The Perfection of Wisdom in One Hundred Thousand Lines

From a historical perspective, a group of “long Prajñāpāramitā sūtras,” including texts that exist variably in Sanskrit, Chinese, and Tibetan, appears to have been one distinctive genre that took form from the early Prakrit literature on the perfection of wisdom that first appeared in writing in the first centuries ʙᴄᴇ and ᴄᴇ. Modern scholars have disagreed about which of the geographically dispersed Buddhist communities of the time may have first given rise to this literature, some favoring its origin among the Mahāsāṅghikas of Andhra in the south of India, while others point to evidence of its early flourishing in the northwest regions such as Gandhāra. Whichever may be the case, a birch-bark scroll from the northwest, in the Gāndhārī language, written in Kharoṣṭhī script, and found in Bajaur (a district of present-day Pakistan near the Afghan border), has been radiocarbon-dated to the first century ᴄᴇ and is currently the oldest known Prajñāpāramitā manuscript.⁷ It is fragmentary and cannot be matched to any extant recension of the complete sūtras, or identified as belonging to the “long sūtra” group. If anything, it may be most closely related to The Perfection of Wisdom in Eight Thousand Lines, and thus supports the hypothesis that another group of sūtras similar to the Eight Thousand may be older in form than both the longer sūtras and the shorter ones, which evolved from texts of the Eight Thousand subfamily via processes respectively of expansion and contraction.⁸

The earliest surviving manuscript that can be identified as a “long” (Mahāprajñāpāramitā) version is another birch-bark scroll, this one found along with a large number of other texts in Gilgit in 1931. It is in Sanskrit and can be dated by details of its script to the sixth or seventh century ᴄᴇ.⁹ Although it was thought at first by Edward Conze¹⁰ to be a hybrid consisting of parts of The Perfection of Wisdom in Twenty-Five Thousand Lines and parts of The Perfection of Wisdom in Eighteen Thousand Lines, it is now taken as one among the many coexisting versions of the “long” sūtra series.¹¹ As the most complete of the Gilgit Prajñāpāramitā manuscripts, it is a very important source; the others are incomplete fragments. Another, similar Sanskrit manuscript of a generic “long” version was found in Dunhuang.¹²

It is important to bear in mind that the naming of the different versions by the number of lines¹³ they contain is likely to have been a later development, applied as a means of classifying the profusion of circulating texts of different lengths. It was already in use by the time these texts were first translated into Tibetan in the late eighth and early ninth centuries, but it is not a feature of the oldest Chinese translations. The earliest evidence of this nomenclature appears to be in the Chinese literature, in the record of a lecture by the sixth-century translator Bodhiruci,¹⁴ and its widespread adoption in the centuries that followed may have served to limit further profusion and even reduce the variety of different texts by fixing their number. Those texts in different languages that can be seen as belonging to the “long” sūtra group (as distinct from the mid-length Perfection of Wisdom in Eight Thousand Lines and the many short versions, each of which followed their own evolutionary path), differentiated as they are by greater or lesser degrees of expansion of the lists of dharmas, show complex patterns of textual proximity that do not necessarily follow the numerical denominations that were retrospectively applied to them. Indeed, these numerical titles may obscure rather than clarify the recensional affinities.¹⁵

With that important reservation, there are three recensions within the overall group of “long” Perfection of Wisdom sūtras that can nevertheless justifiably be labeled The Perfection of Wisdom in One Hundred Thousand Lines:

The Tibetan and Sanskrit recensions are quite similar to each other, while the Chinese differs from both in a number of respects.

The Chinese translation was made by Xuanzang in the mid-seventh century from the massive collection of material he had brought from his travels in India. It takes the form of the first of the sixteen sections or “assemblies” that represent historically independent texts and make up the voluminous compilation of perfection of wisdom works he translated as The Large Perfection of Wisdom Sūtra (Taishō 220).¹⁶

Just how the extant Sanskrit and Tibetan versions differ compared to this single comparable Chinese translation has not been fully documented, but the differences are not to be ignored. The Sanskrit original from which Xuanzang translated this section is said to have been 132,600 ślokas in length, and thus possibly even longer than the Sanskrit texts that were translated into Tibetan and have also survived in the Nepalese tradition. Moreover, the sections of Xuanzang’s compilation, despite being explicitly differentiated, are presented together as in some sense comprising a single work, and it is thought that this arrangement was not Xuanzang’s own invention but may have been a feature of his Sanskrit source texts.

Nevertheless, it is practical, and a close approximation, to consider the first section as identifiable with the texts known in Sanskrit and Tibetan as The Perfection of Wisdom in One Hundred Thousand Lines.

The second and third sections of Xuanzang’s Chinese translation correspond in similar respects to the Twenty-Five Thousand Line and Eighteen Thousand Line sūtras as found in both Sanskrit and Tibetan (for the Twenty-Five Thousand Line) and in Tibetan alone (for the Eighteen Thousand Line). It is also significant that the first and second sections, despite their differences in the degree of compression, are strikingly similar to each other in language, content, and order, but less closely related to the third. The same pattern of similarities and differences exists between the Hundred Thousand, Twenty-Five Thousand, and Eighteen Thousand Line sūtras in Tibetan and (for the first two) Sanskrit.

A final point to be made concerning the Chinese translation is that the relatively late appearance in China of this equivalent of the Hundred Thousand, the longest of the long sūtras, almost five centuries after the first translation of the equivalent of the Eight Thousand and four centuries after the first appearance of the equivalent of the Twenty-Five Thousand, provides supporting evidence for the notion that‍—for the long sūtras‍—a process of expansion from shorter to longer versions, rather than contraction from longer to shorter, may provide the better account of their evolution.¹⁷

Early Gāndhārī and Sanskrit manuscripts of generic Prajñāpāramitā sūtras are mentioned above (1.9–10). Specifically of this Hundred Thousand Line version of the long sūtras, however, no Sanskrit manuscript has survived that can be dated as early as the Chinese and Tibetan translations. But there are Nepalese Sanskrit manuscripts of more recent date that, from their content, must be closely related to the Sanskrit original from which the Tibetan translation was made. Some of them may even be copied descendants of a Sanskrit manuscript in the temple of Phamthing that Tibetan histories say was consulted by Ngok Loden Sherab in the eleventh century to correct the earlier Tibetan translation (see below).

Nepalese manuscripts of varying dates are presently to be found not only in Nepal but also in Kolkata, Cambridge, New Delhi, Paris, and Tokyo. A critical edition of the first twelve chapters was published in three installments by Ghoṣa between 1902 and 1914, and more recently has been extended by Kimura in four further volumes, published between 2009 and 2014.

The Sanskrit manuscripts of the Hundred Thousand and Twenty-Five Thousand resemble each other closely in terms of language, terminology, content, and order, as is the case with their Tibetan translations, and like them differ mainly in the degree of expansion of the different groups of dharmas. The Sanskrit of the Hundred Thousand matches the Tibetan translation in content closely.¹⁸

For anyone interested in how the early Tibetan translations of canonical works in general were carried out, it is a disappointing fact that much of the detailed information about the process was either not recorded or has been lost. However, in the case of The Perfection of Wisdom in One Hundred Thousand Lines‍—thanks to the unique status of this monumental scripture‍—traditional historical accounts include far more detail of the successive versions that led to the Tibetan translation preserved in the Kangyurs we have today than is the case for any other work.

A succession of Tibetan translations were made from Sanskrit in the late eighth and early ninth centuries. There are specific terms for these imperial-period manuscript versions of the Hundred Thousand: Labum (bla ’bum) and Lagyur (bla ’gyur), meaning, respectively, a Hundred Thousand (Tibetan ’bum), or a translation (’gyur), that is commissioned or owned by one who is “foremost” or “paramount” (bla), i.e., the emperor.¹⁹

As a terse summary in his list of canonical translations, Butön’s fourteenth-century History of the Dharma states that “it is well known that there were six versions attributed to the translators Nyang Khampa Gocha,²⁰ Vairotsana, Che Khyidruk, Zhang Yeshe Dé, and others.”²¹ Other accounts mention only four versions, while being in broad agreement on the rough outline of how the Tibetan translations evolved over time.

Much fuller details than Butön sets out are provided by other histories, including a historical survey that opens the commentary on this text by the great prajñāpāramitā commentator Rongtönpa (rong ston shes bya kun rig, 1367–1449);²² the lineage records of Minling Terchen Gyurmé Dorjé (smin gling gter chen ’gyur med rdo rje, 1646–1714); a survey of the prajñāpāramitā literature in a commentary to the Heart Sūtra by Alaksha Tendar (a lag sha bstan dar, 1758–1839); and an account in the encyclopedic Treasury of Knowledge by Kongtrul Lodrö Thayé (kong sprul blo gros mtha’ yas, 1813–99).²³ The catalog of the Degé Kangyur,²⁴ and particularly the historical section of the detailed Narthang Kangyur catalog, written by the fifth Olkha Lelung Jedrung, Lobsang Trinlé (’ol kha / dga’ sle lung blo bzang ’phrin las, 1697–1740),²⁵ are also very informative. Rongtönpa’s work is the earliest of these histories and most of the others may have drawn from it, or perhaps from other even earlier accounts. The information to be gleaned from these different works is not always consistent, and sometimes conflicting. Here is a tentative synopsis:

• The earliest translation was one made by Khampa Gocha, who had been sent with offerings of gold to India by King Tri Songdetsen²⁶ to bring back the sūtra; his four-part²⁷ translation was called The Translation from Memory (thugs ’gyur) because Khampa Gocha had memorized the text while in India‍—some accounts say he had attained the dhāraṇī of perfect recall‍—and only wrote the Tibetan translation when he was back in Tibet. The king, as an offering of merit to the queen who had died, had a copy of that translation written in ink made from his own blood,²⁸ using goat’s milk as a binder; this version was therefore called The Red Manuscript (reg zig dmar po) or The Red Goat Manuscript (ra gzigs dmar po) and also had four parts. In later times it was taken to Lhasa and is said to have been housed in the brick caitya near the Trulnang temple. It is not clear whether the above names and descriptions refer to a single manuscript or to two different ones of the same translation, but in any case this version, abridged and condensed as it is into four volumes, is also called The Short Imperial Translation (bla ’gyur chung ngu).

• Since the king did not have full confidence in The Translation from Memory, he commissioned Nyang Indrawaro and Wé Mañjuśrī to seek out the text in India and translate it. This they did, and their translation was written using ink made from indigo and the king’s singed hair, with goat’s milk binder. This manuscript was therefore called The Blue Manuscript (reg zig sngon po) or The Blue Goat Manuscript (ra gzigs sngon po); it was also called The Authorized Hundred Thousand (bca’ ’bum), and because the translators’ needs were met through a levy collected from the people, it was known as The Levy Hundred Thousand (dpya ’bum) as well. It too is said to have consisted of four parts, and because they were not held together with cloth bands but had iron fasteners it was also called The Iron Fasteners (lcags thur can). It is said to have been kept at Samyé.

• Later, Pagor Vairotsana, at the instigation of Mutik Tsenpo (also known as Senalek), compared the Sanskrit text with the translation in The Iron Fasteners, revised it, and filled out the abridgements of the earlier version by adding the missing repetitions and lengthening some lines. He wrote out a new six-volume manuscript, which was called The Medium-Length Imperial Translation (bla ’gyur ’bring po). According to Rongtönpa, it was also known as The Snowy One Promised by the King (rgyal po’i thugs dam par kha ba can), from the name of (or on) the binding boards, and both Rongtönpa and the Narthang catalog add, too, that it was called The Bats (pha wang can) because there was a bats’ nest near the place in Samyé where it was kept. Kongtrul says that it was still in Samyé Chimphu in his time (i.e., the mid-nineteenth century). Other authors, including Situ Panchen in the Degé catalog and Kongtrul, say that this is the version called The Deerskin Case Hundred Thousand (’bum sha sgro can), and the Narthang catalog explains in some detail how all the fascicles came to be stored as scrolls in a deerskin. However, the Narthang catalog says that the same name was also applied, for similar reasons, to the next revision; this may explain the apparent inconsistencies between authors in regard to this particular moniker.

• Later still, in the reign of Mutik Tsenpo’s son Tri Ralpachen, in the period when a number of Indian scholars were working with Tibetan translators in Tibet on the translation and revision of many texts and the great language revision edicts were drawn up, Surendrabodhi, Kawa Paltsek, Chokro Lui Gyaltsen, and others made an extensive revision of the translation in six parts (according to Rongtönpa) or sixteen (according to the Narthang catalog, Tendar, and Kongtrul), which was called The Long Imperial Translation (bla ’gyur chen mo). Rongtönpa calls this version The Deerskin Case Hundred Thousand (’bum sha sgro can) rather than the preceding one but, as mentioned above, the Narthang catalog explains how both versions were kept in deerskin wrappers and the name can therefore refer to either one or the other.

• At a point in this sequence that remains to be determined and is not mentioned in any of these accounts, but lies probably in the late eighth or early ninth century period, as many as eight copies of the whole text were made by scribes in Dunhuang. They were probably commissioned by or on the behalf of one of the kings (probably Tri Ralpachen). Pages from these copies, as well as rejected pages subsequently used for writing practice, make up a large proportion (along with manuscripts of The Aparimitāyurjñāna Sūtra²⁹) of the manuscripts found in Dunhuang.³⁰

• The Narthang catalog then summarizes the six early manuscripts, and gives them yet more monikers:

“At that time, since to produce a great Mother (i.e., a Hundred Thousand) was possible only for the king and not for his subjects, only these six were made:³¹ (1) Tri Detsuk’s Monochrome Imperial Hundred Thousand (khri sde gtsug gi bla ’bum skya bo); (2) Senalek Jingyön’s Innermost Hundred Thousand (mjing yon gyi sbug ’bum); (3) Lhasé Tsangma’s Demarcated³² Hundred Thousand (gtsang ma’i bye ’bum); (4) Ralpachen’s Six-Volume Hundred Thousand (ral pa can gyi drug ’bum); (5) Prince Namdé’s³³ Red-Faced Version (gnam lde lha’i zhal dmar can); and (6) Darma’s Yellow-Paper Version (dar ma’i shog ser can).”³⁴

• At this point in the story there seems to have been a proliferation of further Hundred Thousands produced as copies of one or another of these six, perhaps resulting from a royal prerogative on sponsoring them coming to an end. Rongtönpa provides a detailed list of seventeen named Hundred Thousands and the places they were kept,³⁵ including the earliest ones he had already described, culminating in one made by Chang Dorjé Tsultrim (lcang / cang rdo rje tshul khrims)³⁶ of Ru Tsam (ru ’tshams), who produced the seventeenth from a detailed comparison of the other sixteen. Rongtönpa goes on to classify the seventeen into groups according to the short, medium-length, or long Labum from which they were copied.³⁷ The Narthang catalog has less detail but correspondingly mentions nineteen Hundred Thousands, including those of the subsequent period that could be produced not just by the kings but by subjects. It also points to the importance of the same significant figure, Chang Dorjé Tsultrim, who soon afterward compared all nineteen and made a “highly corrected version.”³⁸ Proliferating from that version, the catalog says, about sixty copies were made, and indeed Rongtönpa proceeds to list a large number of these and their locations to a total of sixty-five.³⁹

• In the later translation period, in the late eleventh century, Ngok Lotsāwa, according to the Degé catalog and others, having consulted the Indian manuscripts to be found in Tibet and the Iron Fasteners translation, revised and corrected the above-mentioned Long Imperial Translation by comparing it against a Sanskrit manuscript in the Phamthing temple in Pharping, Nepal. These catalog accounts also mention that the corrections included the addition of the names of three meditative stabilities that had been missing in earlier versions.⁴⁰ It is Ngok’s revised translation that, according to most Kangyur catalogs, is preserved in the Kangyur. However, although Rongtönpa places Ngok’s version in fifth and last place among the most significant, major translations,⁴¹ it is difficult to be entirely certain where to place it in time relative to the other revisions and simple copies that he also details, mentioned above.⁴²

• The Narthang catalog, indeed, seems to place less importance on Ngok’s revisions and mentions his version only in passing. Rather, it attributes the final establishment of a fully correct version, a “later descendant of the Imperial Hundred Thousands” (bla ’bum gyi bu phyi), to a Yarlung Jowo Chöjé (yar klung jo bo chos rje).⁴³ The catalog also gives more extensive detail than any other of the differences between versions and the corrections made to establish the definitive version, including the varying numbers of meditative stabilities but also some of the sections and passages that had been missed in earlier manuscripts.

The translation preserved in most Kangyurs is thus the result of this long process of evolution. It is commonly divided into twelve volumes, but the Narthang catalog mentions other numbers of volumes into which different manuscripts were divided, and Rongtönpa, saying that there was even one version in thirty volumes, sets out schemes with the details of which bampo (fascicles) were included in which volume for several different volume arrangements: twelve (as in the Degé and many other Kangyurs), fourteen (as in the Berlin and Qianlong Kangyurs), and sixteen (as in the Choné, Phukdrak, Lithang, London, Stok Palace, Ulaanbaatar, and some of the peripheral Kangyurs).⁴⁴

The Degé catalog, describing the translation as being the eleventh-century one that resulted from Ngok Lotsāwa’s revision of previous Tibetan translations against the Sanskrit manuscript of Phamthing, also specifies that the text has seventy-two chapters. It mentions the traditional legend according to which the four final chapters that conclude the Twenty-Five Thousand Line sūtra‍—the questions of Maitreya, the story of Sadāprarudita and Dharmodgata, and the entrusting of the text‍—are missing from the Hundred Thousand because when Nāgārjuna brought the sūtra from the realm of the nāgas, the nāga king had withheld those four chapters to ensure that Nāgārjuna would return. There is, the catalog says, a tradition of appending those four chapters, copied from the other sūtras. Although the Degé Kangyur does not follow that tradition, the Narthang, Lhasa, Namgyal, Hemis, and Shey Kangyurs do, and thus have seventy-five or seventy-six chapters.

The text in the Degé Kangyur is comprised of 301 bampo (fascicles), with between twenty-two and twenty-seven in each volume. The bampo numbers begin again with each volume, unlike most other Kangyurs where they are numbered consecutively throughout the text. Kangyurs in which the extra final chapters have been added have three more bampo, bringing their total to 304.

The Tibetan text in the Degé Kangyur, interestingly, preserves the old orthography of a subscript ya, “ma yata” (ma ya btags), i.e., the subscript ya under the letter ma in certain words, and of the “dadrak” (da drag), i.e., da as a second suffix in certain words ending in na, ra, or la. Spellings such as myi and myed for mi and med, rkyend for rkyen, tshuld for tshul, and smyind for smin are widespread, although in the first volume particularly some of these archaic spellings have been inconsistently revised. In most other Kangyurs these spellings are absent, but the Urga, Namgyal Collection, and Gondlha Kangyurs have also preserved them. Indeed, in the Namgyal Collection version of the sūtra, these archaic features are considerably less revised and spellings such as pha rold tu phyind pa (“the perfection of wisdom”) and rab ’byord (“Subhūti”) are preserved.

Given the complex process by which many translators contributed to the extant translation, it is perhaps not surprising that in the Degé and in the majority of other Kangyurs, there is no translators’ colophon. In some Kangyurs, however, the text does have a colophon, and these fall into two groups:

“This is the golden Hundred Thousand called Not Mixed with Wrongs in twelve sections, and is based on the corrected, limit-defining Black Hundred Thousand of Gyan-gong, which itself had resulted from being proofed and edited sixteen times after being compared against the early limit-defining, abridged manuscripts in Samyé and Lhasa, and so forth, and which, since the great lotsāwa, the omniscient Butön Rinchen Drup, had confirmed it as the prototype, was used as master copy for the Hundred Thousand produced by the great scholar Rinchen Gyaltsen.”⁴⁶