. Berkwitz, Juliane Schober and Claudia Brown, ed., Buddhist Manuscript Cultures: Knowledge, Ritual, and Art. pp. 95-105. London: Routledge. von Hinüber, Oskar. 1996. A Handbook of Pāli Literature. Berlin, New York: Walter de Gruyter.

Hirakawa, Akira, 1963. “The Rise of Mahāyāna Buddhism and Its Relationship to the Worship of .” In , trans., Memoirs of the Research Department of the Toyo Bunko 22: 57–106. de Jong, J.W. 1977. “Sanskrit Fragments of the Kāśyapaparivarta.” In H. Härtel, ed., Beiträge zur Indienforschung Ernst Waldschmidt zum 80. Geburtstag gewidmet. pp. 247–255. Berlin: Museum für Indische Kunst.

Karashima, Seishi. 2002. “Some Features of the Language of the Kāśyapaparivarta.” In Annual Report of the International Research Institute for Advanced Buddhology at Soka University for the Academic Year 2001: 43–66.

——— 2010. A Glossary of Lokakṣema’s Translation of Aṣṭasāhasrikā Prajñāpārmitā 道行般 若經詞典 (Bibliotheca Philologica et Philosophica Buddhica, XI). Tokyo: The International Research Institute for Advanced Buddhology, Soka University.

93

——— 2015. “Who Composed the Mahāyāna Scriptures?: The Mahāsaṃghikas and Vaipulya Scriptures.” In Annual Report of the International Research Institute for Advanced Buddhology at Soka University for the Academic Year 2014: 113-162.

Lenz, Timothy. 2010. Gandhāran Avadānas: British Library Kharoṣṭhī Fragments 1–3 and 21 and Supplementary Fragments A–C. Seattle and London: University of Washington Press.

Lopez, Donal S. Jr. 1995. “Authority and Orality in the Mahāyāna”. In Numen 42.1: 21-47.

Maggi. 2015. “A folio of Ratnakūṭa (Kāśyapaparivarta) in Khotanese.” In Dharma Drum Journal of Buddhist Studies 17: 101-143.

Marino, Joseph. 2017. “Metaphor and Pedagogy in Early Buddhist Literature: An Edition and Study of Two Sūtras from the Senior Collection of Gāndhārī Manuscripts.” Ph.D. Dissertation. University of Washington.

Martini, Giuliana. 2011. “A Large Question in a Small Place: The Transmission of the Ratnakūṭa (Kāśyapaparivarta) in Khotan.” In Annual Report of the International Research Institute for Advanced Buddhology at Soka University 14: 135-183.

McMahan, D. 1998. “Orality, Writing, and Authority in South Asian Buddhism: Visionary Literature and the Struggle for Legitimacy in the Mahāyāna.” In History of Religions 37.3: 249–274.

Morris, R (ed.). 1976 (1888). Aṅguttara-Nikāya of the Sutta-Piṭaka, Part II. Catukkanipāta. London: Pali Text Society.

Nattier, J. 2000. “The Realm of Akṣobhya: A Missing Piece in the History of Buddhism.” In Journal of the International Association of Buddhis Studies 3.1: 71-102.

——— 2003. A Few Good Men: The Bodhisattva Path According to the Inquiry of Ugra (Ugraparipṛcchā). Honolulu: University of Hawaii Press.

——— 2008. A Guide to the Earliest Chinese Buddhist Translations: Texts from the Eastern Han and Three Kingdoms Periods. Bibliotheca Philologica et Philosophica Buddhica, X. Tokyo: The International Research Institute for Advanced Buddhology, Soka University.

Norman, K.R. 1983. Pāli Literature, Including the Canonical Literature in Prakrit and Sanskrit of all the Hīnayāna . Wiesbaden: Otto Harrassowitz.

Osto, Douglas. 2015. “Orality, Authority, and Conservatism in the Prajñāpāramitā Sūtras”. In Brian Black and Laurie Patton, ed., Dialogues in Early South Asian Religions: Hindu, Buddhist and Jain Traditions. pp. 115-135. London: Routledge.

94

Pagel, Ulrich. 1992. “The Bodhisattvapiṭaka: Its Doctrines, Practices, and their Position in Mahāyāna Literature.” Ph.D. Dissertation. University of London.

Paranavitana, S. 1933. “Indikatusaya Copper Plaques”. In EZ 3.20: 199-212.

Pāsādika, Bhikkhu. 1980. “The Kāśyapaparivarta (`Od-srung-gi le`u)-Prolegomena.” In The Tibetan Journal 5.4: 48-58.

——— 1991. “Bibliographical remarks bearing on the Kāśyapaparivarta.” In Buddhist Studies Review 8.1-2: 59-70.

——— 2015. The Kāśyapaparivarta. New Delhi: Aditya Prakashan.

Pedersen, K. Priscilla. 1980. “Notes on the Ratnakūṭa Collection.” In Journal of the International Association of Buddhist Studies 3.2: 60–66.

Pollock, Sheldon. 2006. The Language of the Gods in the World of Men: Sanskrit, Culture, and Power in Premodern India. Berkeley: University of California Press.

Rocher, Ludo. 1994. “Orality and Textuality in the Indian Context.” In Sino-Platonic Papers 49: 1-28.

Ruegg, David S. 2004. “Aspects of the Investigation of the (earlier) Indian Mahāyāna.” In Journal of the International Association of Buddhist Studies 27.1: 3-62.

Salomon, Richard. 1995. “On the Origin of the Early Indian Scripts.” In Journal of American Oriental Society 115.2: 271-279.

——— 2017. “On the Evolution of Written Āgama Collections in Northern Buddhist Traditions.” In Dhammadinnā, ed., Research on the Madhyama-āgama, pp. 239–268. Taipei, Taiwan: Dharma Drum Publishing Corporation.

——— 2018. The Buddhist Literature of Ancient Gandhāra: An Introduction with Selected Translations. Maine: Wisdom Publication.

Schopen, Gregory. 1975. “The Phrase ‘Sa Pṛthivīpradeśaś Caityabhūto Bhavet’ in the Vajracchedikā: Notes on the Cult of the Book in Mahāyāna.” In Indo-Iranian Journal 17.3/4: 147– 181.

——— 2009. “On the Absence of Urtexts and Otiose Ācāryas: Buildings, Books, and Lay Buddhist Ritual at Gilgit”. In Gérard Colas and Gerdi Gerscheimer, ed., Écrire at transmettre en Inde Classique. Études Thématiques 23. pp. 189-219. Paris: École françaises d’Extrême-Orient.

95

Silk, Jonathan A. 1994. “The origins and Early History of the Mahāratnakūṭa Tradition of Mahāyāna Buddhism: With a Study of the Ratnarāśisūtra and Related Materials.” Ph.D. Dissertation, The University of Michigan.

——— 2002. “What, If Anything, Is Mahāyāna Buddhism?: Problem of Definitions and Classifications.” In Numen 49.4: 355-405.

——— 2009a. “Remarks on the Kāśyapaparivarta Commentary.” In M. Straube et al., ed., Pāsādikadānaṁ: Festschrift für Bhikkhu Pāsādika. pp. 381-397. Marburg: Indica et Tibetica Verlag.

——— 2009b. “The Nature of Verses of the Kāśyapaparivarta.” In Carol Altman, Timothy J.Lenz, and Jason Neelis, ed., Bulletin of the Asia Institute: Evo Ṣuyadi Essays in Honor of Richard Salomon’s 65th Birthday. pp. 181-190.

——— 2010. “Test Sailing the Ship of the Teachings: Hesitant Notes on Kāśyapaparivarta §§153–154.” In From Turfan to Ajanta: Festschrift for Dieter Schlingloff on the Occasion of his Eightieth Birthday. pp. 897–924. Bhairava: International Research Institute.

Skilling, Peter. 2009. “Redaction, Recitation, and Writing: Transmission of the Buddha’s Teaching in India in the Early Period”. In Stephen C. Berkwitz, Juliane Schober and Claudia Brown, ed., Buddhist Manuscript Cultures: Knowledge, Ritual, and Art. pp. 53– 75. London: Routledge,

——— 2013. “Vaidalya, Mahāyāna, and Bodhisatva in India: An Essay towards Historical Understanding.” In Bhikkhu Nyanatusita himi, ed., The Bodhisattva Ideal: Essays on the Emergence of the . pp. 69-162. Kandy: the Buddhist Publication Society.

Söhnen, Renate. 1995. “On the Concept and Presentation of Yamaka in Early Indian Poetic Theory”. In Bulletin of the School of Oriental and African Studies, University of London 58.3: 495-520. von Staël-Holstein, A. (Baron). 1926. The Kāçyapaparivarta: A Mahāyānasūtra of the Ratnakūṭa Class, edited in the original Sanskrit, in Tibetan and in Chinese. Shanghai: Shangwu Yinshuguan.

Strauch, Ingo. 2014. “Looking into Water-pots and over a Buddhist Scribe’s Shoulder – On the Deposition and the Use of Manuscripts in Early Buddhism.” In ASIA 88.3: 797-830.

Vorobyova-Desyatovskaya, Margarita I. in collaboration with Karashima Seishi and Kudo Noriyuki. 2002. The Kāśyapaparivarta: Romanized Text and Facsimiles. Tokyo: The International Research Institute for Advanced Buddhology, Soka University.

Weller, F. 1933/1935. Index to the Indian Text of the Kāśyapaparivarta. 2 volumes. Cambridge, Mass. 96

——— 1961a. "Passive ausdrucksweisen im mongolischen texte des Kāśyapaparivartasūtra." In Wissenchaftliche Zeitschrift der Karl-Marx-Universität (Leipzig) 10.4: 563-602.

———1961b. "Qayan und qan in der mongolischen übersetzung des Kaśyapaparivarta." In MOF 8: 218-228.

——— 1962. Zum Kāśyapaparivarta. Berlin: Abhandlungen der Sachsischen Akademie der Wissenschaften zu Leipzig, ph.-hist, Klasse. Bd. 54, Heft 2.

———1987 (1963). "Betrachtungen über einen Ratnakūṭa-text." In W. Rau, ed., Forscjimgem und Fortschritt, 369-374. Reprinted in Friedrich Weller, Kleine Schriften. pp. 537-542. Stuttgart

——— 1964. "Kāśyapaparivarta nach der Tjin-Übersetzung verdeutscht." In Wissenschaftliche Zeitschrift der Karl-Marx-Universität 13: 771-804.

——— 1965. "Buyu und bolai im mongolischen texte des Kāśyapaparivarta." In CAJ 10: 3-43.

——— 1966a. "Kāśyapaparivarta nach der Djin-Fassung verdeutscht." In MOF 12: 379-462.

——— 1966b. “Die Sung-Fassung des Kāśyapaparivarta Versuch einer Verdeutschung.” In Monumenta Serica 25: 207-361.

———1987 (1968-69). "Kāśyapaparivarta nach der Han-Fassung verdeutscht." In Buddhist Yearly. pp. 57-221. Reprinted in W. Rau, ed., Friedrich Weller: Kleine Schriften: 1035- 1459.

Wynn, Alexander. 2004. “The Oral Transmission of the Early Buddhist Literature.” In Journal of the International Association of Buddhist Studies 27.1: 7-128.

Zimmermann, Michael. 1999. “The Tathāgatagarbhasūtra: Its Basic Structure and Relation to the Lotus Sūtra”. In Annual Report of the International Research Institute for Advanced Buddhology at Soka University for the Academic Year 1998: 143-168.

97