The Stem Array
Introduction

The Stem Array (Gaṇḍa­vyūha) is a unique sūtra in that most of its narrative takes place in South India, far from the presence of the Buddha. It follows the journey of the young Sudhana from teacher to teacher, or kalyāṇamitra (literally “good friend”), beginning with his meeting Mañjuśrī when that bodhisattva came to South India. Another unique characteristic is that Sudhana’s teachers include children, non-Buddhists, a courtesan, merchants, and so on, among them a number of women. His teachers are both humans and deities, including eight night goddesses around the Bodhi tree and the forest goddess of Lumbinī, the birthplace of the Buddha. These teachers are often described as having received teachings from numerous other buddhas. For example, the bhikṣu Sāgara­megha describes how he received, from a buddha who appeared out of the ocean, teachings that would take more than a kalpa to write out. The kalyāṇamitras are described as having realizations and miraculous powers that test the limits of the imagination.

The Gaṇḍa­vyūha forms the forty-fifth and final chapter of the Buddhāvataṃsaka (A Multitude of Buddhas) Sūtra, where it is called a “chapter” rather than a “sūtra.” According to the Degé colophon, the previous forty-four chapters form six sections, or sūtras, of the Avataṃsaka, with the Gaṇḍa­vyūha as the seventh sūtra.1 In his sixteenth-century survey of the major sūtras, Pekar Zangpo (pad dkar bzang po) divides the first group of chapters into two, so that the Gaṇḍa­vyūha is the eighth section of the Buddhāvataṃsaka.2 The Gaṇḍa­vyūha is one of the four sections that consist of a single sūtra, but it is by far the longest sūtra or chapter, comprising about a third of the Avataṃsaka Sūtra.

In the Buddhāvataṃsaka Sūtra, the Buddha Śākyamuni never speaks: all the teachings in the forty-five chapters of the Avataṃsaka are given by others. In the first forty-four chapters or sūtras this is done in the Buddha’s presence. The Gaṇḍa­vyūha is unique in that most of this lengthy chapter takes place far from his presence, with other buddhas being presented as the sources of teachings received by the kalyāṇamitras whom Sudhana meets. However, the previous chapters of the Avataṃsaka have already presented the view that various buddhas are manifestations of the Buddha Vairocana, and it is by the name Vairocana that Śākyamuni is referred to in this sūtra.3

The previous forty-four chapters of the Avataṃsaka Sūtra take place during the two weeks after the Buddha’s enlightenment, at which time he sits in silence under the Bodhi tree yet is simultaneously present, still in silent meditation, in other locations throughout our universe: the Trāyastriṃśa paradise of Indra on the summit of Sumeru, the Yāma and Tuṣita paradises high above Sumeru, and the highest paradise in the realm of desire‍—the Para­nirmita­vaśa­vartin paradise. Bodhisattvas congregate around him, inspired by his presence to give such teachings as the Daśa­bhūmika Sūtra (Ten Bhūmi Sūtra),4 which is taught by the bodhisattva Vajragarbha in the Para­nirmita­vaśa­vartin paradise. The Daśa­bhūmika Sūtra had a great influence on the development of Buddhism, eclipsing the previous seven bhūmis of the Prajñāpāramitā (Perfection of Wisdom) sūtras.

The Gaṇḍa­vyūha, on the other hand, begins with the Buddha in silent meditation in his Jetavana Monastery in Śrāvastī, where he spent most of his summer retreats. Human pupils are gathered around him along with a multitude of bodhisattvas that his human pupils are not advanced enough to perceive. While the Buddha sits silently in meditation, the bodhisattva Samanta­bhadra gives a teaching to the assembled bodhisattvas. The bodhisattva Mañjuśrī leaves the assembly for South India, and, rather than continuing to describe events and teachings in the presence of the Buddha, the sūtra follows Mañjuśrī to South India, where he meets Sudhana, and the narrative then follows Sudhana for the rest of the long sūtra. Although the beginning of the sūtra is set at a time later than that of the Buddha’s enlightenment, further on, in the night-goddess chapters, the Buddha is depicted as being present under the Bodhi tree. There are other temporal anomalies: the bodhisattva Maitreya, in the chapter where Sudhana meets him, is portrayed as being on earth and not yet passed away to be reborn in Tuṣita, even though he is said in the Māyādevī chapter, as is generally said in other Buddhist sources, to be already present in Tuṣita. Māyādevī, the Buddha’s mother, appears to Sudhana in Kapilavastu, the Buddha’s hometown, even though she is traditionally said to have passed away shortly after the Buddha’s birth and been reborn as a male deity in the Trāyastriṃśa paradise.

The sūtra primarily describes (in successive long compounds in Sanskrit) both the inner qualities and the external displays of miraculous powers that have been attained by the various kalyāṇamitras whom Sudhana meets. It concludes with the bodhisattva Samanta­bhadra composing the Samanta­bhadra­caryā­praṇidhāna (“The Prayer for Completely Good Conduct”), which is regularly recited by contemporary Tibetan Buddhists.

Mahāyāna sūtras first appeared through the medium of revelations after the tradition of written sūtras had developed. These sūtras appeared in various Buddhist traditions and in various locations.5 Mahāyāna sources show doctrinal similarities to the Mahāsāṃghika tradition.6 The Mahāsāṃghika was especially prevalent in southwest India, and a substantial number of Mahāyāna sūtras have indications of a South Indian provenance, with passing references to South Indian music, or, to take the Samādhirāja Sūtra7 as just one example, to a prominent South Indian personage, in this case the ṛṣi Ananta. Similarly, much of the Gaṇḍa­vyūha Sūtra takes place in South India, depicting teachers who appear to operate independently of Buddhist communities in the north. Douglas Osto echoes Qobad Afshar in suggesting that the site of Dhanyākara referenced in the Gaṇḍa­vyūha is in fact Dhānyakaṭaka/Dharaṇi­koṭa, an ancient city that existed on the banks of the Kṛṣṇa River in the southern region of Andhra. This place, Afshar and Osto believe, was probably where the Gaṇḍa­vyūha was first composed.8

In terms of its language, the Sanskrit of the Gaṇḍa­vyūha Sūtra has numerous nonclassical Buddhist Hybrid Sanskrit (BHS) features and vocabulary. This is especially true of the verses, which are less prone to revision to Classical Sanskrit than the prose. It is frequently the case that the verses in a sūtra are older than the prose that accompanies them, or they at least retain the original form of the language in which the sūtra was composed. In the Gaṇḍa­vyūha, they favor the -u ending for the nominative case, where it would be -a in Sanskrit, -e in the ancient northeastern dialect, and -o in that of the northwest (and its continuation in Pali). The difference between these two kinds of Sanskrit is not evident in the Tibetan or the English translations.

Concerning the relative chronology of the Gaṇḍa­vyūha, chapter 41 makes a clear reference to the Satyaka Sūtra (formally known as The Teaching of the Miraculous Manifestation of the Range of Methods in the Field of Activity of the Bodhisattvas),9 which describes a Jain master by the name of Satyaka, who advises a king on the polity of rulership10 and is eventually revealed by the Buddha to be a bodhisattva who takes on various forms in order to benefit beings.11 The Satyaka Sūtra briefly presents the single-yāna view that was expounded in The Lotus Sūtra,12 but it goes further by stating that all religious traditions in India occur through the blessing of the Buddha and are therefore included within the single yāna.13 This view of the Satyaka Sūtra is crucial for understanding one of the surprising elements in the Gaṇḍa­vyūha Sūtra‍—some of Sudhana’s teachers do not appear to be following a Buddhist path. There is the ṛṣi Bhīṣmottara­nirghoṣa, who is teaching young brahmins; and Jayoṣmāyatana, who is following the non-Buddhist ascetic practice of “the five heats” (sitting amid four fires under the noonday sun); and there is even Mahādeva, also known as Śiva. Therefore, in terms of the succession of sūtras, it would appear that the Gaṇḍa­vyūha postdates both The Lotus Sūtra (though not necessarily its later chapters) and the Satyaka Sūtra.

The depiction of the bodhisattva Avalokiteśvara in the Gaṇḍa­vyūha is surprising in that he is presented as an apparently human guru living on the Potalaka Mountain in South India, with no mention of his association with Amitābha’s pure realm of Sukhāvatī, where earlier sūtras locate him. The Potalaka Mountain was an important place of pilgrimage for both the Buddhists and Śaivites of South India, as the abode of both Śiva, who was known as Lokeśvara (Lord of the World), and Avalokiteśvara. This earthly abode of Avalokiteśvara, in Tibet known as the Potala Mountain, would become prominent in Tibetan and Chinese Buddhism, as evidenced by the Potala Palace in Lhasa and Mount Putuo Island in China.

The Gaṇḍa­vyūha Sūtra first existed in India as an independent sūtra and still exists as an independent sūtra in Sanskrit manuscripts. The successive Chinese translations reveal a gradual growth in the contents of the sūtra, with the addition of more teachers in the Indian version before its eventual translation into Tibetan. Even so, the number of kalyāṇamitras met by Sudhana is still smaller than the number that Maitreya, toward the end of the sūtra, proclaims that Sudhana has met on his journeys‍—110.

There was an Indian version longer than the one that was translated into Tibetan, though no Sanskrit manuscript of this version has survived. It is known only from the version sent to China by the king of Orissa, who gave a copy to the Chinese emperor in 795.14 This version was translated by the Kashmiri monk Prajñā in 798, two or three decades before the Tibetan translation was made. The Chinese translations also indicate that the Gaṇḍa­vyūha had a different title in the seventh century, which can be reconstructed from the Chinese 入法界品 (Ru fa jie pin) as Dharma­dhātu­praveśana (Entry into the Realm of the Dharma).

Earlier versions concluded with the Maitreya chapter. The Mañjuśrī and Samanta­bhadra chapters were added subsequently, and finally the sixty-two-verse “Prayer for Completely Good Conduct,” which has continued to exist as an independent text, was added as the sūtra’s conclusion. This prayer was translated into Chinese in a forty-four-verse version by Buddhabhadra in the early fifth century. Amoghavajra’s (705–74) eighth-century Chinese translation of the prayer has sixty-two verses because of the addition of fifteen verses on Amitābha. This longer version appeared as the conclusion of the Buddhāvataṃsaka in the translation into Chinese made in 798 and in the early ninth-century translation into Tibetan. “The Prayer for Completely Good Conduct,” particularly its first twelve verses, is regularly recited in Tibetan Buddhism, and it also exists independently in the Kangyur with an additional concluding verse.15

There are no surviving Sanskrit manuscripts of the Gaṇḍa­vyūha from the first millennium, but there is a complete Sanskrit text that dates to 1166 ᴄᴇ, three hundred years later than the Tibetan translation. It consists of 289 palm-leaf pages and was sent from Nepal to the Royal Asiatic Society in London by Brian Houghton Hodgson (1800–1894) in the early nineteenth century. Cataloged as Hodgson 2 (A), this is the earliest extant Sanskrit manuscript of the Gaṇḍa­vyūha.16 The Sanskrit Buddhist tradition has continued in Nepal, where the Gaṇḍa­vyūha remains one of the nine central works of Newar Buddhism.17

The Gaṇḍa­vyūha Sūtra was first translated into Chinese as an independent text by a monk named Shengjian sometime between 388 and 408 ᴄᴇ.18 In that translation there are only twenty-five kalyāṇamitras. Compared to the extant Sanskrit version, the first nine kalyāṇamitras after Mañjuśrī and those after the thirty-fourth are not present.19 Therefore, it may represent an unfinished translation or an earlier form of the sūtra, or both.

Not long afterward, in 420, Buddhabhadra (359–429 ᴄᴇ), an Indian monk who had migrated to China, translated with his team the entire Buddhāvataṃsaka Sūtra, which at that time was composed of thirty-four chapters with the Gaṇḍa­vyūha Sūtra as the final chapter.20 Buddhabhadra’s translation contains not only additional chapters when compared to Shengjian’s translation, but also additional verses and passages within the chapters.

Interestingly, the Indian monk Paramārtha (499–569 ᴄᴇ), who flourished not long after Buddhabhadra’s time, refers to the Buddhāvataṃsaka as a Bodhisattva­piṭaka (“basket” or “collected teachings for bodhisattvas”). This is echoed by a copy of this sūtra found in the Mogao Caves at Dunhuang, which bears the title Bodhisattva­piṭaka Buddhāvataṃsaka.21 While the presently available Sanskrit does not give the title Bodhisattva­piṭaka, the Tibetan colophons feature this designation, though the sense of it shifts depending on the given colophon’s grammar,22 which may suggest that there was some confusion among Tibetan scholars about the term’s significance.

The Buddhāvataṃsaka had grown even further in size by the time of its translation into Chinese under the direction of the Khotanese Śikṣānanda (于闐國實叉難陀, 652–710 ᴄᴇ). This was made between 695 and 699 ᴄᴇ23 and had an additional five chapters, with the Gaṇḍa­vyūha Sūtra still in final place as the thirty-ninth chapter. The translation of the Gaṇḍa­vyūha in this version varies little from that of Buddhabhadra, with a few exceptions: the final two verses of the Veṣṭhila section have been added, Avalokiteśvara’s mountain is now named Potalaka rather than Prabha, and a short verse greeting appears in the final section.24

Śikṣānanda’s version of the Buddhāvataṃsaka became the basis for the Huayan school of Chinese Buddhism, huayan being the Chinese translation of avataṃsaka as “flower garland,” and it is Śikṣānanda’s version that was translated by Thomas Cleary into English as The Flower Ornament Scripture.25 In Śikṣānanda’s Chinese, the title of the Gaṇḍa­vyūha Sūtra was 入法界品 (Ru fa jie pin), which could have come from the Sanskrit Dharma­dhātu­praveśana­parivarta and in English could be translated as The Chapter on Entering the Realm of the Dharma. Thomas Cleary translated this as Entry into the Realm of Reality, while Douglas Osto translated it as Entry into the Realm of Dharma, and Patrick Carré as l’Entrée dans la dimension absolue (Entry into the Absolute Dimension).” The term dharma­dhātu­praveśana does occur in the text, as for example in verse 44 of chapter 1.

There is a longer version of the Gaṇḍa­vyūha Sūtra that was translated into Chinese as an independent sūtra in 798 by the Kashmiri monk Prajñā. This was the first among the Chinese translations to include “The Prayer for Completely Good Conduct.”26 Prajñā’s translation was based on a Sanskrit manuscript that the king of Orissa sent as a gift to the Chinese emperor, who received it in 795.27 The additions in Prajñā’s version are not found in any surviving Sanskrit edition, nor are they found in the early ninth-century Tibetan translation made just a few decades after Prajñā’s translation. In Prajñā’s translation the Gaṇḍa­vyūha Sūtra is called The Vow Concerning the Course of Conduct of Samanta­bhadra and the Entry into the Range of Inconceivable Liberation (入不思議解脫境界普賢行願品, Ru bu si yi jie tuo jing jie pu xian xing yuan pin), which could be reconstructed in Sanskrit as Acintya­vimokṣa­gocara­praveśana­samanta­bhadra­caryā­pranidhāna).

As mentioned, the Buddhāvataṃsaka Sūtra became the basis for the Chinese Huayan school of Buddhism. Li Tongxuan (635–730 ᴄᴇ) was particularly influential in the spread of this tradition, and he wrote a commentary (華嚴論, Huayan lun) on the Buddhāvataṃsaka Sūtra, in which he identifies a deeper structure and meaning in the Gaṇḍa­vyūha’s narrative, and that part of Li Tongxuan’s commentary has been translated into French by Patrick Carré. Li also composed a summary of that commentary and a chapter-by-chapter synopsis of the Buddhāvataṃsaka Sūtra itself.28 This tradition spread to Korea and Japan, where it became, respectively, the Hwaeom and Kegon schools of Buddhism.

The Gaṇḍa­vyūha Sūtra, and in particular its Maitreya chapter, was an inspiration for what is arguably the greatest Buddhist monument ever built: Borobudur in Indonesia. This was built in the ninth century by the Buddhist kings of the Śailendra dynasty in Java. The massive structure has a series of encircling terraces that hold 504 statues and 2672 carved panels. The upper terraces, the third and fourth galleries, are entirely dedicated to the Gaṇḍa­vyūha, with 460 panels illustrating the sūtra.29 Of these, it appears that 218 panels are dedicated to the Maitreya chapter, and from among those twenty are dedicated to the description of Maitreya’s kūṭāgāra and thirty-five to the various manifestations of Maitreya.30 In fact, 334 panels are dedicated to the conclusion of the Gaṇḍa­vyūha: Maitreya, the brief return to Mañjuśrī, Samanta­bhadra, and the “Prayer of Good Conduct.” This reflects the importance of the Gaṇḍa­vyūha and particularly of Maitreya’s kūṭāgāra‍—an edifice that embodied enlightenment‍—for the constructors of Borobudur, who were also trying to create an edifice that embodied enlightenment.

Although the number of accounts of Sudhana meeting kalyāṇamitras grew in succeeding recensions, they did not reach the number given in the sūtra itself in chapter 54, where Maitreya states that Sudhana, following his initial meeting with Mañjuśrī, has visited 110 kalyāṇamitras. Borobudur, possibly to accord with that statement, does have that number of panels dedicated to the illustration of that part of the Gaṇḍa­vyūha, but even so it does not represent 110 kalyāṇamitras but instead repeats the illustration of certain visits, in addition to portraying Sudhana traveling and depicting incidents in past lives of the kalyāṇamitras.31

The Buddhāvataṃsaka as translated into Tibetan is composed of forty-five chapters and 115 fascicles, with the Gaṇḍa­vyūha Sūtra still retaining its position as the last of the chapters. According to the Degé Kangyur, the entire Buddhāvataṃsaka Sūtra, including the Gaṇḍa­vyūha, was translated into Tibetan by Yeshé Dé, Jinamitra, and Surendrabodhi, which would have been during the reign of King Senalek (r. ca. 800/804–15) or King Ralpachen (r. 815–36). Yeshé Dé and Jinamitra had been working at Samyé Monastery since the reign of King Trisong Detsen (r. 742–98), but Surendrabodhi appears to have come to Tibet after Senalek, Trisong Detsen’s youngest son, became king. Senalek was in turn succeeded by his son Ralpachen, the end of whose reign also saw the end of state-sponsored translation. Therefore, this translation appears to have been made sometime between 800 and 836, but it may incorporate even earlier translation work, particularly in the case of the Gaṇḍa­vyūha.

By the time of the sūtra’s translation into Tibetan at the beginning of the ninth century, the number of chapters in the Avataṃsaka Sūtra had increased from thirty-nine to forty-five. But this is primarily because of the division in Tibetan of what is the Chinese chapter 5 into chapters 5 through 9, with some additional material, and the additional chapters 11 and 32. This Tibetan translation provides the earliest indication of when the forty-fifth chapter was named Gaṇḍa­vyūha.

According to Tashi Wangchuk, who wrote the colophon to the eighteenth-century Degé edition, and also according to the historian Ngorchen Könchok Lhundrup (1497–1557), the Buddhāvataṃsaka is composed of seven sūtras or sections, while Pekar Zangpo divides the first of these into two, making eight sections.

Whereas the Chinese version of the Buddhāvataṃsaka retained the traditional beginning of the Gaṇḍa­vyūha Sūtra as an independent sūtra, commencing with “Thus did I hear…” and so on, the version translated into Tibetan omits it, as do the surviving Sanskrit versions.

The Gaṇḍa­vyūha Sūtra, which in Tibetan is interpreted to mean, ambiguously, The Stem Array Sūtra, is the forty-fifth and last chapter in the Tibetan version of the Buddhāvataṃsaka Sūtra, which is made up of four volumes as found in the Degé Kangyur. This chapter is by far the longest, beginning halfway through the third volume and occupying the entire fourth volume of the Buddhāvataṃsaka. It is composed of 72 of the 115 fascicles that make up the entire sūtra, beginning with fascicle 44 (the twenty-fourth in volume Ga). Fascicles refer to the bundles of pages in the original Sanskrit manuscripts, usually joined up through two holes in the center of each page. In this translation the beginning of a fascicle is simply marked with the letter B (from the Tibetan for fascicle: bam po) and a number, for example, [B24].

The quality of the Tibetan translation differs from the rest of the Buddhāvataṃsaka, either because of scribal corruption or choices of translation. The Tibetan has peculiarities not shared with other parts of the Avataṃsaka. For example, it retains the archaic spelling of myi and myed instead of mi and med. The translation is less reliable than usual, as it contains frequent, possibly inadvertent omissions and misspellings that must have occurred early in the scribal transmission, as some of these errors in the sūtra are found in all Kangyurs.

It also exhibits a certain idiosyncrasy of translation, in that the terms may not match what was established in the Mahāvyutpatti (bye brag tu rtogs par byed chen po) and Madhya­vyutpatti (sgra sbyor bam po gnyis pa), the early ninth-century Sanskrit–Tibetan dictionaries produced within the same state-sponsored translation project responsible for the translation of this sūtra. For example, vyūha is regularly translated as rgyan (“adornment”) instead of bkod pa (“array”), even in the title of the sūtra, in spite of its being generally known in Tibetan as sdong po bkod pa. These and other variations of the title are touched upon in the Avataṃsaka’s editorial colophon in the Degé Kangyur (c.­11).

According to that colophon, this edition of the Avataṃsaka was prepared in 1722. This was eight years before the eighth Tai Situpa Chökyi Jungné (1700–1774) began his work as chief editor of the Degé Kangyur. He states in his account of the creation of the Kangyur: “I began in the Iron Dog year,”32 which was 1730, and the carving and printing of the Degé woodblocks did not begin until 1737 and was completed in 1744. It was nevertheless done under the command of the Degé King Tenpa Tsering (1678–1738), as was the entire Degé Kangyur, and therefore may be an earlier edition that Situ incorporated into his edition of the Kangyur. The colophon also states that it was based on the Lithang Kangyur, also known as the Jangsa Tham Kangyur. The creation of this Kangyur took five years, from 1609 until 1614. The Lithang was the second printing of the Kangyur, which otherwise only existed in manuscript form. The first printing was the Yongle in 1411.

The colophons of the Narthang, Lhasa, and Stok Palace Kangyurs ascribe the translation of the Buddhāvataṃsaka to a Vairocana­rakṣita: “It was translated and revised by the chief editor Lotsawa Vairocana­rakṣita.” The great Sakya master Ngorchen Könchok Lhundrup also states that the translator was Vairocana­rakṣita.

In the editor’s colophon in the Degé it states, “It is taught that Surendrabodhi and Vairocana­rakṣita became principal editors for a Chinese translation.” As the Indian master Surendrabodhi came to Tibet during the reign of King Senalek (ca. 800/804–15), the identity of this Vairocana­rakṣita is a mystery, as he could not be the eleventh-century translator Vairocana­rakṣita, nor could he be the eighth-century Vairocana. Tashi Wangchuk adds, “I have not seen any other text or history of a translation made by any lotsawa or paṇḍita other than those listed in the colophon to this translation into Tibetan,” thus apparently rejecting the attribution of Vairocana­rakṣita as the translator.

The Gaṇḍa­vyūha Sūtra was translated into German from Buddhabhadra’s Chinese version by Dōi Torakazu as Das Kegon Sutra, Das Buch vom Eintreten in den Kosmos der Wahrheit in 1978.

The entire Avataṃsaka Sūtra has been translated from the Chinese by Thomas Cleary and published in 1993 as The Flower Ornament Scripture. The Gaṇḍa­vyūha Sūtra is therefore included as the final chapter, chapter 39, under the title “Entry into the Realm of Reality.”

The Śikṣānanda version has been translated from the Chinese into French by Patrick Carré. There is as yet no translation of the longest Chinese version, which was translated by the Kashmiri Prajñā in 798.

There have been partial unpublished translations from the Sanskrit by Mark Allen Ehman in 1977 and Yuko Ijiri in 2005.

Douglas Osto has translated the first part of chapter 1 and chapters 3, 54, and 55 from the Sanskrit of the Gaṇḍa­vyūha, with its title given as The Supreme Array. They are available to read on his website.33 He has also included excerpts from other chapters of the sūtra in his book Power, Wealth and Women in Indian Mahāyāna Buddhism: The Gaṇḍa­vyūha-sūtra.

As mentioned above, the sūtra’s title in Chinese translations differs from that in Tibetan, and it was evidently known by other names in earlier centuries. By the ninth century, however, it was known by this obscure title Gaṇḍa­vyūha. This translation follows the ambiguous meaning assigned to it by the early ninth-century translators into Tibetan.

There are two versions of the Tibetan title. In the Kangyur, the title is only mentioned in the colophon, where it is given as sdong pos rgyan. As stated above, rgyan (“adornment”) is used throughout as the translation of vyūha instead of the usual bkod pa (“array”). Nevertheless, it is usually referred to in Tibetan literature as sdong po bkod pa. The Sanskrit compound does not indicate the grammatical connection between the two terms gaṇḍa and vyūha, but the Kangyur colophon’s sdong pos rgyan ascribes an instrumental case to gaṇḍha, while its popularly known title sdong po bkod pa has no such case. Neither Sanskrit nor Tibetan specifies whether gaṇḍa is singular or plural. The title appears to have no connection with the content, unless it is taken to refer to the successive joints in a bamboo stem, as an analogy to the successive episodes in Sudhana’s journey.

Gaṇḍa in Buddhist Hybrid Sanskrit can have two meanings: “stem” or “stalk” and “pieces” or “parts” or “sections,” and the Pali specifies that, as a variation of gaṇṭha, it can mean the section between the joints of a stem, in addition to such things as a swelling, a boil, an excrescence, and so on. As the sūtra is composed of a series of episodes in which Sudhana meets a succession of teachers, the intended meaning could well have been “an array of parts” or, more freely, “a series of episodes.” The only use of the word gaṇḍa in the sūtra itself is within a compound in verse 112 in the Maitreya chapter: pañca­gaṇḍa­gati­cakra­mohitam. Pañcagaṇḍika is a standard BHS term for the five classes of existence, and therefore that compound could be translated as meaning “the ignorance of the wheel of the five sections of existence,” referring to hells, pretas, animals, humans, and devas. However, because the meanings of gaṇḍa can include boil, blister, abscess, goiter, cheek, or bubble (as well as harness, button, joint, bone, and so on), the Tibetan here translates gaṇḍa as “blister” (shu ba) so that the Tibetan translation of this compound is shu ba’i lam rgyud lnga yi ’khor lor rmongs (“the ignorance of the wheels of the blisters of the five existences,” which seems unlikely to have been the original intended meaning).

Douglas Osto also points out that gaṇḍa has been used as the first element of a compound in Sanskrit to mean “great” or “supreme,” and he therefore has translated gaṇḍa­vyūha as “Supreme Array,” which would have made for a reasonable title.

However, the English translation of the title here, “Stem Array,” follows the better-known version of the Tibetan title, preserving its peculiar ambiguity, while the less familiar title as given in the colophon could have been translated as “An Adornment by Stems.”

The title of the sūtra in which the Gaṇḍa­vyūha is the final chapter has also been interpreted variously. The word avataṃsaka is a substantiate of avataṃsa. In Classical Sanskrit, avataṃsa describes a garland or any circular ornamentation. For example, karṇāvataṃsa (ear avataṃsa) means “earring.” One peculiar Tibetan translation of buddhāvataṃsaka is “Buddha’s earring” (snyan gyi gong rgyan). A kusumāvataṃsa (flower avataṃsa) is a flower garland that is worn by a person, hence another Tibetan translation of buddhāvataṃsaka is “Buddha’s garland,” using an obscure archaic word for garland that has various spellings (rmad ga chad, rma ga chad, or rmag chad). Thomas Cleary, translating into English from the Chinese Huayan, calls it Flower Ornament. However, in Buddhist Hybrid Sanskrit (BHS), avataṃsaka means “a great number,” “a multitude,” or “a collection.” Therefore, we have the Tibetan version of the title as A Multitude of Buddhas (sangs rgyas phal po che). phal po che is used elsewhere in the Kangyur to translate Sanskrit words meaning “multitude,” such as nicaya, for “a great assembly of beings” (skye bo phal po che for mahat janakāya). Although this is the title given in all Kangyurs, some, such as the Urga and Degé have the title Flower Garland (rma ga chad) at the conclusion of each chapter, a possible indication that this was the earlier translation of the title, which has been left unchanged within the body of the text. However, the Mahāvyutpatti dictionary has phal po che for avataṃsaka, and neither rma ga chad nor its variant forms appear anywhere in the dictionary. This contradiction between the chapter colophons and the main title is absent in the Lhasa, Stok Palace, Narthang, Lithang, and Shelkar Kangyurs. This translation follows the example of those latter versions so as to avoid such an evident contradiction.

Whatever the intended meaning of the title, the Buddhāvataṃsaka Sūtra does depict a multitude of buddhas, among which are multitudes of the Buddha Śākyamuni, all of whom are emanated by the Buddha Vairocana.

During the course of the Buddhāvataṃsaka Sūtra we find that the Buddha Śākyamuni is but one of countless manifestations of the Buddha Vairocana. Śākyamuni is even referred to as the Buddha Vairocana. The Buddha Śākyamuni is depicted as being simultaneously present in various locations in our world realm: at the Bodhi tree, in the Trāyastriṃśa paradise of Indra, which is on the summit of Sumeru, in the Yāma and Tuṣita paradises high above Sumeru, and in the highest paradise in the realm of desire‍—the Para­nirmita­vaśa­vartin paradise. Śākyamuni is also depicted as being present in these same locations not only in our world realm but in countless other world realms. The Buddha Vairocana prayed to manifest in this way and to have vast assemblies that his manifestations would teach to, and these buddhas are the result of his prayer. According to the Buddhāvataṃsaka Sūtra, not only is Vairocana the source of all buddhas everywhere, but all the bodhisattvas whom those buddhas teach were previously pupils of Buddha Vairocana.

This depiction of Śākyamuni as a Vairocana emanation has its precedent in a sūtra that was never translated into Tibetan, the Brahma­jāla­sūtra, which introduces the Buddha Vairocana as the buddha who is the source of ten billion Śākyamunis simultaneously existing in various worlds. This sūtra should not be confused with the early Buddhist Brahma­jāla­sūtra, which has an identical title but entirely different content. That Brahma­jāla­sūtra was translated into Tibetan34 and is included within the Pali canon.

The Buddha Vairocana is therefore portrayed in the Buddhāvataṃsaka as the fundamental buddha who is the source of countless manifestations of the Buddha Śākyamuni. The Buddha Vairocana would also later become the central buddha in tantric traditions such as the Shingon tradition of Japan, which is based particularly on the Mahā­vairocanābhisambodhi Tantra (Toh 494).35 Even in the higher tantras Vairocana still retains his position as the central buddha in the five-buddha-family system.

The Borobudur panels portray Sudhana as a prince-like young man with a retinue, whereas there are Chinese and Japanese depictions of him as a chubby child. Many years pass in the course of his wanderings. In chapter 8 it is stated that he spent twelve years searching for the head merchant Muktaka, so even if he were a child at the beginning, he would be an adult by the end. However, the conventional passage of time is not a feature of this sūtra. Sudhana is introduced as being part of one of four groups that come to see Mañjuśrī when he goes to South India. There are laypeople‍—the male upāsakas and female upāsikās‍—and dārakas and dārikas (“sons and daughters” or “boys and girls”), which, like the compound strī­puruṣa­dāraka­dārikāḥ much later in the sūtra, appears to imply parents and their children, and this is specifically indicated when Mahāprajña, the first of the eleven named upāsakas, is identified as the father of the first of the daughters. However, the definition of the terms dāraka and dārika includes unmarried males and females up to the age of twenty, and each of these sons and daughters who come to meet Mañjuśrī is accompanied by a retinue. Therefore, the implication is that Sudhana is not a child but presumably around eighteen or nineteen years old. As Sudhana is the first of the eleven named sons listed, the implication appears to be that, as the most prominent of the sons, he is also the son of Mahāprajña. The only description of his family is the vast, miraculous wealth they obtained upon his birth.

Throughout the sūtra he is referred to as a śreṣṭhidāraka. The word śreṣṭhin in Classical Sanskrit can mean “distinguished,” “eminent,” “a person of high position,” and, more specifically, the president of a guild or a head merchant. Edgerton, for the Buddhist Hybrid, gives “guild leader” and “capitalist.” The Pali equivalent seṭṭhi, according to the Pali Text Society’s dictionary, can mean “the foreman of a guild,” “a treasurer,” “a banker,” or “a wealthy merchant.” Douglas Osto, in Power, Wealth and Women in Indian Mahāyāna Buddhism: The Gaṇḍa­vyūha-sūtra, used “merchant-banker’s son” to describe Sudhana, but subsequently in his translation of parts of the sūtra he used the simpler “merchant’s son.” Cleary and Carré, translating from the Chinese, have, respectively, “youth” and jeune (“young”), omitting a translation of śreṣṭhin. Here I follow the Tibetan translation of śreṣṭhin as tshong dpon, literally “head merchant.”

In chapter 8, Muktaka is simply called a śreṣṭhin, translated into Tibetan as tshong dpon (“head merchant”), and this translation follows the Tibetan, although his work is not described. Cleary translates śreṣṭhin as “distinguished man” and also as the adjective “noble”; Carré, translating from the Chinese, has aîné (“elder” or “superior”).

In chapter 18, Ratnacūḍa is said to be a dharmaśreṣṭhin, which was translated into Tibetan as chos kyi tshong dpon, literally “head merchant of the Dharma,” presumably meaning that he is a wealthy patron of the Dharma. Cleary translates as both “eminent person” and “religious eminent,” while Carré translates from the Chinese simply as aîné (“elder” or “superior”). Ratnacūḍa is twice associated with a market in the narrative and therefore does seem to be an extremely wealthy merchant, and his ten-story house of gold is filled with Dharma activities and even visions of buddhas and bodhisattvas.

In chapter 19, Samanta­netra is said to be a gāndhika­śreṣṭhin, translated into Tibetan as tshong dpon spod tshong (“head-merchant perfume seller”). Cleary translates this as “eminent perfumer,” which would correspond to the Sanskrit, and Carré has simply aîné (“elder” or “superior”), omitting reference to gāndhika.

In chapter 24, Utpalabhūti is also called a gāndhika­śreṣṭhin, translated into Tibetan slightly differently than in chapter 19 as spod tshong gi tshong dpon. Cleary again translates this as “eminent perfumer,” while Carré has simply parfumeur (“perfumer” or “perfume seller”), omitting reference to śreṣṭhin.

In chapter 26, Jayottama is simply called a śreṣṭhin, translated into Tibetan as tshong dpon (“head merchant”). Cleary translates as “eminent man,” and Carré, translating from the Chinese, has marchand (“merchant”). However, in the narrative there is no mention of the work he does, but instead he is said to be advising householders on civic duties, doing so in terms of the Dharma.

Chapters 10 and 15 of the Gaṇḍa­vyūha have two long lists of numbers that demonstrate the innumerable beings for whom bodhisattvas seek enlightenment. Any number, no matter how large, does not encapsulate the scope of their salvific agenda. These two lists of numbers should have been identical. As this is a unique list of numbers, they have proved to be very susceptible to corruption or omission. There are differences between the two lists in the surviving Sanskrit and also for the same lists in different Sanskrit editions. Moreover, the same numbers were translated differently into Tibetan in the two chapters even though the translation in chapter 10 of the Gaṇḍa­vyūha is reproduced exactly in chapter 36 or the 36th sūtra in the Avataṃsaka as The Teaching on Using Numbers on Being Questioned by Cittarāja.36 The Gaṇḍa­vyūha lists were recorded in the Mahāvyutpatti Sanskrit–Tibetan dictionary, revealing not only that variations already existed but that they also differed from the Kangyur versions. The Chinese should be the earliest witness to the condition of the lists. However, Buddhabhadra and Śikṣānanda omitted most of the numbers in the Gaṇḍa­vyūha chapter, instead writing “and so on” and jumping to the last number. Therefore, they are not found in Carré’s translations, and Cleary reproduces the version found in the Sanskrit. Meanwhile, Prajñā recorded the list in Chinese phonetics. The phonetics of Chinese ideograms have varied over time and regions, but they are still able to give an indication of the list. However, because of being a long sequence of names for incalculable numbers, it may have suffered from the same defects through successive copying. In addition, Śikṣānanda transliterated the list as it occurs in chapter 36 of the Avataṃsaka. In our translation there is an attempt to supplement omissions that have occurred in the two lists and to find the correct Sanskrit spelling for the numbers, with the Mahāvyutpatti record of the numbers being particularly important, so as to create a uniformity between the lists. Some of the recorded forms of the numbers in chapter 15 have the nominative case in -u, which may well be a remnant of the original Buddhist Hybrid Sanskrit of the sūtra. It may never be possible to ascertain the original condition of the lists, but their purpose was not to create a reliable mathematical tool but to overwhelm the mind with an inconceivable vastness of numbers. Therefore, a disproportionate amount of time has been spent on these pages of numbers, even though a reader may very well skip over them.

The system of enumeration reaches numbers of such enormous value that they exceed even the extensive system of names for large numbers that have been developed in English in modern times, the highest being the googolplex (a 1 followed by a hundred zeros, and the source for the name and verb Google). Even such a number is dwarfed by the vastness of the universe of the Gaṇḍa­vyūha, where even within each atom there are as many buddhas as there are atoms in total. Therefore, an attempt to provide equivalent English names foundered, and the inconsistency in the Tibetan has resulted in choosing the Sanskrit names for this translation, though the various Tibetan, Chinese, and Sanskrit equivalents or variants are supplied in the endnotes.

Although the translation is based on the Degé edition, and the numbers of the xylograph Degé folios are included within the translation, it is the Comparative Edition of the Degé that has proved invaluable, as it contains detailed annotations of the discrepancies with other editions of the Kangyur, often revealing errors in the Degé. Some of these errors are evidently the result of the copyist mishearing a word being dictated by a reader. The sūtra existed only in handwritten copies for five centuries before the first woodblock printing (the Yongle edition made in China) in 1410. Not included in the Comparative Edition but consulted for this translation is the Stok Palace Kangyur, which belongs to another group of Kangyurs and has occasionally been alone in preserving the correct spelling, for example, rnyi (“snare”) translating the Sanskrit jāla (“net”) while all other available Kangyurs have snying (“essence” or “heart”).

The Sanskrit often provides a witness as to which Kangyur has the correct spelling. Sanskrit manuscripts are themselves just as subject to corruption, and surviving Sanskrit manuscripts date to a later period than the Tibetan translation. At times the Tibetan is evidently translating from a word similar to that in the presently available Sanskrit, and therefore it may have been translated from a manuscript in which particular words were not yet corrupted, or it could be translating from a manuscript that contained an error. The Chinese translations are the earliest witness to the condition of the sūtra in Sanskrit; at times the Chinese agrees with the Tibetan and at times with the available Sanskrit. The work of our Chinese consultant, Ling Lung Chen, has been illuminating in this respect. An example of where the Tibetan and the Chinese are correct and the present Sanskrit is not is where mukha (“gateway”) has been misspelled as sukha (“bliss”). There are also cases where the Classical Sanskrit meaning of a word has been incorrectly adopted for the Tibetan translation instead of the Buddhist Hybrid Sanskrit meaning. Another problem is inadvertent omission of words, particularly in lists, which may have occurred in the Sanskrit manuscript being used or early in the process of manuscript copying in Tibet. These missing words are included in the translation, even though the meaning would not be affected by their omission.

The Sanskrit has also been helpful when the meaning in Tibetan translation is vague and open to interpretation or seems dubious in its accuracy. Another challenge particular to this sūtra is the repetition of long descriptive compounds that are not stock phrases and are meant to be overwhelming, describing qualities beyond the grasp of the ordinary mind. The meaning of words in this context and their interrelationship is open to a wider interpretation than usual; for example, if sarva (“all”) begins a compound, it has been open to interpretation as to which part of the compound it refers to. Generally, however, in this translation the Tibetan interpretation is followed unless there appears to be an egregious (and noted) error or an inadvertent omission. Instances of discrepancies among the Tibetan, the Sanskrit, and the Chinese are included in the notes.

Tibetan does not have the precise cases that Sanskrit has; therefore, the Sanskrit has been invaluable in indicating singular or plural as well as where in long passages the compounds change case, such as from nominative to instrumental in the long passage describing Sudhana at the beginning of chapter 56.

Also, when Sanskrit compounds are translated into Tibetan, it is not always clear where one compound ends and the other begins. This is sometimes indicated by the shad, the vertical line in Tibetan orthography, but as the Tibetan editor’s colophon indicates, the placement of the shad is at times arbitrary and confusing, the likely result of the process of copying, as some of the errors are evidently transcription errors from listening to the text being read. The Vaidya Sanskrit edition in Roman letters has added punctuation that does not always agree with the Tibetan interpretation of the structure of the sentence. The Suzuki edition of the Sanskrit has also been useful for finding transcription errors in the Vaidya. Tibetan also tends to translate the prefixes of Sanskrit words, while a direct translation from Sanskrit into English would choose a word from the more extensive English vocabulary. This translation tends to follow the latter approach rather than the translation of prefixes. Words can be multivalent in Sanskrit, their meaning altered according to context, and can even be interpreted as having more than one meaning simultaneously. There are Sanskrit poems written with a carefully chosen vocabulary to create verses that can be read as having two different meanings. There are some terms in this sūtra where a choice of translation necessarily means rejecting another possible meaning; for example, vajra can refer to a thunderbolt or a diamond, and these two can be seen in that culture as identical. In particular, however, there are two prominent terms in this sūtra that present such a problem: dharmadhātu and samantabhadra­bodhi­sattva­caryā.

The word dharmadhātu is a compound made of two elements: dharma and dhātu. Dharma has a great number of meanings, from “the Buddha’s teachings” to “law,” “justice,” “qualities,” “nature,” “phenomena,” and so on, while dhātu can indicate a realm, an element, or a constituent and can be used to mean sensory elements, the relics from an enlightened being’s cremation, and so on. In the term lokadhātu, used over and over in this sūtra, dhātu can mean “realm,” as in “world realm” or “a realm of worlds.” The term dharmadhātu has been used to mean “essence or true nature of phenomena,” such as emptiness, and so the Chinese title, which could be translated as Entering the Dharmadhātu, has been translated by Carré as l’Entrée dans la dimension absolue (Entry into the Absolute Dimension) and by Cleary as Entry into the Realm of Reality. Osto has used the more literal “realm of the Dharma,” which has in certain contexts been used in this translation. However, often the meaning appears to be “the realm of phenomena” when something is said to spread everywhere, and occasionally it could mean either, or simultaneously both.

The ambiguity of samantabhadra­bodhi­sattva­caryā is reflected in differing translations. It is a compound used repeatedly in the sūtra, and the question is whether samantabhadra (“completely good”) is an adjective for bodhi­sattva­caryā (“bodhisattva conduct”) or whether it means the conduct of the bodhisattva Samanta­bhadra. The sequence of the words in the compound (this is not evident in Tibetan) would appear to signify the former meaning, but its use in the Samanta­bhadra chapter appears to indicate the latter. At times it could mean both simultaneously.

Neither the Tibetan nor the Chinese‍—nor therefore the Sanskrit from which they were translated‍—had a division of the Gaṇḍa­vyūha into chapters. The later Sanskrit has a division into fifty-six chapters, and because this will make the reading of the sūtra much easier and more accessible for the reader in English, that chapter division has been adopted for this translation.