1
TANTALISING TRACES OF THE LABOURS
OF THE LOTSĀWAS:
ALTERNATIVE TRANSLATIONS OF SANSKRIT SOURCES IN
THE WRITINGS OF RJE TSONG KHA PA
CHRISTIAN K. WEDEMEYER (CHICAGO, USA)
I. INTRODUCTORY REFLECTIONS
Much has been made, in both scholarly and more popular literature, of
the marvel of the Tibetan translations of Buddhist texts between the
eighth and the fourteenth centuries. In particular, the Tibetan translations have been praised for their alleged extreme precision—an exactitude the presumption of which has led some to attempt ‘back translating’ of works from Tibetan into Sanskrit.1 Walter Eugene Clarke, in his
1951 Presidential Address to the American Oriental Society, enthusiastically praises the Tibetan translations, by way of contrasting them
favorably with their counterparts in Chinese, writing:
The Tibetan translations are marvelous for their word-for-word fidelity to
the original. They are of great help in dealing with badly mutilated texts
and in giving a clear impression of the original. With a good knowledge
of Tibetan and of Buddhist Sanskrit the Tibetan texts could be rewritten
in a Sanskrit that would approximate very closely to the original.2
The author would like to thank the members of the International Association for
Tibetan Studies in attendance when this paper was first delivered on 8 September 2003,
under the title “Alternative translations of Sanskrit sources in the writings of Rje Tsong
kha pa: a survey and analysis of the criteria for preference”. Particular thanks are due
to Dr Helmut Eimer and Dr Isabelle Onians for helpful comments at that time. Also
beneficial were discussions held between sessions with Professors Matthew Kapstein
and Ronald Davidson. The input of all of these colleagues resulted in the present paper
being substantially recast and, it is to be hoped, at least slightly improved.
1 For example, in the 1930s, Rāhula Sā#k$tyāyana described his work to “restore
some of the great works of the Buddhist logicians, from Tibetan to Sanskrit”—in particular the Pramā"avārttika of Dharmakīrti. See Sā#k$tyāyana 1935: 21. Similar efforts
continue to this day at, for example, the Central Institute for Higher Tibetan Studies,
Sarnath, Varanasi, India.
2 Clark 1951: 210.
CHRISTIAN K. WEDEMEYER
2
On the whole, of course, there is some truth to this. The normative
practice enjoined of Tibetan translators was to adhere to a rather strict
policy of following prescribed translation-equivalents—which terminology is set out in the Bye brag tu rtogs par byed pa chen po (a.k.a.
the Mahāvyutpatti). Guidance, prescription and proscription, was also
to be had from a companion work, the Sgra sbyor bam po gnyis pa—
which encourages, for instance, the preservation of the syntax of the
Sanskrit original, including (in so far as possible) retaining verbal prefixes (upasarga) and word order. The compilers of this work were careful to stress that deviation from the Sanskrit syntax is allowable in order
to produce a comprehensible Tibetan product;3 however, in practice,
translators seem to have been somewhat reluctant to follow this advice,
seeking instead to preserve these elements of syntax even when the
result was a rather artificial (not to say clumsy and misleading)
Tibetan.4
3
The Sgra sbyor bam po gnyis pa specifies that the method of translating the Holy
Teaching should be such that, without deviating from the meaning, the translation
should be easy to understand in Tibetan. That if verses can be rendered into Tibetan
without disturbing the Sanskrit word-order, it should be done so. However, if this is not
the case, they (as well as any prose passages) should be rearranged for ease of understanding word and sense (dam pa’i chos bsgyur ba’i lugs ni don dang mi ’gal la bod
skad la yang gar bde bar gyis shig | dharmma bsgyur ba la rgya gar gyi skad kyi go
rims las mi bsnor bar bod kyi skad du bsgyur na don dang tshig tu ’brel zhing bde na
ma bsnor bar sgyur cig | bsnor na bde zhing go ba bskyed pa zhig yod na | tshigs bcad
la ni rtsa ba bzhi pa’am | drug pa’ang rung ste | tshigs su bcad pa gcig gi nang na gang
bde ba bsnor zhing sgyur cig | rkyang pa la ni don gang snyegs pa yan chad kyi tshig
dang don gnyis ka la gar bde bar bsnor zhing sgyur cig |). It further stipulates that
upasarga-s should be translated literally and as an extra element only if they have a
semantic effect; those which do not add anything semantically (i.e. upasarga-s used
pleonastically) need not be translated as an extra element, but rather the entire verb may
be translated according to the meaning (pari dang | sam dang | upa lta bu la sogs te |
tshig gi phrad dang rgyan lta bur ’byung ba rnams bsgyur na don dang mthun zhing
’byor pa’i thabs ni | yongs su zhe’am | yang dag pa zhe’am | nye ba zhes sgra bzhin du
sgyur cig | don lhag par snyegs pa med pa rnams ni tshig gi lhad kyis bsnan mi dgos
kyis [sic for kyi] don bzhin du thogs shig |). See Ishikawa 1990: 2–3. For a valuable discussion of this, see Kapstein 2003: esp. pp. 755–58.
4 Contrary to the flexibility dictated by the Bam gnyis (see above, note 3), it would
seem that de facto prevalent Tibetan translation practice in the Phyi dar period dictated retaining all Sanskrit upasarga-s and word-order. In regard to the former, the translator could then choose whether also to translate literally the root dhātu (apparently a
rarer practice) or to follow the literally-translated upasarga with a Tibetan verb rendering the meaning of the entire upasarga-dhātu complex. An example of the former may
be found in Lo chen Rin chen bzang po’s translation of verse 16 of the
Cittaviśuddhiprakara"a of Āryadeva (Sems kyi sgrib pa rnam par sbyong ba, Tōh.
1804), wherein he renders the term vyavasthiti$ (‘distinction’) with the literal rnam par
gnas, instead of e.g. rnam par ’byed. An example of the second, more common, type (as
found, for example, in Lo chen’s translation of the Caryāmelāpakapradīpa, chapter 5)
TANTALISING TRACES
3
On the other hand, the widespread myth of the absolute precision
(and the consequently-assumed univocality and transparency) of the
Tibetan translations is misleading in a number of ways.5 It occludes, for
example, a) the variety of the Sanskrit manuscripts imported into Tibet
during these periods. Univocality of translation implies univocality of
the translated—a situation which we know, from a variety of sources
(including some we will consider below), not to have been the case.
There were numerous lines of manuscript transmission for many (if not
most) of the works imported to Tibet; any one translation—no matter
how faithful, no matter how transparently it renders its source—gives
access to only one ‘original’.6
It further overlooks b) the variation in the quality of these manuscripts. As will be seen in more detail in what follows, it can be determined from surviving materials that many of the Tibetan translations
were based upon idiosyncratic texts. In addition, it neglects c) variation
in the quality of the translations. Even those which may likely have
been based on reliable texts, simply do not translate their source texts
with consistent accuracy. While most translations do tend to follow a
strict set of lexical equivalencies—which suggest ‘word-for-word
is translating anu bhū (‘to experience’) as rjes su myong ba, rather than the literal *rjes
su ’gyur ba (which latter is actually prescribed in the Mahāvyutpatti as one of three
possible equivalents for anuvidhānam). As an example of preserving Sanskrit wordorder, one may consider Lo chen’s rendering of the line from the Guhyasamāja Tantra,
Chapter 17, daśa-kuśalān karma-pathān kurvanti jñāna-varjitā$ | as dge ba bcu’i las
kyi lam | byed pa ye shes spangs pa’o |. In this instance, as Tibetan is an SOV language,
the retaining of the inflected verb in clause-medial position is confusing. I would have
suggested, for example, dge ba bcu’i las kyi lam | ye shes spangs pa rnams byed do |.
5 It may be noted that I am by no means the first to point this out. The inadequacy
(or, rather, the tentativity) of Sanskrit reconstruction based on Tibetan translations has
been indicated by, for example, Bhik'u Prāsādika, in the Foreword to his own attempt
at reconstructing the text of the Vimalakīrti-nirdeśa-sūtra, wherein he notes the rather
different results obtained by P.L. Vaidya and V. Bhattacharya in reconstructing Āryadeva’s Catu$śataka (though it might also be entertained that a rather partial command of
Tibetan idiom also played a hand here). See Prāsādika and Joshi 1981: ix–x. As he notes
therein, Étienne Lamotte had already made a similar point in his 1962 French translation of the same scripture.
As will be seen below, however, I make the case anew in light of new and rather different materials. Given the widespread currency of the myth of reconstructability, I
believe the point is worth making again.
6 An exception, of course, would be a translation based upon multiple source texts,
in which case the translation would consequently be an ‘eclectic’ one and hence
would—unless remarkably successful in its critical enterprise—if ‘reconstructed’, provide an ‘original’ text that may never have existed.
4
CHRISTIAN K. WEDEMEYER
fidelity’—significant variation is seen in the areas of syntax and morphology. Deviations of the translators and their collaborators are evident in the misconstrual (or, at least, alternative construal) of Sanskrit
idioms, alteration of grammatical case and number, unwarranted ‘rectification’ of cited passages, and the like.
Finally, this conception fails to take cognisance of the fact that d) the
seeming uniformity of Tibetan translations is attributable neither to the
translators themselves nor to the restrictive imperial guidelines, but to
historical developments in canon and its associated technology (in particular the influence of xylography) which resulted in the disappearance of variant translations. That is to say, it is not the case that the
Tibetan translations are so uniform due to the uniformly extreme fidelity of their translators. Rather, what one sees is the result of an extrinsic
historical development, which resulted in the standardisation of the
Tibetan Buddhist canons.
With the work of assembling the Tibetan translations of Indian
Buddhist literature into the great collections of Bka’ ’gyur (containing
sūtra-s and tantra-s) and Bstan ’gyur (śāstra-s), a process was initiated
which ultimately led to the standardisation and univocality of canonical reference works in the latter half of the second millennium. While
the prototypical ’Jam Lha khang canon (early fourteenth century)
included several ‘redundant’ translations (i.e. included more than one
translation of any given Indic work), the subsequent editing of the
Bka’’gyur and Bstan ’gyur by Kun dga’ Rdo rje and Bu ston Rin chen
grub in the early-mid-fourteenth century eliminated those which had
been included previously. Those translations selected for inclusion in
the canonical collections assumed thereby a privileged status and soon
eclipsed the alternative translations that had not been so selected. This
process was accelerated considerably by the later adoption of the practice of block-printing (reaching its apogee with the xylographic canons
of the eighteenth century),7 which allowed the mass reproduction—and
7 According to Jackson 1983: 5, though the xylographic reproduction of Tibetan
books had started in Mongolia as early as the thirteenth century, “large printing projects were quite rare [in Tibet proper] until the fifteenth century”. The Bka’ ’gyur was
first set into a xylographic edition at the beginning of the fifteenth century (specifically, 1410), when the Peking (Yunle) edition was prepared. The Bstan ’gyur, on which we
will largely focus below, seems not to have been rendered xylographically until after
1683. Cf. Smith 2001: 312–13, fn. 551. As Gene Smith has further indicated (Smith
2001: 183), “Bu ston and his Zhwa lu redaction would come to exercise an overwhelming influence on all of the xylographic editions”.
TANTALISING TRACES
5
thus wider and easier availability—of the standard collections. With the
ready availability of the now ‘standard’ versions, manuscript copies of
the excluded works were no longer ‘needed’, and so were no longer produced. As a result, almost none of the excluded translations have come
down to us today.8 Thus, I would argue, the apparent uniformity of the
Tibetan translations of Sanskrit Buddhist texts is not (as usually maintained) exclusively—or even chiefly—the result of widespread standards of fidelity in translation. Rather, it is an epiphenomenon of factors which have led to there only being one canonical translation.9
In this light, I believe it behoves scholars of Tibetan literary history,
religion and culture to revisit this issue—to reassess the nature and status of Tibetan translations, their qualities and limitations, in light of
their original, historical polyvocality during the period between the
tenth and fourteenth centuries (gradually tapering off until the eighteenth, whereupon univocality quickly becomes the standard). Several
bodies of material which can shed light on the details of the processes
of Tibetan translation make this task possible, and closer attention to
these should allow this work to proceed. Some of these materials have
already been noted and exploited (at least for the light they shed on the
Indic source texts, if not the dynamics of Tibetan translation). For
instance, although Bu ston chose Jinākāra and Nag tsho Lotsāwa’s
translation of the Cittaviśuddhiprakara"a (Sems kyi sgrib pa rnam par
sbyong ba zhes bya ba’i rab tu byed pa) attributed to Āryadeva for
8 One notable exception is the Phug brag Bka’ ’gyur, which preserves many alternative translations not found in other canonical editions. See Samten 1992a, Samten
1992b, and Eimer 1993. Futher, Ronald Davidson indicated to me, during a discussion
between sessions at the 2003 Oxford IATS Seminar, that he had seen complete alternative translations conserved in the collected works of Sa skya translator/pa()ita-s. He
cautioned, however, that such ‘alternative translations’ frequently turn out to be little
more than recensions of earlier translations, lightly-revised or even merely reattributed.
9 In fact, it would seem as if, as time went by, the xylographic canons, in concert
with a certain Tibetan monastic conservatism, functioned to suppress more contemporary translations as well. As Yoshiro Imaeda has indicated, when Si tu Chos kyi ’byung
gnas undertook the preparation of a new xylographic edition between 1730 and 1733—
at least the seventh such xylographic print of the Bka’ ’gyur—he was constrained by
the objections of a certain ‘great abbot’ (mkhan chen) from including new translations
prepared by Tāranātha and himself. Imaeda also comments, however, that (according to
the dkar chag of the Rgyud sde kun btus) certain new translations of Bstan ’gyur materials were inserted into the Sde dge redaction in the nineteenth century by’Jam dbyangs
Mkhyen brtse’i dbang po. So, one may note that the practice of xylography seems to
have worked both ways—though I would argue that its exclusionary function far outweighed such rare moments of its transmitting new or alternative texts. Cf. Imaeda
1981: 229–30 and 234–35.
6
CHRISTIAN K. WEDEMEYER
inclusion in the (Zha lu MS) Bstan ’gyur10—thereby marginalising other
versions which may have existed in the Old Snar thang Bstan ’gyur
MS—somewhere in the transmission process of the Snar thang, an alternative translation of this work (by Dīpa*karaśrījñāna and Khu Dngos
grub) crept into a Snar thang xylographic version under the title Sems
rin po che rnam par sbyong ba.11 Both of these are available in a published romanised, as well as xylographic, form and are well suited for
undertaking the type of research and analysis I suggest below.12
However—turning at last to the central point of this paper—there are
other, less obvious, sources which may also be of use to scholars working in this area. I would here draw attention to a previously-overlooked
source of valuable materials in this regard: passages from alternative
translations which have survived the standardising processes of canon
formation and xylography through being incorporated as citations into
Tibetan scholastic treatises. Such citations seem to exist in the works of
several Tibetan authors. One source of such materials which has attracted my attention as of particular importance and scope, is that found in
the writings of Rje Rin po che Blo bzang grags pa, also known as Rje
Tsong kha pa. In what follows, I will explore some of the textual
remains preserved in Tsong kha pa’s œuvre, examining them and the
‘standard’ translations against the extant Sanskrit materials,13 and begin
10 As reported in the ‘catalogue section’ of his Chos ’byung: slob dpon ’phags pa
lhas mdzad pa’i sems kyi sgrib sbyong nag tsho’i ’gyur. See Nishioka 1983: 83.
11 This fact was first noted by Prabhubhai Patel in his posthumously-published
work Cittaviśuddhiprakara"a of Āryadeva (Patel 1949), wherein he makes use of both
translations. The inclusion of this text (and its consequent availability to modern scholarship) may perhaps be attributable to the fact that the work not only bears a distinct
title, but is ascribed in its colophon to the authorship of Indrabhūti, rather than Āryadeva. Bu ston may (like Patel) have noted this fact, and excluded the text, and a less astute
successor may have reintroduced it.
12 See, for example, Patel 1949: 175ff. Another important, potential source is the
criticism leveled by Ngor chen Kun dga’ bzang po against the Jo nang new translation
of the Kālacakra Tantra, commissioned by Dol po pa Shes rab rgyal mtshan. In this
work (Sa bzang bsod nams dpal gyi dris lan), Ngor chen compares this translation
against the ‘standard’ version. Cf. Stearns 1999: 189 fn 79.
13 There is a legitimate objection which may be raised at this point. It could reasonably be argued that it is not valid, in comparing the Tibetan translations against extant
Sanskrit texts, automatically to privilege the received Sanskrit texts as if they constituted a necessarily reliable standard. As has been noted more than once, since many
Buddhist Sanskrit texts have come down to us only in nineteenth-century copies, often
riddled with errors introduced over the course of time, they are not always more reliable witnesses to the state of the text in an earlier period than the Chinese or Tibetan
translations. These latter can sometimes bear witness to an older and more reliable state
of the text. It is thus problematical, when faced with a divergence between the Sanskrit
TANTALISING TRACES
7
to explore what they reveal about the diversity of translations in the
early Phyi dar period, discursive strategies of legitimation of authority
in fourteenth/fifteenth century Tibetan scholastic discourse, and the
benefit these sources provide for contemporary scholars of the textual
traditions of India and Tibet.
II. THE CORPUS
Before (and seemingly also during) the lifetime of Rje Tsong kha pa
Blo bzang grags pa (1357–1419)—a half-century after the redaction of
the Zha lu canon, yet over two centuries before the widespread adoption of block-printing—a variety of translations were still available to
Tibetan religious in manuscript form; and tantalising traces of these
‘alternative’ texts are to be found in his surviving works. While such
passages are found in both his exoteric and his esoteric writings, it is in
his writings on the esoteric Tantras that one finds the bulk of this material. In these works, for example, Tsong kha pa makes frequent reference to such alternative translations, often expressing a preference in
his exegesis for one or the other over the ‘standard’ translations of Śraddhākaravarman and Lo chen Rin chen bzang po.
Such citations are particularly widespread in his works on the
Guhyasamāja Tantric system, such as his magnum opus on Nāgārjuna
and Āryadeva’s presentation of the yogic system of five stages,14 his
pair of commentaries on two of the explanatory tantras (vyākhyātantra, bshad rgyud) of the Guhyasamāja system,15 and his voluminous
interlinear sub-commentary (mchan gyi yang ’grel) on the Pradīpodand the Tibetan, to presume a privileged status to the Sanskrit.
While this is no doubt true in many cases, in the discussion below we will be dealing with works for which earlier manuscript evidence exists and is quite consistent.
Most of the examples are drawn from the Caryāmelāpakapradīpa, a work which I know
quite well (having edited and translated it, see Wedemeyer 2006) and in which the two
extant Sanskrit MSS are in overall concord.
14 Rim lnga gsal sgron: Rgyud kyi rgyal po dpal gsang ba ’dus pa’i man ngag rim
pa lnga rab tu gsal ba’i sgron me; in Rje Tso# kha pa 1978c.
15 The commentaries on the Enquiry of the Four Goddesses Tantra (Caturdevīparip(cchā, Lha mo bzhis zhus) and the Gnosis Vajra Compendium (Jñānavajrasamuccaya, Ye shes rdo rje kun las btus pa): the ’Dus pa’i bshad rgyud lha mo bzhis
zhus kyi rgya cher bshad pa srog rtsol gyi de kho na nyid gsal ba, and the Dpal gsang
ba ’dus pa’i bshad pa’i rgyud ye shes rdo rje kun las btus pa’i )īkka; in Rje Tso# kha
pa 1977: 350–449 and 450–586, respectively.
8
CHRISTIAN K. WEDEMEYER
dyotana of Candrakīrti.16 In this latter work, reference to alternative
translations of the root text appear on virtually every folio—including
versions by ’Gos Khug pa Lhas btsas (fl. eleventh century), Chag Chos
rje dpal (1197–1264), and Pa tshab Nyi ma grags (b. 1055), in addition
to that of Lo chen Rin chen bzang po (958–1055), readings in independently-translated commentaries, and, apparently, several Sanskrit
manuscripts. As such, the corpus is quite large, and what follows—my
own first foray into mining the richness of this resource—will be necessarily tentative and incomplete.
Given the vast dimensions of the corpus, I will focus in what follows
on citations of alternative translations, and the use Rje Rinpoche makes
of them, in his major work on the yogic practice of the Guhyasamāja
Tantra, the Brilliant Lamp of the Five Stages (Rim lnga gsal sgron).17 In
this work, one finds roughly 108 distinct references to alternative translations of Indian tantra-s and śāstra-s in a work of 344 folios—that is,
approximately one every three folios. It thus falls somewhat in the middle between the extreme frequency of such citations in the mchan ’grel
on the Pradīpoddyotana and their more sparse citation in his other
works.
In all, in this text, eight works of the Guhyasamāja literature are
cited in alternative forms, five tantra-s and three śāstra-s. The former
include the Guhyasamāja Tantra itself, its ‘Appendix Tantra’ (Uttara
Tantra), the explanatory tantras Vajramālā and Sandhyāvyākara"a, and
the Sarvabuddhasamāyoga Tantra; the latter comprise the Pañcakrama
of Nāgārjuna, the Caryāmelāpakapradīpa of Āryadeva, and the
Pradīpoddyotana Commentary of Candrakīrti. The alternative translations cited are: Chag and Pa tshab’s translations of the Guhyasamāja
Tantra and its Uttara Tantra; Chag’s translation of the
Sandhyāvyākara"a; translations of the Vajramālā by Zhi ba ’od (and,
separately, its further revision by Dar ma brtson ’grus), Rwa Lo tswā
ba, a version nested in another commentary, citations found in the old
translation of the Caryāmelāpakapradīpa, and unnamed ‘others’; a
separate translation of the Sarvabuddhasamāyoga (in addition to its
many cited versions); translations of the Pañcakrama by Chag and
another ‘new’ translation (presumably by Pa tshab, but unspecified);
16 Sgron gsal mchan: Rgyud thams cad kyi rgyal po dpal gsang ba ’dus pa’i rgya
cher bshad pa sgron ma gsal ba’i tshig don ji bzhin ’byed pa’i mchan gyi yang ’grel, in
Rje Tso# kha pa 1978a and b.
17 See above, note 14.
TANTALISING TRACES
9
two ‘old’ and two ‘new’ translations of the Caryāmelāpakapradīpa (the
latter being those by Chag and Pa tshab); and versions of the Pradīpoddyotana by Chag, Pa tshab, and ’Gos Khug pa Lhas btsas. In addition,
and of particular interest, is the fact that Rje Rinpoche also refers to
several Sanskrit manuscripts (rgya dpe), including several of the
Pradīpoddyotana, two variant manuscripts of the Vajramālā, and at
least one each of the Uttara Tantra, Pañcakrama, and
Caryāmelāpakapradīpa. This is most intriguing, given the period and
the conventional wisdom that Tsong kha pa did not know Sanskrit, and
we will have occasion to remark on it below.
III. ANALYSIS OF SELECTED EXAMPLES
In his exegesis of the esoteric literature available to him, Tsong kha pa
indicates many of the typical problems addressed in classical textual
editing. He describes the following: bad Tibetan texts (bod dpe ma dag
pa, pp. 58 and 426), translations based on bad Sanskrit texts (rgya dpe
ma dag pa las bsgyur ba, p. 58), old scribal errors (bris nor ring du
brgyud pa, p. 156), as well as several forms of variant readings (pp. 367
and 425). Many of his criticisms are delivered ex cathedra, without justification, yet there are a significant number of cases wherein the relevant texts are cited, compared, and judged. In these instances, we have
the materials to evaluate in more detail the nature and qualities of the
cited translations and reflect on the use made of them by Tsong kha pa.
In the first example,18 Tsong kha pa cites two versions of two brief
passages from the Caryāmelāpakapradīpa—Lo chen’s and Chag’s. The
Sanskrit of both is clear and univocal in the extant MSS. The first reads,
“the one who understands isolation of body, speech, and mind, having
obtained the tenth stage, attains the phantasmical samādhi”; the second
is a dependent clause reading “having made manifest a buddha-body”.
Interestingly, both translators have rendered the (nominal) past passive
participle adhigata as a (verbal) gerund (suggesting a ‘reconstruction’
of *adhigamya),19 but the sense is not materially altered. There are only
two substantial differences. In the first passage, Lo chen has omitted
18 The complete, parallel texts of all examples discussed below, including the relevant comments of Tsong kha pa, are included in the Appendix.
19 In what follows, I use a prefixed asterix (*) to mark the hypothetical, ‘back-translated’, reconstructions I believe are implied by the cited Tibetan text.
10
CHRISTIAN K. WEDEMEYER
reference to ‘isolation’ and has rendered the ‘body, speech, and mind’
in the Tibetan honorific register. The latter is a stylistic choice and may
be passed over without comment. The former suggests either that Lo
chen was working from a Sanskrit text which lacked the reading viveka, that the term was accidentally (or intentionally) omitted by him, or
that the term rnam par dben pa was lost in the process of manuscript
transmission of his translation. It is impossible to adjudicate between
these hypotheses without further data.
We are on firmer ground, however, with the latter variant, i.e. ‘buddha deeds’ vs. ‘buddha body’. The surviving Sanskrit texts both read
buddhakāya; Lo chen’s rendering of this as ‘buddha deeds’ (sangs
rgyas kyi bya ba) indicates that he understood the text to read buddhakārya.20 While one must allow for the possibility of erroneous reading (and/or sloppy handwriting), this suggests evidence for a (potentially significant) variant reading which has not been transmitted by the
extant Sanskrit manuscript sources.
Tsong kha pa, in his comments, does not assert that Chag’s translation is better per se; his only comment is that Chag’s translation “clarifies the earlier translation” (’gyur snga ma gsal du btang ngo). This
suggests that he may have believed ‘isolation’ to be an interpolation by
Chag—that is, an interpretative translation on his part—rather than an
omission by Lo chen. In the second case, it is not at all clear to me how
reading ‘body’ clarifies the reading ‘deeds’. Either deeds or body
would be appropriate enough in this context, so the preference of one
over the other remains something of a mystery. We have here an example where the text Tsong kha pa prefers indeed appears to be superior,
though it is not clear upon what line of reasoning his preference
depends.
Our next example also lends us little or no insight into the editorial
criteria of Tsong kha pa, but it does provide a rather striking example
of the variability (and sometimes rather dubious reliability) of Tibetan
translations. In example two, drawn from the Pañcakrama (Rim pa lnga
pa) attributed to Nāgārjuna, the Sanskrit texts read “The forty prototypes are momentary and born from the extreme void” (catvāri*śat
prak(taya$ k+a"ikāś cātiśūnya-jā$ |).21 The ‘standard’ translation of Lo
20 Many thanks to Prof. Helmut Eimer and Dr Isabel Onians for stressing this point,
which I had overlooked in my original presentation.
21 Pañcakrama II.20cd; cf. Mimaki and Tomabechi 1994: 18.
TANTALISING TRACES
11
chen has several problems. For one, it renders the technical term prak(ti
by mtshan nyid (which usually translates lak+a"a and, to my knowledge,22 is nowhere else attested as an equivalent for prak(ti). Further, it
reads a relative pronoun (gang yin pa, *ya$) not found in the source;
and it reads the word ‘born from the extreme void’ (atiśūnyajā$) as a
genitive (shin tu stong pa’i, *atiśūnyasya? *atiśūnyatāyā$?) modifying
a nominal ‘moment’ (skad cig, *k+a"a), rather than as two independent, quasi-adjectival terms modifying the prototypes (prak(ti).23
In this context, Tsong kha pa cites the ‘alternative’ translation of
Chag. But is his translation any better? Chag does render the central
term prak(ti by the authorised term rang bzhin. However, his grammar
is equally—albeit differently—problematical. He successfully captures
the essence of the final, predicative term atiśūnyajā$. But rather than
signaling the predicative role of the term ‘momentary’ (k+a"ika), he
renders this as the primary (nominal plural) subject ‘moments’ (skad
cig rnams, *kśa"ā$?) modified by ‘forty prototypes’ in the genitive
(rang bzhin bzhi bcu’i, *catvāri*śata$ prak(tyā$? prak(te$?).
Thus, we have here a rather clear example in which two alternative
translations of a Sanskrit source text exist, each gives a divergent
impression of the original text, and yet neither adequately captures its
meaning (or grammar). Chag’s translation is clearly closer to the mark,
adequately translating the basic lexical units, though it too misconstrues the subtending grammar.24 Were a Sanskrit text not available,
‘back-translating’ either form of this half-verse would likely provide a
misleading sense of the Sanskrit original.
Tsong kha pa makes no explicit comment on the alternative text. In
his discussion, he cites a lengthy passage from the Pañcakrama (in Lo
chen’s translation) which presents the forty prototypes (prak(ti) associated with the luminance-radiance (ālokābhāsa) level of the subtle
mind, and discusses its significance. At the conclusion of this com22
It does not appear in any of the Tibetan-Sanskrit lexicons (Mahāvyutpatti, L.
Chandra, J.S. Negi), nor have I personally witnessed such a usage. Matthew Kapstein
(personal communication) believes, however, he has seen it elsewhere. Even if it is not
a unique case, this is nonetheless at least a rare, atypical usage and contrary to the
prevalent rendering of what is a central technical term of the Mahāyoga Tantras.
23 Though, as Matthew Kapstein has commented, “the relativising use of the genitive is a common way of dealing with appositional strings in Sanskrit” (personal communication), I do not find Lo chen’s use of it here effective as it rather loses the force
of the upapada compound.
24 One would prefer, for example: rang bzhin bzhi bcu skad cig pa | shin tu stong
pa las skyes te |.
CHRISTIAN K. WEDEMEYER
12
mentary, he says “with regard to the last two lines, Chag’s translation
reads...”.25 We may infer from the mere fact of his drawing attention to
this alternative text, that Rje Rinpoche here preferred Chag’s (marginally better) translation to that of Lo chen.
In example three, however, the situation is rather different. Tsong
kha pa ci tes t wo ver sions of a st atem en t fro m t h e Caryāmelāpakapradīpa, contrasting the Lo chen translation with the ‘two
new translations’ (’gyur gsar gnyis). In this case, it is fairly clear that
the latter are, in fact, not only preferable to the former, but actually
more successfully live up to the reputation of Tibetan translation. The
Sanskrit texts read: tilamātre+v api vastu+u parigraha-buddhi* parityajet | (“one should forsake the mind [desiring to] take possession of
objects even [those] merely the size of a sesame seed”). The two new
translations communicate this meaning quite nicely, with only the
expedient of changing the word order slightly. That is, the main clause
is intact and word-for-word (yongs su ’dzin pa’i blo dor bar bya’o). The
locative phrase at the beginning is modified slightly to better accommodate Tibetan idiom (though it loses the plural number), reading: dngos
po til ’bru tsam la yang.26
Lo chen’s standard translation takes a different approach, being more
rigorous with regard to preserving the word order, yet less rigorous
with regard to preserving Sanskrit prefixes (upasarga-s).27 The main
verb is rendered by the equivalent spangs par bya’o, rather than dor bar
bya’o. While this is unobjectionable, the manner in which the remainder of the sentence is translated causes problems for Tibetan understanding. Following the Sanskrit word order precisely, Lo chen has rendered the object phrase as follows: til ’bru tsam la’ang dngos por ’dzin
pa’i blo. This is a more mechanical rendering and—except for the loss
of the plural number and the prefix pari—it may be more easily ‘backtranslatable’. However, the loss of the plural number (and the attendant
use of the locative ending r, rather than a locative particle) creates
major difficulties for the interpretation of the translation in a
Tibetophone context. The instinct of a reader is to construe dngos por
as subordinate to ’dzin pa, rather than construing it with til ’bru tsam
25
rkang pa tha ma gnyis la chag ’gyur las | (Rje Tso# kha pa 1978c: 424).
This is likely merely an accommodation to Tibetan idiom (like my own English
rendering, which also uses the singular). Nonetheless, it is important to note that it suggests a ‘back-translation’ of *vastuni tilamātre ’pi.
27 For alternative examples of this see above, note 4.
26
TANTALISING TRACES
13
la’ang.28 Thus, a Tibetan would interpret this to mean, “one should forsake the mind which grasps at the [substantial] reality of even a sesame
seed”.
This is, indeed, how Tsong kha pa interprets this passage in Lo
chen’s translation; and it is on this basis that he criticises it. Given these
two slightly different translations—“one should cast off the mind
[desiring to] take possession of objects even the size of a sesame seed”
and “one should abandon the mind which grasps at the [substantial]
reality of even a sesame seed”, Tsong kha pa comments,
Concerning the second [of the four distinctive intentions], since the
translation found in the two new translations, to wit “one should forsake
the mind [desiring to] take possession of objects even the size of a
sesame seed”, is better, one should take this to refer to not taking, i.e. not
accumulating, goods even [the size of] a sesame seed. One should not
take this to refer to ‘object grasping’ in the sense of conceptual insistence
on reality.
In this case, I believe we can reconstruct the reasoning behind Rje
Rinpoche’s comments. Although, again, it is not explicitly so stated,
one can with some confidence infer that Tsong kha pa was basing his
editorial judgment on the context of the passage. The other three of the
four ‘special intentions’ under discussion are: to abandon having (conventionally speaking) a good time and to meditate on the suffering of
the (conventionally) happy; to give up concern for life and limb; and to
give up desire for siddhi-s. Thus, all four are oriented around sacrificing things which, conventionally speaking, are desired: fun, possessions, a nice body, and superpowers. In this list, grasping at substantial
existence does not seem to fit. It is on this basis—rather than, say,
philological concerns—that, I believe, Tsong kha pa preferred one over
the other.
In our next example (no. 4), drawn once more from the Pañcakrama,
we again find Rje Rinpoche citing Chag’s translation—implicitly
endorsing it over that of Lo chen. He does not cite Lo chen’s text, but
merely gives Chag’s translation in the context of describing the central
idea of the Tantric goal of communion (yuganaddha or zung ’jug). If
we measure the two versions against the Sanskrit text, we find mixed
results.
28
This was, for example, my own experience when I first translated this text, without reference to either Tsong kha pa’s comments or the Sanskrit texts. Cf. Wedemeyer
1999: 347.
14
CHRISTIAN K. WEDEMEYER
In the first of the two verses cited, Chag’s translation is clearly better. Lo chen misinterprets the dvandva compound svādhi+)hāna-prabhāsvarau as a genitive tatpuru+a (bdag byin brlab pa’i ’od gsal)—
very strange given its dual ending. He further misconstrues tayo$ as
locative (de nyid la), rather than genitive (*de nyid kyi)—possible, but
highly improbable. He does, however, preserve the gerund form of the
verb to know (jñātvā shes nas). Chag loses the gerund—putting the
verb ‘to know’ in the indicative—and leaves the genitive implicit (de
dag kho na rather than de dag kho na’i) but correctly construes the rest
of the grammar.
The second verse is another story, however. Here, the last two lines
are identical. Both translators lose the optative sense of the Sanskrit
bhavet—obviating accurate ‘back-translation’, but capturing the sense
adequately enough. The translations diverge in their renderings of the
first two lines. The Sanskrit text begins with the two nouns sa*v(ti and
paramārtha (‘conventional’ and ‘ultimate’ [reality])—each in the accusative singular—joined by the conjunctive particle ca. The second line
begins with the indeclinable p(thak (‘separately, individually’), continues with the gerund jñātvā (‘having known, knowing’) and concludes
with vibhāgata$—that is, the term vibhāga (‘division, section’) with an
ablative suffix—in this case, presumably, giving an adverbial significance (‘proportionately, in their own measure’).
Here, Chag correctly translates jñātvā as a gerund (shes nas), yet
construes vibhāga (rnam dbye) as its object and puts conventional and
ultimate reality in a genitive relationship to it, while translating p(thak
(suitably enough) as so sor. Lo chen on the other hand, while altering
the gerund, successfully interprets conventional and ultimate reality as
the objects of shes pa, marking them with the dual particle dag. I
believe he also seeks to capture the adverbial import of vibhāgata$ with
the phrase so so’i char.
Thus, I believe that—while Tsong kha pa was correct to prefer
Chag’s translation of the first verse—with the second verse, he should
have stuck with good old Lo chen. Both translations reveal idiosyncratic choices on the part of their translators. Indeed, both reveal their
translators failing reliably to construe the grammar of the Sanskrit texts
on which they were working. Neither would allow consistently accurate
reconstruction of the original Sanskrit.
The next pair of translations (see no. 5) is quite interesting. The pas-
TANTALISING TRACES
15
sage is from the famous description in the Caryāmelāpakapradīpa of
the enlightenment process of Buddha Śākyamuni.29 Lo chen’s translation reads “having arisen from the phantasmical samādhi, [the Lord]
taught beings”. Chag’s translation is more involved. It reads, “having
arisen by means of the phantasmical samādhi, [the Lord] turned the
wheel of Dharma for beings who were to be disciplined”. How to evaluate these two versions?
As it turns out, these two Tibetan translations correspond (with
some very minor variants) to the divergent readings of the two extant
manuscript witnesses of the Caryāmelāpakapradīpa.30 The opening,
gerundive phrase is consistent between the two; the object of the gerund
being in the accusative in the Sanskrit, the variation between the ablative and the instrumental in our two Tibetan versions are merely two
idiosyncratic attempts to render the phrase in idiomatic Tibetan. The
main clause of the sentence, however, diverges in the two manuscripts.
Lo chen’s version corresponds in the main to the text found in the
Calcutta-Kathmandu MS, which reads janebhyo dharma* pravartitavān ([the Lord] set forth the Dharma for beings). Lo chen cut some
corners in rendering ‘set forth the Dharma’ as ‘taught’, but it seems
clear that the text he had before him read something like this one.
Chag’s version, on the other hand, corresponds exactly to the text found
in the Ngor manuscript, which reads vineya-janebhyo dharma-cakra*
pravarttitavān.31
Thus, in this case, the two divergent translations given in Tsong kha
pa’s comment reflect two divergent states of the text—presumably
reflecting two divergent manuscript traditions. Neither, then, is
29 ‘Famous’ among the Tibetan religious élite, at least. Mkhas grub rje considers
this to be the locus classicus for the ’Phags lugs understanding of the “method by
which the Teacher, the Lord, became enlightened” (ston pa bcom ldan ’das sangs rgyas
tshul). Cf. Lessing and Wayman 1968: 34.
30 These MSS are composed of three manuscript parts. The first MS consists of two
halves of one (nearly complete) palm leaf manuscript, now divided between the Asiatic
Society of Bengal and the Nepalese National Library. The second MS is available as a
nearly-complete photograph (taken by Rahul Sā#k$tyāyana) of a text in the library of
Ngor monastery in Tibet. The photograph is now in the archives of the Bihar Research
Society, Patna. For more information, see Christian K. Wedemeyer, Āryadeva’s Lamp
that Integrates the Practices (Caryāmelāpakapradīpa): The Gradual Path of Vajrayāna
Buddhism according to the Esoteric Community Noble Tradition (forthcoming [2006])
for more information on the manuscripts, a critical edition of the original Sanskrit and
Lo chen’s Tibetan translation, an annotated English translation, and textual analysis.
31 Actually, it doesn’t correspond exactly to the Ngor text. The MS reads vijaneyajanebhyo. However, this can be emended unproblematically to vineya-janebhyo.
CHRISTIAN K. WEDEMEYER
16
‘wrong’—though Lo chen’s is less exacting than Chag’s. In this case,
one could very well envision an accurate reconstruction of (one of) the
Sanskrit text(s) from Chag’s version. Tsong kha pa does not comment
on these texts, does not express a valuation of one over the other. He
merely notes the variant. One may presume that he preferred Chag’s for
its greater specificity.
The next example (no. 6) is extremely intriguing for the light it sheds
on the vagaries of Tibetan translation. The two translations, again Lo
chen and Chag versions of the Caryāmelāpakapradīpa, reveal two translators parsing a Sanskrit sentence in two completely divergent ways. The
Sanskrit text reads: yannāmarūpātmako mahāvajradhara iti |. Tsong
kha pa prefers Chag, for the stated reason that it better conforms to the
exegesis given in the Moonlight Commentary (’grel pa zla ’od); and,
indeed, at first glance, it may seem preferable. Taking nāma-rūpa as a
typical compound (which it undeniably is), Chag seems to have parsed
the sentence as follows: yan nāmarūpātmako mahāvajradhara iti, reading gang ming dang gzugs kyi bdag nyid can rdo rje ’chang chen po |
(“that which has the nature of mind and matter [nāma-rūpa] is
‘Mahāvajradhara’”). Lo chen, on the contrary, parses the sentence as follows: yannāma rūpātmako mahāvajradhara iti, reading gzugs kyi bdag
nyid can gyi ming ni rdo rje ’chang chen po zhes bya’o | (the name of
that which has the nature of form is ‘Mahāvajradhara’). In so doing, he
construes the grammar to reflect the Sanskrit idiom yan-nāma (having
which name).32
However, both renderings (at least as I have construed them above)
take some liberty with the Sanskrit grammar.33 In particular, they
ignore the gender (not marked in Tibetan) of the pronoun yat. Whether
taken as a relative pronoun in apposition with nāmarūpātmaka$ (as
Chag seems to do), or understood to function idiomatically with the
indeclinable nāma (as per Lo chen), yat ought to be inflected in the
masculine singular, i.e. yo (ya$). In the neuter singular, as we find it
here, I believe yat can only be understood as an indeclinable in which
capacity, as V.S. Apte notes, yat “is frequently used...in the sense of
‘because’, ‘since’”.34 This reading is, I believe, supported by the context of the passage in the Caryāmelāpakapradīpa.35
32
Monier-Williams [1899] 1990: 844.
I thank Professor Gary A. Tubb, of Columbia University, for very helpful clarifying comments and suggestions on this point.
34 Apte [1890] 1998: 1303.
35 Caryāmelāpakapradīpa, Chapter VIII: jalān mīnam iva suptaprabuddham iva
33
TANTALISING TRACES
17
If this reading is correct, Lo chen’s translation may be confidently
declared misguided. What, however, of Chag’s? There is a certain sense
in which Chag’s rendering could be understood as, in fact, correct.
Each element of the Sanskrit clause is rendered in the Tibetan text in an
exacting way; and, as Tibetan does not mark gender, we have no way of
determining whether Chag intended gang to imply ya$ or yat. That
said, however, in a Tibetophone context, the translation could only have
been (mis-) construed as we have done above (i.e. reading gang as ya$).
To my knowledge, gang does not function in Tibetan as an indeclinable
in the same way as yat does in Sanskrit; it is only and always a relative
(or interrogative) pronoun. In the final analysis, there is no compelling
evidence that Chag understood the grammar of the sentence. If he did,
he was clearly less concerned about whether that meaning would be
understood by a Tibetan reader than with mechanically rendering the
lexical elements, preserving even the word-order. His rendering may be
‘exact’ in a certain technical sense but, read as natural Tibetan, it does
not accurately convey the meaning of the passage.
In example seven, Tsong kha pa notes two variant versions of a verse
from the Guhyasamājottaratantra. The verse describes the breath as
composed of the five gnoses and as the nature of the five elements;
which breath should be emitted in the form of a drop and fixated either
on ‘the tip of the nose’ (as found, according to Rje Rinpoche, in Lo
chen’s translations of the Pañcakrama and Caryāmelāpakapradīpa, the
two old translations of Chapter 12 of the Pradīpoddyotana, and some
Indian manuscripts) or, ‘on the tip of the lotus nose’ (the reading
allegedly found in Chag’s translation of the Pañcakrama and
Caryāmelāpakapradīpa, all translations of Pradīpoddyotana Chapter
6, the two new translations of Pradīpoddyotana Chapter 12, and some
other Indian manuscripts). Tsong kha pa expresses his preference for
the former rendering, commenting:
Although it is clear that both variants occur in the Indian texts, because
the practice of vajra recitation of the three-syllabled (yi ge gsum,
tryak+ara) on the winds of the four ma()alas at the upper nose is taught
in the Caryāmelāpaka by means of the six-lined verse ‘ye shes lnga’i’
and so on, [the version] without ‘lotus’ (padma) is manifestly [the] correct [one].
paramānanda-mūrti-svarūpa* ni+padyate | yan nāmarūpātmako mahāvajradhara iti |
sa*sāra-bandhanān nirmuktatvān mok+a ity ucyate |. The last clause uses the ablative
to convey this sense of ‘because’, ‘since’. Our clause, the penultimate, uses the indeclinable yat to communicate the same sense.
18
CHRISTIAN K. WEDEMEYER
What is noteworthy here is that both readings are, in fact, to be found
in the surviving Sanskrit manuscript record. However, the reading
without padma is only found in the manuscripts of the Uttaratantra
itself, and the reading with padma is not attested in any extant text of
the Uttaratantra. Conversely, however, the reading with padma is the
only reading found in the surviving manuscripts of the commentarial
literature—the Pañcakrama, Caryāmelāpakapradīpa, and Pradīpoddyotana. What does this mean? Tsong kha pa’s comment that both readings are clearly found in the Sanskrit texts would suggest that this is the
source of the discrepancy—that we have here another example of two
independent lines of manuscript transmission. How, then, are we to
interpret the consistent difference in the Sanskrit MSS of the source
texts? The answer—which contradicts Tsong kha pa’s testimony—is, I
believe, to be found in what we know of Tibetan translational practice.
As the excellent work of Anne MacDonald on the translation of
Candrakīrti’s Prasannapadā has demonstrated,36 it was not uncommon
for Tibetan translators—when faced with a quotation from a sūtra or
śāstra in a work they were translating—to ‘cut and paste’ the translation of the cited passage(s) from a previous Tibetan version of that
scripture, rather than to translate the citation afresh from the manuscript before them. I have myself independently confirmed this based
on Lo chen’s translation of the Caryāmelāpakapradīpa, in which—
rather than translating a verse as found nested in the source text (also a
citation of the Uttaratantra)—he inserted a translation of the verse as
found in his earlier translation of the scripture cited, even though the
change made the entire passage a non sequitur!37 I suspect, then, that
what we have here is such a case of textual eclecticism in the four trans36 This work was initially presented at the XIIIth Conference of the International
Association of Buddhist Studies, Bangkok, Thailand, under the title, “From
Prasannapadā to Tshig gsal: Remarks on an 11th/12th Century Translation Project”.
More recently, notice of this feature has appeared in article entitled “Interpreting
Prasannapadā 19.3–7 in Context: A Response to Claus Oetke”. See MacDonald 2002,
and MacDonald 2003: 163–64. Further details are forthcoming in a book on
Prasannapadā Chapter One. Many thanks to Dr MacDonald for providing this reference.
37 This occurs in Lo chen’s translation of Chapter Four of the
Caryāmelāpakapradīpa (Spyod pa bsdus pa’i sgron ma). After a longish discussion of
the doctrine of the eighty ‘prototypes’ (prak(ti, rang bzhin) encoded in the subtle mind,
Āryadeva provides a scriptural verse to give a stamp of authority to his exposition. He
cites the following verse: “Whatever is born from conditions, By means of sense-organs
and objects, is mentation. That mind [is] called ‘eighty’. The syllable ‘tra’ is derived
from ‘protection’ (trā"ana)”. The extant Sanskrit (which makes perfect sense in this
TANTALISING TRACES
19
lations without padma, three of which are the work of Lo chen. This
would, of course, cast doubts on Tsong kha pa’s information about the
Sanskrit manuscripts. Given the preponderance of evidence, I find the
claim that “both [readings] occur in the Indian manuscripts” (rgya gar
gyi dpe la yod med gnyis byung ba) highly unlikely.
Furthermore, from our perspective, Tsong kha pa’s editorial preference appears not only to be mistaken with regard to the correct reading
of the Pañcakrama, but to be rather manifestly based on circular argumentation. For the version of the Caryāmelāpakapradīpa to whose
authority he refers in arguing for the reading without padma, is none
other than the Lo chen redaction—in both of which texts, Lo chen substituted the ‘standard’ reading of the source tantra for the idiosyncratic
reading of the commentarial śāstra. To so argue for the rectitude of the
Lo chen translation (of the Pañcakrama) based on the testimony of one
of the very texts whose reading is in doubt (the Lo chen translation of
the Caryāmelāpakapradīpa) seems methodologically problematical,
and evidently led our author astray.38
In our next and final example, one sees another clear case where the
preference of one translation over another went terribly wrong based on
presumably dubious grounds. This passage (see no. 8) presents the two
Tibetan texts given by Tsong kha pa alongside the extant Sanskrit text.
While the Sanskrit text is quite clear, the Tibetan translations (that by
Lo chen and the ‘two new’ translations) diverge quite remarkably. The
two new translations, while not entirely ‘exact’, are rather close to it.
There are two minor variants,39 but the sense is intact: that is, “one
should practise the practices completely without elaboration by the
insane penance according to the process expressed in the explanatory
tantra(s)”.
context) reads: pratītyotpadyate yad yad indriyair vi+ayair mana$ | tan manas tv aśītikhyāta* tra-kāras trā"anārthata iti || (cf. Wedemeyer forthcoming [2006]). However,
Lo chen—instead of translating this verse as found—inserted the translation of the
‘standard’ version of this verse from the Uttaratantra, although this latter verse, which
does not mention the number eighty, is thus a non sequitur in this context.
38 Also at play here, presumably, is Tsong kha pa’s signature view on Tantric practice in which prā"āyāma may be practised at a variety of ‘nose-tips’ including, especially, the ‘heart nose’, and not merely the ‘lotus nose’, presumably located in the groin.
39 They read tshul dang rim pa (‘method and process’) where the Sanskrit reads
merely ‘process’ (krama); and they omit the term ‘practices’ (spyod pa, caryā) in the
expression ‘practices completely without elaboration’ (Skt. atyanta-ni+prapañcacaryā, Tib. shin tu spros pa med).
20
CHRISTIAN K. WEDEMEYER
Lo chen’s translation, however, conveys an entirely different sense.
For one, the main predicate is not ‘should be practiced’ (caritavya*,
spyad par bya), but ‘should be explained’ (bshad par bya, *vācayitavya*?).40 The object is the same ‘practices completely without elaboration’, but the other two parts of the sentence (which are actually separated in the Sanskrit text by the object) are integrated into the phrase
“according to the process expressed in the Scripture of the Mad
[Spiritual] Discipline” (smyon pa’i brtul zhugs bshad pa’i mdo las
gsungs pa’i rim pa). In so doing, Lo chen has, in essence, created a reference to an Indic scripture which never existed.
Tsong kha pa comments as follows:
The extensive citation in the Caryāmelāpaka that begins “I shall explain
the practices completely without elaboration according to the process
expressed in the Scripture of the Mad [Spiritual] Discipline”, is from the
Sixteenth Chapter of the Vajrama",alāla*kāra Tantra. Concerning this,
although the two new translations of the Caryāmelāpaka read “one
should practice the practices completely without elaboration by the mad
[spiritual] discipline according to the process expressed in the explanatory tantra”, Lo chen’s translation “Scripture of the Mad [Spiritual]
Discipline” is better. That chapter teaches the practices of [spiritual] discipline, which is also called ‘mad [spiritual] discipline’, so the ‘explaining’ means ‘explaining the mad [spiritual] discipline’ and it does not
indicate an explanatory tantra of the Guhyasamāja.
To understand this judgment, one needs to refer back to the extensive
introductory section of the Rim lnga gsal sgron, in which Tsong kha pa
surveys the literature of the Guhyasamāja, and devotes considerable
attention to the issue of explanatory tantras (bshad rgyud). He ends up
basing his enumeration primarily on whether or not works are explicitly cited as such in the central works of Nāgārjuna, Āryadeva, and
Candrakīrti upon which he relies so heavily in general. However, in this
discussion, in which he concludes that the Vajrama",alāla*kāra does
not so qualify as an explanatory tantra (which conclusion is subsequently used, as we have seen, to invalidate the ‘two new translations’),
he has apparently not considered the testimony of the new translations
40 This variant could very well be the result of a corruption in the process of transmitting the text of Lo chen’s translation. Spyad and bshad are similar in sound and such
shifts are understandable in manuscript transmission. If so, this corruption had presumably crept in before the time of Tsong kha pa (if we exclude the possibility of his own
work having been altered by later editing to conform to the received text of Lo chen).
TANTALISING TRACES
21
of the Caryāmelāpakapradīpa, but based himself solely on Lo chen’s
redaction. Thus, once again, we see Tsong kha pa’s exegesis reflecting
a marked preference for Lo chen, based upon an implicit privileging of
his work. We may infer here that the Lo chen corpus had already
acquired a distinctive renown in the Tibetan world, such that his works
were considered ‘standard’ (by at least one major thinker) already by
the early fifteenth century.41
IV. SUMMARY REFLECTIONS
As we have observed in the foregoing, in early second-millennium
Tibet (ca. tenth to fifteenth centuries), a range of alternative translations of Sanskrit scriptural works had been produced and were available
to Tibetan religious. This corpus—though ostensibly constrained by
rigorous standards of consistency—was equally (or nearly) as diverse
as the cacophony of competing translations of the Mūlamadhyamakakārikā which confronts the contemporary (esp. Anglophone) student of Buddhism. To overlook these translations and their authors is to
disregard an important aspect of intellectual and spiritual life in Tibet
and to misunderstand the very processes which enabled Buddhism in
Tibet as we know it to develop and thrive. In the following pages, I will
indicate in brief three areas to which this evidence makes a substantial
contribution to scholarly understanding, noting as well several avenues
for future research.
First, it may be observed that the evidence of these alternative translations establishes quite clearly that—even working under the exacting
normative regimen of the standards prescribed in the glossary and
handbook of the imperial translation system—the products of individual Tibetan translators exhibit a wide range of variant renderings of
their Sanskrit originals. Though one finds only slight variation in the
Tibetan words used to translate Sanskrit lexical items (e.g. either
mtshan nyid or rang bzhin for prak(ti), variability with regard to syntax
and morphology is legion. One sees widespread license being taken
with the rendering of verbal forms (past passive participles being ren41 It is worth noting that, as Ronald Davidson has indicated, Tibetan tradition continued to refer to the Lo chen translation of the Mañjuśrīnāmasa*gīti, even though this
version was excluded by Bu ston from the Zha lu Kanjur in favour of its extensive revision by Blo gros Brtan pa. See Davidson 1981: 13 (incl. fn. 38).
22
CHRISTIAN K. WEDEMEYER
dered as gerunds [ex. 1], gerunds rendered in the indicative [ex. 4],
etc.). One witnesses all manner of deviation in the rendering of inflected nominal forms (e.g. nominative plural rendered as genitive singular,
ex. 2), adjectival forms being rendered as nominal (k+a"ika as *k+a"a,
ex. 2), inflected objects being rendered as if they were in compound
(vastu+u, ex. 3), loss of upasarga-s ([yongs su] ’dzin pa for parigraha
[ex. 3]), misconstrual of compounds (dvandva for tatpuru+a), and the
like. One also notes divergent strategies for parsing Sanskrit sentences
(ex. 7). At times, one or the other of the alternative translations seems
to be more accurate. At others, neither seems adequately to render
either the words or the sense of the Sanskrit.
Hence, the widespread notion of the ‘painstaking accuracy’ or the
‘meticulously faithful rendering’42 of Sanskrit originals produced by
Tibetan translators ought accordingly to be qualified. As suggested
above, depending upon one’s perspective, this conception either a) is
quite simply mistaken, or b) may more profitably be viewed as an
epiphenomenon of the distinctive historical process of canon formation
in Tibet and its attendant technologies. The attempt to impose standards
of consistency by the ninth-century imperial court attests ever-so-clearly to the existence at that time of divergent translations. We may confidently conclude from the evidence of the translations later produced
under the guidelines of this redactional régime that it also represents a
wholly theoretical ideal incapable of realisation due both to the
vagaries of language and to the humanity of translators.43
Furthermore, it is important to note that the very existence of such
variant texts was itself not historically neutral. Rather, as they apparently constituted one vehicle of discursive struggle in Tibetan religious
culture, the historian of religions overlooks them at her own risk. As
42 Cf. Snellgrove concerning the Tibetan translators: “every one of their texts is an
extraordinary linguistic feat, for no other translators have ever succeeded in reproducing an original with such painstaking accuracy. Relying upon them alone, there is no
reason why the exact contents of any Buddhist text should not become known to us”.
Snellgrove 1959: viii. Or, somewhat later, Gene Smith’s remark: “When foreign scholars first compared some of the Tibetan translations of Sanskrit texts found in the Bka’
’gyur and Bstan ’gyur, they were impressed by the meticulously faithful rendering that
the translators achieved”. Cf. Smith 1959: 181.
43 Or, rather, as it actually enjoins a wholly-realistic syntactic flexibility on the
would-be translator (as indicated in note 3 above), even the imperial system itself does
not in fact correspond to the ideal practice as envisioned by many modern commentators. Contemporary translators of Tibetan Buddhist texts—who might otherwise be
tempted to emulate the imagined practices of the great Tibetan lotsāwas—would be
well-advised to bear this observation in mind.
TANTALISING TRACES
23
can be seen from their citation in the writings of Tsong kha pa, the existence of alternative renderings of authoritative texts provided an avenue
by which those advancing competing interpretations could establish
their case. Tsong kha pa is not alone in citing alternative translations—
this also being an exegetical strategy employed by Dol po pa Shes rab
rgyal mtshan (1292–1361), Ngor chen Kun dga’ bzang po (1382–1456),
and numerous others. Hence, I believe it is fair to say that such citations
reveal a practice distinctive of Tibetan scholasticism between the thirteenth and fifteenth centuries, the study of which will shed further light
on the dynamics of scholastic discourse in this period.
What’s more, the citation of such alternative texts and the attendant
acknowledgement of the insecurity of the readings found in texts presenting authoritative works, raises further questions of a more general
nature concerning the existence and/or nature of a tradition of editorial
practice in traditional Tibet. They indicate and assume both an awareness of textual diversity and an acknowledgement of the legitimate
claims of competing voices.44 Though there is evidence of a similar
practice of textual criticism among contemporaneous Indian critics,45 it
is not as clear to what extent a developed tradition may have existed in
Tibet. Tsong kha pa, albeit a rather unsuccessful editor himself (for reasons we will return to briefly below), seems nonetheless to indicate
something of the indigenous Tibetan practice of textual editing.
Elsewhere, awareness of—and editorial action regarding—alternative
manuscript readings is in evidence around this period of Tibetan history. For instance, David Jackson has drawn attention to the work of
Gong dkar ba (1432–1496) in editing and xylographing the Sa skya
bka’ ’bum, asserting that “one of Gong dkar ba’s goals in sponsoring
the printing must have been to get beyond the errors of scribes by establishing and disseminating a standardised text for these crucial works”.46
However, aside from a brief reference by Pieter Verhagen to a forth44 At the same time, however, the trend of Tsong kha pa’s preferences also suggest
that—although the process of xylography had not yet dramatically changed the textual
landscape in Tibet—the Lo chen corpus had already by the early fifteenth century
become the ‘standard’ against which the ‘others’ were judged, presumably reflecting
the already growing influence and prestige of the manuscript Bka’ ’gyur(s) and Bstan
’gyur(s).
45 Thus, for instance, it is not uncommon to find references to variant readings
(pā)hāntara), interpolations (prak+ipta), wrong readings (apapā)ha), and the like
among Indian commentators. Cf. Sarma 1982: 281–88. See also von Hinüber 2000: esp.
p. 25.
46 Jackson 1982: 16.
24
CHRISTIAN K. WEDEMEYER
coming article by Leonard van der Kuijp which allegedly explores the
(apparently more successful) practice of textual editing by Skyogs ston
Lo tsā ba Rin chen bkra shis (b. fifteenth century),47 I know of no work
which has addressed this subject in any detail.48 If there is one striking
thing about Tibetan high culture, it is a kind of hyper-textuality. Integral
as editorial practice is to handling text, it would seem that a closer
attention to the characteristics of indigenous Tibetan textual criticism is
a desideratum.
Second, this corpus of citations also provides some insight into the
more narrow question of Tsong kha pa’s use of these alternative texts,
his command (if any) of the Sanskrit language, and the discursive
strategies in play in his writings. On first glance, his works give the distinct impression of a scholar who has a working command of Sanskrit,
such that he can judge the relative quality of alternative translations, not
merely on the basis of exegetical expediency, but on linguistic grounds
and in light of Sanskrit textual materials (i.e. rgya dpe). He cites, for
example, various terminological discrepancies: e.g., the translation of
dhana*jaya as either gzhu las rgyal or nor las rgyal (Rim lnga gsal
sgron, p. 249); and divergent renderings of the term śyāma as either
ljang gser or ljang sngon (ibid., p. 257). In one instance (p. 423), he
recognises that the variant readings lhan skyes and tsham tshom med pa
are based on a confusion of sahasa (sic for sāhasa, the correct reading)
with sahaja. He notes that this former term may be validly rendered by
either tsham tshom med pa (as Chag renders it) or dka’ ba la sbyor ba
(as do the ‘old translations’), thus suggesting some familiarity with
Sanskrit-Tibetan translational idiom.49 However, he elsewhere (p. 424)
equates the term rang bzhin (in its Tantric usage) with svabhāva, rather
47 Cf. Verhagen 1996: 279 wherein he cites “a forthcoming article [by] Leonard van
der Kuijp” which “describes a unique document providing evidence for the fact that
textual criticism was still being applied to the translated literature, even at that time
[late fifteenth/early sixteenth centuries]”.
48 As this article was nearing completion, I was delighted to hear that Prof. Kurtis
Schaeffer has written on the principles of textual editing set forth by Bu ston Rin po
che, and employed in editing the Bka’ ’gyur. See Schaeffer 2004. He has also written
on editorial practices involved in the production of one of the xylographic canons
(Schaeffer, unpublished [1998]) and a Tibetan ‘collected works’ project (Schaeffer
1999).
49 He notes, however, that his discussion is based on the comments of ‘earlier scholars’ (sngon gyi mkhas pa dag).
TANTALISING TRACES
25
than prak(ti—portending perhaps a closer familiarity with
Madhyamaka Sanskrit than Mahāyoga-Tantric Sanskrit.
Nonetheless, I believe the materials I have examined above lend
strong support to the view that Tsong kha pa did not, in fact, have facility with Sanskrit, nor did he consult the Sanskrit manuscripts he cites
(as indicated, for instance, by the testimony of no. 7 above). While there
are examples wherein the translation he prefers is in fact better (nos. 1,
2, 3, 4a), there are also many examples in which this is not the case; and
even some of the former reveal that extra-linguistic factors were guiding his choice (e.g. no. 3). There are other places in his work, furthermore, where his preference cannot possibly be based on philology per
se, but rather must be based on purely æsthetic judgment. For example,
his preference (p. 671) for Pa tshab’s a ra li rdzogs par byed over Lochen’s a ra li sgrub par byed, for Sanskrit āralli$ sampādayati cannot
be adjudicated on other than æsthetic grounds. In the causative, the
Sanskrit root pad, coupled with the prefix sam, has a range of meanings, comprehending the semantic range of both sgrub par byed and
rdzogs par byed. This suggests that Tsong kha pa’s preferences did not
proceed from a thorough knowledge of Sanskrit or Sanskrit-Tibetan
translation conventions.
Thus, I believe we are entitled to conclude that the copious citations
of alternative translations, and the spurious claim of reference to
Sanskrit texts, reveal not the workings of a philologically-based editorial practice, but rather a rhetorical strategy operative in Rje Rinpoche’s
esoteric writings in which, following the pattern established by figures
such as Sa skya Pa()ita (1182–1251/2), Tsong kha pa sought to validate
his mode of exegesis, and to establish his credentials as a pre-eminent
teacher, by a claim of privileged access to sources of exegetical legitimacy. The religious landscape of the Phyi dar period was largely
defined by competing claims to religious authority. In this context, one
approach to legitimating such authority was to attempt to demonstrate
a superior degree of knowledge of, or other relationship to, the Indic
tradition—the ultimate source of religious legitimation in this context.
As Georges Dreyfus has indicated, “[Dge lugs] scholasticism...has
tended to present itself in Tibet as the inheritor and sole legitimate
interpreter of the classical Indian Buddhist tradition”.50 I think it is fair
to say that this tendency is already manifest in the formative works of
its founder, Rje Rinpoche.
50
Dreyfus 2003: 148.
26
CHRISTIAN K. WEDEMEYER
The sheer mass of references to alternative translations, Sanskrit
texts and terms, provides a prima facie case that this feature of Tsong
kha pa’s writing is in no way peripheral to his literary persona. Rather,
he seems to be making a rather strong statement about the type of tradition he advocates and its sources of authority. Like Sa pa(, who critiqued his contemporaries for their provincial Tibetan ignorance (himself professing an intimate claim to knowledge of the original Indian
sources), the conceit of precision of translation as setting his work apart
from the run-of-the-mill, sloppy old Tibetan interpretations is central to
the authorial identity Tsong kha pa seeks to project. His frequent invectives against ‘earlier Tibetan scholars’ (sngon gyi bod kyi mkhas pa,
etc.) can only be effective against the background of a claim to distinctive, unmediated access to the Indic tradition. As is clear, this was
accomplished, in part, through appeal to alternative translations. The
fact that his claim to authoritativeness on these grounds would seem to
be (by and large) specious, only serves to make the situation more
intriguing. He would seem to have counted on the fact that there were
none to contradict him—that is, no contemporary with the Sanskritic
credentials to undermine his work, or the ability to counterpose the
authority of his alternative readings. Given the documented existence
of such persons, the source(s) of his confidence bears further inquiry.51
Finally, this corpus of texts, in providing a rare glimpse of some of
the many voices speaking for the Indian literary heritage, constitutes a
tremendous resource for contemporary textual criticism, including socalled ‘Bka’ ’gyur (and Bstan ’gyur) Studies’. Passages cited in the
51 As noted above, several near-contemporaries of Tsong kha pa did seem to have
such expertise (pending closer analyses such as this one), so it seems doubtful that he
might have believed he could successfully ‘bluff’ in this regard. If we do, in fact, conclude that he did not actually cite Indic texts he himself personally consulted, his citation of alternative translations also becomes somewhat dubious. In conversation at the
IATS Seminar in Oxford, Prof. Ronald Davidson suggested that Tsong kha pa had likely derived (not to say ‘lifted’) his citations wholesale from other sources. It does seem
likely that the majority of the texts he takes to be superior were indicated as such by
other sources he deemed trustworthy (such as the works of Bu ston)—though there are
writings attributed to him which strongly suggest their author was consulting such
alternative texts directly (See e.g. Bshad rgyud rdo rje phreng ba’i zin bris rje nyid kyis
gnang ba in Rje Tso# kha pa 1999).
One of the major questions unfortunately left unaddressed in the present essay is the
identity of these presumed sources. Some of the citations of the Pradīpoddyotana, for
instance, seem to come from the writings of Bu ston. However, I have not been able to
identify the source of his citations of the Caryāmelāpakapradīpa treated herein. I have
unfortunately not been able to check the writings of Red mda’ ba Gzhon nu blo gros as
suggested to me by Prof. Kapstein, though I agree that this is a likely source.
TANTALISING TRACES
27
works of scholastics such as Tsong kha pa are sometimes the only
extant versions of these independent voices, and are thus an important
source for the literary history of Tibet. They can also be of immense
value in the task of editing Sanskrit literature.52
While we must, it seems, surrender our corporate faith in the rectitude and consistency of the Tibetan translations, what we gain is far
more valuable than what we imagined we had to lose. As we have seen
in the above examples, the extant fragments of alternative translations
give us invaluable information on and insight into the various states of
texts in early second-millennium Tibet. One encounters variant readings which may be taken into account in editing and interpreting Indic
works. One observes divergent manuscript readings reflected in various
Tibetan translations—suggesting once again that we take seriously the
proposition that variation in Tibetan may be legitimately based on variation in the source texts. One also finds further confirmation of the
findings of Anne MacDonald regarding the (apparently common) practice of Tibetan translators who—quite likely seeking to attain the kind
of mythic consistency of canonical translations we have problematised—regularly cannibalised previously completed Tibetan translations, ‘cutting and pasting’ cited passages into their own new translations.
In this material, one encounters translations which are far preferable,
far more faithful to the original, than those preserved in the canonical
collections. It should be clear from the above that to rest satisfied with
the Lo chen corpus (or other canonical translations) as ‘the’ Tibetan
version is to limit ourselves unnecessarily and, in certain cases, to rely
on definitively unreliable texts. Throughout the examples given above,
we can see that the ‘lost’ translations of Chag are often markedly superior to those of Lo chen. This would all commend, as a methodological
principle, that scholars of Tibetan Buddhist literature—in particular,
those Indologists or ‘Indo-Tibetanists’ who are interested in Tibetan
translations largely for the insight they give into lost or corrupted
52 Of course, it should not be forgotten that Sanskrit texts can be of great use also
in the editing of their Tibetan translations. Not only are their many instances where
genitive and instrumental particles have become confused in the course of transmission, but there are other, more gross differences, which can be rectified. For instance,
it is not uncommon for homophones to be confused in the transmission of Tibetan
translations. Thus, for instance, in Lo chen’s translation of the CMP, in Chapter Three,
the received text reads bstan for brtan; elsewhere in the same chapter, the text reads
bzlas for ’das; and later in Chapter Seven, one encounters gzod nas for gdod nas. Cf.
Wedemeyer (forthcoming [2006]).
CHRISTIAN K. WEDEMEYER
28
Sanskrit originals—pay closer attention to the idiosyncrasies of the
translator(s) of each specific text with which they are working.53
In recent years there has been tremendous growth in so-called ‘Bka’
’gyur Studies’, which has shed invaluable light on the creation and
development of these canons. This research, in establishing stemma of
the major exemplars, is an important contribution to the task of editing
the translations included therein. However, these translations are only
one part of a much larger story of the Tibetan translations of Buddhist
texts. If the goal is reliable editions, consulting extra-canonical translations is also a desideratum—not only for the work of editing the Indic
originals, but also in editing the ‘standard translations’ themselves.54
All of this would also suggest a research desideratum of cataloguing
these alternative texts, to make them available to modern editors. It may
also be worthwhile—in addition to mining the scholastic commentarial literature—to search for any surviving texts of these ‘lost’ translations. Some of these are to be found in non-canonical collections, gsung
’bum, and the like. In addition, a Nepalese colleague of mine once suggested to me that the Tibetan collection of the National Library of
Nepal may contain some of these (presumably manuscript) texts,
though others I have spoken with have expressed skepticism that any
such will turn up. Let us hope these latter are mistaken.
53
Nearly twenty years ago, Leonard van der Kuijp made an important foray in this
direction in an article entitled “A Text-Historical Note on Hevajratantra I:v:1–2” (van
der Kuijp 1985). In this piece, he refers to the Hevajratantra edition of Snellgrove. The
latter is a perfect example of text-critical scholarship taking the canonical version as
“the [sic] Tibetan text”. Van der Kuijp, on the contrary, draws on an ‘alternative’ translation by Ngor chen Kun dga’ bzang po to illustrate the textual history of the
Hevajratantra in the early-mid second millennium, and notes references to other (presumably no longer extant) translations by G.yi jo Lo tsā ba and Shong Lo tsā ba.
54 For example, as indicated above (note 40), example eight shows a case where Lo
chen’s text reads bshad par bya for what should be (based on the Sanskrit and Chag)
spyad par bya. I personally believe this to be a case of ‘old scribal error’ (in Tsong kha
pa’s terms, bri nor ring du brgyud pa), which predated Tsong kha pa. That is, sometime in the process of manuscript tranmission, between the eleventh and the thirteenth
centuries, the syllable spyad was altered to bshad. Whether copied from an oral recitation or directly copied from a text, it is easy enough for (homophones or) such pseudohomophones to creep into the textual transmission. One may compare in this regard the
comments made by David Snellgrove regarding the tribulations of having his edited
text of four rnam thar copied by scribes in rural Dolpo: “[With] even the best of
scribes...as there are so many ways of writing the same sound in Tibetan, there is constant risk of error if he tries to write a whole passage without glancing back at the text
he is copying”; cf. Snellgrove [1964] 1992: 69–70.
TANTALISING TRACES
29
V. CONCLUSION
In sum, I believe it is profitable to pay closer attention to this corpus of
texts, lying concealed within the writings of scholastic authors and
elsewhere. They are a uniquely rich source regarding the early literary
history of Tibet; they can shed light on individual authors’ approach to
the rhetoric of legitimation in early second-millennium religious contestation; and they are underutilised witnesses which promise to be of
great aid both in the work of editing Tibetan religious literature and
broader research into canon and criticism in traditional Tibetan culture.
As we have seen, the landscape of Tibetan translation is by no means
the uniform and mechanical place some authors make it out to be. The
agency and creativity of the Tibetan translators, their triumphs and their
bungles, need to be taken very seriously by modern scholars. They
should no longer be treated as invisible (if occasionally venerated),
transparent media through which the Sanskrit Buddhist literary culture
was transmitted to Tibet. Their works should be treated as works in the
true sense, and attributed to those individuals to whose labours they are
owed. Scholars should no longer refer to the products of specific lotsāwa-s as ‘the Tibetan text’—but should refer them to their proper
authors, the translator(s). To do otherwise is not only rude; it is scientifically and historically imprecise. Tibetan translations are not ahistorical entities floating ‘out there’ to be exploited by Tibetans and scholars alike, anymore than is the work of modern scholars. Like the latter,
they are historical products, with their attendant idiosyncrasies, ideological encodings, and multiple cultural entanglements—all of which
are susceptible to analysis, if scholars will attend to them.
From the materials examined herein, a new and more detailed picture emerges of the great translators and their works—one, not of
automaton lotsāwa-s mechanically rendering texts like a kind of premodern Babelfish,55 but of creative, intelligent, and fallible human
beings engaged in the challenging task of interpreting Sanskrit
Buddhist literature through the medium of classical Tibetan. For all that
I have, in the foregoing, identified numerous shortcomings of their
work, this is by no means to disparage it. Rather, it is merely a call for
a more nuanced understanding of these translations as cultural and historical products, and for further reflection on their use by contemporary
55 That is, the electronic translation programme available on the Internet from
AltaVista.
CHRISTIAN K. WEDEMEYER
30
scholars of Indian and Tibetan Buddhism. Indeed, in the end, we might
very well be justified in admiring the ‘marvel of the Tibetan translations of Buddhist texts between the eighth and the fourteenth centuries’
(with which sentiment this paper began)—not on the basis of some
putative, mythical univocality and transparency, but as a tremendous,
polyvocal monument to the spirit of willed human work on the part of
the Tibetan intelligentsia.
APPENDIX: COMPARATIVE TABLE OF TEXTS
‘STANDARD’ TRANSLATION ‘ALTERNATIVE’ TRANSLATION
SANSKRIT TEXT(S)
1. Citations of the Caryāmelāpakapradīpa: from Rje Tso# kha pa 1978c: 587
[Lo chen]
sku dang gsung dang thugs
mngon par rtogs nas
[kyang] | sa bcu thob pa |
sgyu ma lta bu’i ting nge
’dzin so sor thob cing |
[Chag]
lus ngag sems rnam par
dben pa mngon par rtogs
nas kyang sa bcu thob ste |
sgyu ma lta bu’i ting nge
’dzin so sor thob bo |
sangs rgyas kyi bya ba
mngon sum du byas nas
sangs rgyas kyi sku mngon buddha-kāya* sāk'āt k$tvā
sum du byas nas |
kāya-vāk-citta-vivekādhigato ’pi daśa-bhūmi* prāpya
māyopama-samādhi* pratilabhate |
Tsong kha pa’s comment: chag ’gyur las...zhes dang | ...zhes ’byung bas ni ’gyur
snga ma gsal du btang ngo |
2. Citation of Pañcakrama: from Rje Tso# kha pa 1978c: 424
[Lo chen]
[Chag]
mtshan nyid bzhi bcu gang rang bzhin bzhi bcu’i skad catvāri*śat prak$taya.
yin pa |
cig rnams |
k'a(ikāś cātiśūnya-jā. |
shin tu stong pa’i skad cig shin tu stong pa las skyes te | (PK, II.20cd)
ste ||
Tsong kha pa’s comment: rkang pa tha ma gnyis la chag ’gyur las...
3. Citation of the Caryāmelāpakapradīpa: from Rje Tso# kha pa 1978c: 673.
[Lo chen]
[2 New Translations]
til ’bru tsam la’ang dngos dngos po til ’bru tsam la
tilamātre'v api vastu'u
por ’dzin pa’i blo spang bar yang yongs su ’dzin pa’i
parigraha-buddhi* parityabya’o |
blo dor bar bya’o |
jet |
Tsong kha pa’s comments: gnyis pa ni ’gyur gsar gnyis las | dngos po til ’bru tsam la
yang yongs su ’dzin pa’i blo dor bar bya’o | zhes bsgyur ba legs pas yo byad til tsam
yang bsags te mi ’dzin pa la bya’i | bden zhen gyi dngos ’dzin mi byed pa la mi bya’o |
4. Citation of the Pañcakrama: from Rje Tso# kha pa 1978c: 604.
[Lo chen]
[Chag]
de nyid shes pa’i rim shes de nyid shes pas rim pa
jñātvā krame(a tattva-jña.
nas | bdag byin brlab pa’i bzhin | rang byin brlab dang svādhi'/hāna-prabhāsvarau |
’od gsal la | de nyid la ni
’od gsal shes | de dag kho tayor eva samāja* yad
’dus pa gang | zung du ’jug na ’dus pa gang | ’di ni zung yuganaddha-kramo hy
pa’i rim ’di yin ||
’jug rim pa’o ||
ayam || (PK V.11)
TANTALISING TRACES
kun rdzob dang ni don dam
dag | so so’i char ni shes
gyur pa | gang du yang dag
’dres gyur pa | zung du ’jug
par de bshad do ||
kun rdzob dang ni don dam
gyi | rnam dbye so sor shes
nas ni | gang du yang dag
’dres gyur pa | zung du
’jug par de bshad do ||
31
sa*v$ti* paramārtha* ca
p$thag jñātvā vibhāgata. ||
sa*mīlana* bhaved yatra
yuganaddha* tad ucyate ||
(V.13)
Tsong kha pa’s comment: chag ’gyur las...|
5. Citation of Caryāmelāpakapradīpa: from Rje Tso# kha pa 1978c: 565.
[Lo chen]
sgyu ma lta bu’i ting nge
’dzin las bzhengs nas | ’gro
ba rnams la ston par mdzad
pa yin no |
[Chag]
sgyu ma lta bu’i ting nge
’dzin gyis bzhengs nas |
gdul bya’i skye bo rnams la
chos kyi ’khor lo rab tu
bskor ba yin no |
[2 variant manuscripts]
māyopama-samādhi*
vyutthāya janebhyo dharma* pravartitavān | (MS
B), OR māyopama-samādhi* vyutthāya vineyajanebhyo dharmacakra*
pravarttitavān | (MS C)
Tsong-kha-pa’s comment: chag ’gyur las |...| zhes ’byung ngo |
6. Citation of Caryāmelāpakapradīpa: from Rje Tso# kha pa 1978c: 608.
[Lo chen]
gzugs kyi bdag nyid can gyi
ming ni rdo rje ’chang chen
po zhes bya’o |
[Chag]
gang ming dang gzugs kyi yannāmarūpātmako vajrabdag nyid can rdo rje
dhara iti |
’chang chen po |
Tsong kha pa’s comment: gzugs kyi bdag nyid can gyi ming ni | zhes pa la chag ’gyur
las | gang ming dang gzugs kyi bdag nyid can rdo rje ’chang chen po | zhes ’byung ba
ltar ’grel pa zla ’od las kyang | ming phung po lhag ma bzhi dang | gzugs gzugs phung
la bshad do | de yang zung ’jug rdo rje ’chang de rlung sems tsam gyi bdag nyid du
bstan pas | sngar grol ba po su zhig yin zhes pa’i lan no |
7. Citation of Uttaratantra (in CMP, PK, and PU): from Rje Tso# kha pa 1978c: 313.
[Lo chen’s PK and CMP
Some Indian MSS, and
Two old translations of PU
ch. 12]
[Chag’s CMP and PK
All translations PU Ch. 6,
Some Indian MSS, and
The two new translations
of PU Ch. 12]
ye shes lnga yi rang bzhin
dbugs | ’byung ba lnga yi
ngo bo nyid | gong bu’i
gzugs kyis phyung nas ni |
sna yi rtse mor rab tu brtag ||
ye shes lnga yi rang bzhin
dbugs | ’byung ba lnga yi
ngo bo nyid | gong bu’i
gzugs kyis phyung nas ni |
pad ma’i sna tser rab tu
brtag ||
pañca-jñāna-maya*
śvāsa* pañca-bhūta-svabhāvaka* |
niścārya padma-nāsāgre
pi()a-rūpe(a kalpayet ||
(all other)
OR
niścārya pi()a-rūpe(a
nāsikāgre tu kalpayet ||
(UT.147)
Tsong kha pa’s comment: le’u drug pa’i sgron gsal du rgyud phyi ma drang pa’i ’gyur
thams cad dang | bcu gnyis pa’i ’grel par drangs pa’i ’gyur gsar gnyis dang | chag gis
bsgyur ba’i rim lnga dang spyod bsdus las padma’i sna rtser zhes ’byung la | le’u bcu
32
CHRISTIAN K. WEDEMEYER
gnyis pa’i ’grel pa’i ’gyur rnying gnyis dag pa dang | rgyud phyi ma dang rim lnga
dang spyod bsdus lo chen gyis bsgyur ba la padma zhes pa med pas | rgya gar gyi dpe
la yod med gnyis byung bar gsal yang | spyod bsdus su dkyil ’khor bzhi’i rlung la yi
ge gsum gyi rdor bzlas steng gi sna rtser byed ba | ye shes lnga’i zhes sogs rkang pa
drug gis gsal bar bstan par gsungs pas | padma med par dag par mngon no |
8. Citation of Caryāmelāpakapradīpa: from Rje Tso# kha pa 1978c: 678.
[Lo chen]
shin tu spros pa med pa’i
spyod pa smyon pa’i brtul
zhugs bshad pa’i mdo las
gsungs pa’i rim pas kyang
bshad par bya’o |
[2 New (Chag and Pa
tshab)]
bshad pa’i rgyud las
gsungs pa’i tshul dang rim
pas smyon pa’i brtul zhugs
kyis kyang shin tu spros
med la spyad par bya |
unmatta-vratenāpy atyantani'prapañca-caryāyā*
vyākhyātantrokta-krame(a
caritavyam |
Tsong kha pa’s comment: ’dir spyod bsdus las | shin tu spros med kyi spyod pa smyon
pa’i brtul zhugs kyi mdo las gsungs pa’i rim pas kyang bshad par bya’o | zhes mang du
drangs pa ni | rdo rje snying po rgyan kyi rgyud le’u bcu drug pa nas ’byung ngo | ’di la
spyod bsdus ’gyur gsar gnyis las bshad pa’i rgyud las gsungs pa’i tshul dang rim pas
smyon pa’i brtul zhugs kyis kyang shin tu spros med la spyad par bya zhes ’byung yang
| lo chen gyis smyon pa’i brtul zhugs bshad pa’i mdo zhes bsgyur ba legs te | le’u der
brtul zhugs kyi spyod pa bstan la | de yang smyon pa’i brtul zhugs su gsungs pas |
bshad ces pa ni smyon pa’i brtul zhugs bshad ces pa’i don yin gyi ’dus pa’i bshad
rgyud du ston pa min pa’i phyir |
BIBLIOGRAPHY
Apte, Vaman S. [1890] 1998. A Practical Sanskrit-English Dictionary. Delhi: Motilal
Banarsidass.
Chandra, Lokesh. 1992–94. Tibetan-Sanskrit Dictionary. Seven (‘supplementary’) volumes. New Delhi: International Academy of Indian Culture and Aditya Prakashan.
Clark, Walter E. 1951. The Future of Indian Studies. Journal of the American Oriental
Society 71(4), 209–12.
Davidson, Ronald M. 1981. The Litany of Names of Mañjuśrī: text and translation of the
Mañjuśrīnāmasa*gīti. Mélanges Chinois et Bouddhiques 20, 1–69.
Dreyfus, Georges B. J. 2003. The Sound of Two Hands Clapping: the Education of a
Tibetan Buddhist Monk. Berkeley, Los Angeles, and London: University of
California Press.
Eimer, Helmut. 1993. Location List for the Texts in the Microfiche Edition of the Phug
brag Kanjur: Compiled from the Microfiche Edition and Jampa Samten’s
Descriptive Catalogue. Tokyo: International Institute for Buddhist Studies.
von Hinüber, Oskar. 2000. Der vernachlässigte Wortlaut. Die Problematik der
Herausgabe buddhistischer Sanskrit-Texte. In K. Gärtner and H.H. Krummacher,
(eds) Zur Überlieferung, Kritik und Edition alter und neuerer Texte. Stuttgart:
Franz Steiner Verlag.
Imaeda, Yoshiro. 1981. Note sur le Kanjur de Derge. Mélanges Chinois et Bouddhiques
20, 227–36.
TANTALISING TRACES
33
Ishikawa, Mie. 1990. A Critical Edition of the Sgra sbyor bam po gnyis pa. Tokyo: The
Toyo Bunko.
Jackson, David. 1983. Notes on two early printed editions of Sa skya pa works. Tibet
Journal 8(2), 3–24.
Kapstein, Matthew T. 2003. The Indian literary identity in Tibet. In Sheldon Pollock
(ed.) Literary Cultures in History: Reconstructions from South Asia. Berkeley:
University of California Press, 747–802.
van der Kuijp, Leonard. 1985. A text-historical note on Hevajratantra I:v:1–2. Journal
of the International Association of Buddhist Studies VII(1), 83–89.
Lessing, Ferdinand, and Alex Wayman 1968. Mkhas grub rje’s Fundamentals of the
Buddhist Tantras. The Hague: Mouton.
MacDonald, Anne. 2002. From Prasannapadā to Tshig gsal: remarks on an 11th/12th
century translation project. Unpublished paper presented at the XIIIth Conference
of the International Association of Buddhist Studies, Bangkok, Thailand.
–––– 2003. Interpreting Prasannapadā 19.3–7 in context: a response to Claus Oetke.
Wiener Zeitschrift für die Kunde Südasiens XLVII, 143–95.
Mimaki, Katsumi, and Toru Tomabechi (eds) 1994. Pañcakrama: Sanskrit and Tibetan
Texts Critically Edited with Verse Index and Facsimile Edition of the Sanskrit
Manuscripts. Tokyo: The Centre for East Asian Cultural Studies for Unesco.
Monier-Williams, Monier. [1899] 1990. A Sanskrit-English Dictionary. Delhi: Motilal
Banarsidass.
Negi, J.S. 1993–2003. Bod skad dang legs sbyar gyi tshig mdzod chen mo [TibetanSanskrit Dictionary]. Sarnath: Central Institute for Higher Tibetan Studies.
Nishioka, Soshū. 1983. Index to the catalogue section of Bu ston’s “History of
Buddhism”. Annual Report of the Institute for the Study of Cultural Exchange, no.
6, 47–201.
Patel, Prabhubhai B. 1949. Cittaviśuddhiprakara"a of Āryadeva. n.p.: Visvabharati.
Prāsādika, Bhikshu, and Lal Mani Joshi (eds and transl.) 1981. Vimalakīrti-nirdeśasūtra: Tibetan Version, Sanskrit Restoration, and Hindi Translation. Sarnath:
Central Institute for Higher Tibetan Studies.
Rje Tso# kha pa Blo bza# grags pa. 1977. The Collected Works (gsu- ’bum) of Rje Tsokha pa Blo bza- grags pa. New Delhi: Ngawang Gelek Demo, vol. 8.
–––– 1978a. The Collected Works (gsu- ’bum) of Rje Tso- kha pa Blo bza- grags pa.
New Delhi: Ngawang Gelek Demo, vol. 6.
–––– 1978b. The Collected Works (gsu- ’bum) of Rje Tso- kha pa Blo bza- grags pa.
New Delhi: Ngawang Gelek Demo, vol. 7.
–––– 1978c. The Collected Works (gsu- ’bum) of Rje Tso- kha pa Blo bza- grags pa.
New Delhi: Ngawang Gelek Demo, vol. 11.
—–– 1999. Rje tsong kha pa chen po’i gsung ’bum zhugs so (cha dang ja). n.p.: Mtsho
sngon mi rigs dpe skrun khang.
Samten, Jampa. 1992a. Preliminary notes on the Phug brag bKa’ ’gyur: a unique edition of the Tibetan Buddhist Canon. In Shooren Ihara and Zuihoo Yamaguchi
(eds) Tibetan Studies: Proceedings of the 5th Seminar of the International
Association for Tibetan Studies, Narita 1989. Narita: Naritasan Shinshoji, vol. 1,
115–20.
—–– 1992b. Phug brag Bka’ ’gyur Bris pa’i Dkar chag = A Catalogue of the Phug
brag manuscript Kanjur. Dharamsala: Library of Tibetan Works and Archives.
Sā#k$tyāyana, Tripi/akācharya Rāhula. 1935. Sanskrit palm-leaf mss. in Tibet. Journal
of the Bihar and Orissa Research Society XXI, part 1, 21–163.
Sarma, K.V. 1982. Manuscriptology and textual criticism in medieval India. Indologica
Taurinensia X, 281–88.
34
CHRISTIAN K. WEDEMEYER
Schaeffer, Kurtis R. 1998. Editorial perfection and spiritual failure: creating the Kanjur
and Tanjur at Derge. Unpublished paper presented at the VIIth Seminar of the
International Association for Tibetan Studies, Indiana.
—–– 1999. Printing the words of the master: Tibetan editorial practice in the collected
works of ’Jam dbyangs bzhad pa’i rdo rje I (1648–1721). Acta Orientalia 60, 159–77.
—–– 2004. A letter to the editors of the Buddhist Canon in fourteenth century Tibet:
the Yig mkhan rnams la gdams pa of Bu ston Rin chen grub. Journal of the
American Oriental Society 124(2), 265–281.
Shāstri, Mahamahopadhyaya Haraprasad. 1898. The discovery of a work by Āryadeva
in Sanskrit. Journal of the Asiatic Society of Bengal 67, part 1, 175–84.
Smith, Eugene G. [1959] 2001. Buddhist literary and practical arts according to Bo
dong Pa( chen Phyogs las rnam rgyal. In E.G. Smith. Among Tibetan Texts:
History and Literature of the Himalayan Plateau. Boston: Wisdom Publications,
179–208.
Snellgrove, David L. 1959. The Hevajra Tantra: a Critical Study, Part 2: Sanskrit and
Tibetan Texts. London: Oxford University Press.
–––– ed. and trans. [1964] 1992. Four Lamas of Dolpo: Tibetan Biographies.
Kathmandu: Himalayan Booksellers, vol. I.
Stearns, Cyrus. 1999. The Buddha from Dolpo. Albany: SUNY Press.
Verhagen, Pieter. 1996. Tibetan expertise in Sanskrit grammar: ideology, status and
other extra-linguistic factors. In J.E.M. Houben (ed.) Ideology and Status of
Sanskrit: Contributions to the History of the Sanskrit Language. Leiden: Brill,
275–87.
Wedemeyer, Christian Konrad. 1999. Vajrayāna and its Doubles: a Critical
Historiography, Exposition, and Translation of the Tantric Works of Āryadeva.
Doctoral dissertation, Columbia University.
—— forthcoming 2006. Āryadeva’s Lamp that Integrates the Practices
(Caryāmelāpakapradīpa): the Gradual Path of Vajrayāna Buddhism according to
the Esoteric Community Noble Tradition. New York: American Institute of
Buddhist Studies/Columbia University Press.
Christian K Wedemeyer