JIABS

Journal of the International

Association of Buddhist Studies

Volume 32 Number 1–2 2009 (2010)

 JIABS
Journal of the International
Association of Buddhist Studies

Volume 32 Number 1–2 2009 (2010)

Obituaries
HU Hayan VON HINÜBER
In memoriam, 季羡林 Ji Xianlin (6. August 1911 – 11. Juli
2009) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Articles
James A. BENN
The silent saṃgha: Some observations on mute sheep monks. . . 11
Vincent ELTSCHINGER
Ignorance, epistemology and soteriology – Part I . . . . . . . . 39
Frances GARRETT
Eating letters in the Tibetan treasure tradition . . . . . . . . . . 85
Sarah H. JACOBY
“This inferior female body:” Reflections on life as a Tibetan
visionary through the autobiographical eyes of Se ra mkha’
’gro (bde ba’i rdo rje, 1892–1940). . . . . . . . . . . . . . . . . 115
Andrew MCGARRITY
Āryadeva’s gradual stages: Their transmission from India to
Tibet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Jan-Ulrich SOBISCH
Interpreting the tantras: A Tibetan debate on the numbers of
adepts admissible to tantric consecration . . . . . . . . . . . . . 213
2 Contents

Tibetan scholasticism in the 11th and 12th centuries

Contributions to a panel at the XVth Congress of the International
Association of Buddhist Studies, Atlanta, 23–28 June 2008
Guest editors: Pascale Hugon and Kevin Vose

Pascale HUGON and Kevin VOSE

Introduction – Unearthing the foundations of Tibetan Bud-
dhist philosophy . . . . . . . . . . . . . . . . . . . . . . . . . . 237
Kazuo KANO
rṄog Blo ldan śes rab’s position on the Buddha-nature doc-
trine and its influence on the early gSaṅ phu tradition . . . . . . 249
Kevin VOSE
Making and remaking the ultimate in early Tibetan readings
of Śāntideva . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
Pascale HUGON
The origin of the theory of definition and its place in Phya pa
Chos kyi seṅ ge’s philosophical system . . . . . . . . . . . . . . 319
Jonathan STOLTZ
Phywa pa’s argumentative analogy between factive assess-
ment (yid dpyod) and conceptual thought (rtog pa) . . . . . . . . 369
Georges DREYFUS and Drongbu TSERING
Pa tshab and the origin of Prāsaṅgika . . . . . . . . . . . . . . 387
Thomas DOCTOR
In pursuit of transparent means of knowledge – The Madhya-
maka project of rMa bya Byaṅ chub brtson ʼgrus . . . . . . . . 419
Chizuko YOSHIMIZU
Źaṅ Thaṅ sag pa on theses (dam bcaʼ, pratijñā) in Madhya-
maka thought. . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
 Making and remaking the ultimate in
early Tibetan readings of Śāntideva

Kevin Vose

Śāntideva’s Bodhicaryāvatāra has long been celebrated, alongside

Candrakīrti’s Madhyamakāvatāra, for its explication of empti-
ness from the Prāsaṅgika-Madhyamaka viewpoint, set within a
broader presentation of the Mahāyāna path structure. While the
features that these two texts share – and, indeed, the features of
Prāsaṅgika itself – have been oft-debated in the history of Tibetan
Madhyamaka, we have good textual and doctrinal evidence for
associating them. One compelling reason is Śāntideva’s declara-
tion that “the ultimate is not a referent of awareness,”1 a claim
that echoes Candrakīrti’s statement that the ultimate “is ineffable
and just not a referent of consciousness.”2 Further, in explication
of Śāntideva’s assertion, Prajñākaramati’s (c. 950–1030) commen-
tary on the Bodhicaryāvatāra quotes the Madhyamakāvatāra four
times, linking these two texts’ views on this central Madhyamaka
doctrine.3
The kind of ultimate suggested in these passages, an ultimate
that transcends thought and language, would prove to be a stum-
bling block for those early Tibetan Mādhyamikas with strong com-

1
Stanza IX.2c; La Vallée Poussin, Prajñākaramati’s Commentary
(1905), 352: buddher agocaras tattvaṃ.
2
La Vallée Poussin, Madhyamakāvatāra, 109,2–3: don dam paʼi bden
pa bstan par ʼdod pas de ni brjod du med paʼi phyir daṅ śes paʼi yul ma
yin pa ñid kyi phyir dṅos su bstan par mi nus pas.
3
Prajñākaramati cites Madhyamakāvatāra VI.23, 25, 28, and 29 in
his comments to Bodhicaryāvatāra IX.2; La Vallée Poussin, Prajñāka-
ramati’s Commentary (1905), 353, 361, and 366.

Journal of the International Association of Buddhist Studies

Volume 32 • Number 1–2 • 2009 (2010) pp. 285–318
286 Kevin Vose

mitments to the Buddhist epistemological tradition. Logic-minded

Mādhyamikas, particularly those connected with gSaṅ phu Neʼu
thog Monastery, tended to reject Candrakīrti’s philosophy follow-
ing its spread in Central Tibet around the year 1100, in large part
because of its perceived difficulties in explaining how one real-
izes an ultimate that transcends human intellect.4 In contrast, the
Bodhicaryāvatāra, having long since been translated, retranslat-
ed, and commented upon, was universally acclaimed, making an
accounting of its views incumbent upon any author. Rather than
accept or reject Śāntideva’s seemingly transcendent ultimate and
the host of problems attendant on this view, early bKaʼ gdams pa
scholars found a variety of ways to interpret it. How one interpret-
ed Śāntideva’s ultimate, whether aligning with Candrakīrti’s tran-
scendent portrayal or with gSaṅ phu’s logic-based model, in turn
became a dividing line for a series of categories of Madhyamaka,
including the well-known Svātantrika-Prāsaṅgika divide.
The authority of the Bodhicaryāvatāra was forged at gSaṅ phu
by the monastery’s intellectual founder, rṄog Blo ldan śes rab
(1059–1109), who is credited with establishing the final version of
the Tibetan translation of the text. The colophon of the Tibetan
translation states that the text was first translated from a Kashmiri
exemplar by Ka ba dpal brtsegs (prior to 840), then was revised
in accordance with a central Indian exemplar and its (unnamed)
commentary by Rin chen bzaṅ po (958–1055) and Shākya blo gros
(eleventh century). Finally, rṄog – apparently on the basis of no
new manuscripts or commentaries, but under the guidance of his
Kashmiri teacher Sumatikīrti – corrected and finalized the text.5
Despite the credit given to rṄog, early Tibetan commentaries re-
veal a variety of readings of the root text, suggesting that it cir-
culated in many forms during this period and not just in rṄog’s
“finalized” version.6

4
Candrakīrti’s twelfth-century ascension and the debates it touched
off in Tibet are treated in Vose, Resurrecting Candrakīrti.
5
This information is drawn from the colophon to the Tibetan transla-
tion found in the bsTan ʼgyur; sDe dge edition, vol. ya, 40a5–7.
6
Just what constitutes rṄog’s finalized translation of the Bodhicaryā-
 Making and remaking the ultimate 287

rṄog is known to have written both a commentary and a topical

outline on the Bodhicaryāvatāra.7 The text likewise figured promi-
nently at gSaṅ phu in the generations following rṄog, as a series of
scholars in teacher-to-student relationship composed commentar-
ies on it: rṄog’s student rGya dmar pa Byaṅ chub grags; rGya dmar
pa’s student Phya pa Chos kyi seṅ ge; and Phya pa’s student gTsaṅ
nag pa brTson ʼgrus seṅ ge.8 To these we may add the commen-
tary of the second Sa skya pa hierarch, bSod nams rtse mo (1142–

vatāra is difficult to pinpoint. Akira Saito has analyzed Bu ston’s suspi-

cion that the version of the Bodhicaryāvatāra available to him (which he
included in his bsTan ʼgyur collection) contains unwarranted “emenda-
tions” made by gTsaṅ nag pa; see Saito, “Bu ston on the sPyod ʼjug,”
79–85. In one example, which he takes to be representative, Saito (p. 84)
suggests that “the alteration [of rṄog’s translation] appears to have been
made with rather careless consultation of the old translation(s).” A thor-
ough evaluation of this textual conundrum will require an examination
of the Bodhicaryāvatāra stanzas embedded in the various Indian and
Tibetan commentaries, as compared to the stanzas preserved in the bsTan
ʼgyur editions and in the Dunhuang manuscript (Stein 628) edited in Saito,
A Study of the Dun-huang Recension. My initial investigation shows that
gTsaṅ nag pa’s commentary offers readings of the Bodhicaryāvatāra
stanzas that accord with the stanzas preserved in the bsTan ʼgyur but not
with those found in the Dunhuang version (in cases where the bsTan ʼgyur
edition and Dunhuang version disagree). If gTsaṅ nag pa was indeed Bu
ston’s culprit, he does not seem to have been utilizing a translation of the
Bodhicaryāvatāra related to the Dunhuang version.
7
For an overview of rṄog’s compositions, see Kano, rNgog Blo-
ldan-shes-rab’s Summary, 125–128 and Kramer, The Great Translator.
Fragments of rṄog’s commentary are cited in bSod nams rtse mo’s com-
mentary (discussed below); the whereabouts of the complete commentary
remain unknown. I thank Kazuo Kano for alerting me to his discovery in
Lhasa of rṄog’s sPyod ʼjug gi bsdus don, a “topical outline” of Śāntideva’s
text, which he is now preparing for publication.
8
The available texts are rGya dmar pa Byaṅ chub grags’s Byaṅ chub
sems dpaʼi spyod pa la ʼjug paʼi tshig don gsal bar bśad pa; Phya pa’s
sPyod ʼjug bsdus don, a topical outline (Phya pa’s full commentary is not
presently known); and gTsaṅ nag pa brTson ʼgrus seṅ ge’s sPyod ʼjug gi
rnam bśad.
288 Kevin Vose

1182), which according to its colophon relates the comments of his

teacher Phya pa and also occasionally cites rṄog’s commentary.9
These four generations of commentaries allow us to see the evolu-
tion of gSaṅ phu exegesis, revealing a variety of ways to accom-
modate and even champion the transcendent ultimate found in the
Bodhicaryāvatāra, while at the same time holding to gSaṅ phu’s
emphasis on inferential logic. These early commentators’ solutions
to the problems attendant upon Śāntideva’s ultimate took the shape
of two interrelated discussions of the nature of ultimate truth and
of the cognitive processes involved with realizing it. These in turn
gave rise to distinct Madhyamaka categories to classify views on
each.

The ultimate: “like an illusion” or “not abiding at all?”

bKaʼ gdams pa discussions of ultimate truth, and the Madhyamaka

classifications these discussions engendered, center round Śānti-
deva’s famous proclamation, “The ultimate is not a referent of
awareness; awareness is said to be conventional,”10 as well as the

9
bSod nams rtse mo, sPyod pa la ʼjug paʼi ʼgrel pa. In the colophon
(515.2,4–6), we read: “Indeed, the Commentary composed by the lord
(btsun pa) is exceedingly clear. However, … I write this for the ease of
realization of myself and those like me. For the sake of easy realization
even of the wisdom chapter, the spiritual friend bSod nams rtse mo clearly
arranged (ñe bar sbyar) this from the concise (tshig bsdus, “a summary”)
and difficult to understand (go dkaʼ ba) Explanation of Engaging in the
Bodhisattva’s Practices composed by the monk Chos kyi seṅ ge.” In in-
cluding bSod nams rtse mo’s commentary in this discussion of early bKaʼ
gdams pa commentaries, I do not intend to portray him as a bKaʼ gdams
pa, but rather take his attribution of “arranging” his teacher’s comments
on the Bodhicaryāvatāra as rendering his comments germane to this in-
vestigation. As will be seen, his comments would prove to be more faith-
ful to Phya pa’s views than those of another of Phya pa’s students, gTsaṅ
nag pa. Even still, the subtle criticism implicit in calling Phya pa’s sum-
mary “difficult to understand” may suggest a certain distance between
teacher and student that we would be unlikely to see in later master-to-
disciple relationships within the established orders of Tibet.
10
La Vallée Poussin, Prajñākaramati’s Commentary (1905), 352:
 Making and remaking the ultimate 289

stanzas that follow this, which discuss how more advanced yogis’
views “harm,” or invalidate, those of less advanced yogis.11 This
latter consideration allows commentators to flesh out just who
harms whom, a doxographical project that in some treatments in-
cludes higher and lower types of Mādhyamikas.12 Unfortunately,
the continued absence of rṄog Blo ldan śes rab’s Bodhicaryāvatāra
commentary prevents us from exploring his views on these stan-
zas. However, in his extant works, rṄog touches upon Śāntideva’s
proclamation on the ultimate, albeit in a quite different context,
his commentary on the Ratnagotravibhāga. There, rṄog relates
“The ultimate is not a referent of speech and … is not a referent of
conceptuality, since conceptuality is conventional.”13 This clearly

buddher agocaras tattvaṃ buddhiḥ saṃvṛtir ucyate // The Tibetan (vol.

ya, 31a1) reads: don dam blo yi spyod yul min / blo ni kun rdzob yin par
brjod // Saito (“Śāntideva in the History of Mādhyamika,” 261, n. 3) re-
ports that the Dunhuang version for pāda d reads: blo daṅ sgra ni kun
rdzob yin // = buddhiḥ śabdaś ca saṃvṛtiḥ (“awareness and speech are
conventional”), which would seem to associate “awareness” and “speech”
in the way rṄog (just below) associates “conceptuality” and “speech.”
11
Stanza 4ab states this concisely (La Vallée Poussin, Prajñākaramati’s
Commentary [1905], 370): bādhyante dhīviśeṣeṇa yogino ʼpy uttarotta-
raiḥ / “Through refinements of yogis’ awareness, the progressively higher
harm [the lower].”
12
Stanza 4c, which notes how higher yogis “harm” lower yogis, would
seem to be germane to this discussion of Madhyamaka classification. It
reads (La Vallée Poussin, Prajñākaramati’s Commentary [1905], 371):
dṛṣṭāntenobhayeṣṭena; “through examples accepted by both [parties].”
This could well be an allusion to Candrakīrti’s Prasannapadā critique
of Bhāviveka’s logical procedures, a discussion that led Tsoṅ kha pa to
distinguish between Candrakīrti’s “Prāsaṅgika” Madhyamaka and Bhā-
viveka’s “Svātantrika” view (on Candrakīrti’s critique and Tsoṅ kha pa’s
distinction, see Ruegg, Two Prolegomena to Madhyamaka Philosophy).
However, none of the bKaʼ gdams pa or Sa skya pa scholars whose com-
mentaries I examine here make this association.
13
Kano’s edition of the Tibetan reads (Kano, rNgog Blo-ldan-shes-
rab’s Summary, 286): rdo rje gnas ʼdi bdun dmigs pa med paʼi ye śes kyis
rtogs par bya ba tsam yin gyi / brjod du med pa yaṅ don dam paʼi raṅ bźin
yin paʼi phyir ro // don dam pa ni ṅag gi yul ma yin paʼi phyir te / rnam par
290 Kevin Vose

parallels Śāntideva’s statement, substituting “conceptuality” for

“awareness,” and suggests that, in rṄog’s estimation, the ultimate
is only beyond the ken of conceptuality but can be accessed by
non-conceptual, yogic awareness. This would delimit the ultimate’s
transcendence, returning it to the realm of (highly refined) human
intellect, an approach characteristic of later gSaṅ phu authors, such
as Phya pa.
In the generation following rṄog, available sources reveal fur-
ther discomfort with a literal reading of Śāntideva’s proclamation.
Discussing how the two truths are divided, rGya dmar pa first
notes that “the basis of division” (dbye baʼi gźi) into two truths is
“mere object of knowledge, taken as a referent by awareness.”14
This would entail that the ultimate, like the conventional, can be
brought within the scope of human awareness: it is knowable.
Awareness itself is of two types: “All awarenesses that engage ob-
jects of knowledge are counted as only two: mistaken conscious-
nesses that engage erroneously and reasoning consciousnesses that
engage non-erroneously.”15 As rGya dmar pa explains, a “reason-
ing consciousness” (rigs paʼi śes pa) is the type of consciousness
(which is here used synonymously with “awareness”) that knows
ultimate truth, through a logical process of investigating the final
nature of phenomena.16 The ultimate would seem then to be a refer-
ent of awareness, directly contradicting Śāntideva.

rtog pa ni kun rdzob yin pas don dam pa rtog paʼi yul ma yin paʼi phyir
ro // “The seven vajra topics are only to be realized by the wisdom that
lacks an intentional object; they are ineffable due to being the nature of
the ultimate. This is because the ultimate is not a referent of speech and
because the ultimate is not a referent of conceptuality, since conceptual-
ity is conventional.” Kano’s translation is on p. 383.
14
rGya dmar pa, Tshig don gsal bar bśad pa, 125,3 (59a3): blos yul du
bya paʼi śes byaʼ tsam. An interlinear note on śes byaʼ tsam explains the
phrase as śes bya ma yin pa las log pa.
15
rGya dmar pa, Tshig don gsal bar bśad pa, 127,1 (60a1): śes bya la
ʼjug paʼi bloʼ mthaʼ dag ni nor bar ʼjug pa ʼkhrul śes daṅ / ma nor bar ʼjug
pa rigs paʼi śes pa gñis kho nar graṅs ṅes pas na //
16
Ultimate truth is “true in the perspective of a reasoning [conscious-
ness].” rGya dmar pa, Tshig don gsal bar bśad pa, 127,1 (60a1): rigs paʼi
 Making and remaking the ultimate 291

However, when commenting on Śāntideva’s proclamation, rGya

dmar pa seems to contradict himself:
[The ultimate] is passed beyond all referents of awareness, conceptual
and non-conceptual. It also has no proliferations (spros pa) because it
is not a determined referent of speech or conceptuality. … The ulti-
mate is not to be characterized by any quality and has no character-
istic; thus [Śāntideva] said, “not a referent of awareness.” … Since all
awarenesses are mistaken and conventional, all referents of aware-
nesses are conventional truths. Thus it is established that the opposite
of conventional truth, ultimate truth, is not a referent of awareness.17
Whereas rGya dmar pa’s previous discussion divided awareness
into “mistake” and “reason,” here awareness is uniformly mistak-
en, functioning only in the conventional, dualistic world of charac-
teristics and the characterized. How are we to reconcile rGya dmar
pa’s conflicting positions? We might read “not a determined refer-
ent of speech or conceptuality” as delimiting the scope of aware-
ness, the same move that rṄog made, thereby allowing for a non-
conceptual form of awareness that accesses the ultimate. However,
the first sentence of the quote, implicating both conceptual and
non-conceptual awareness, would seem to preclude this interpreta-
tion. To make sense of rGya dmar pa’s earlier statement, that the
two truths divide “objects of knowledge” (śes bya), we would have
to tease out some kind of consciousness (śes pa) that is not included
in “all awarenesses.” The ultimate would be accessible to (some

ṅor bden pa.

17
We can note in this passage that rGya dmar pa uses spyod yul (found
in Śāntideva’s text) and yul (in rGya dmar pa’s comments) interchange-
ably. Both are used to translate the Sanskrit viṣaya, but typically only
the former is used to translate gocara, which is the reading found in the
Sanskrit of Śāntideva’s text. rGya dmar pa, Tshig don gsal bar bśad pa,
127,2–4 (60a2–4): rtog pa daṅ mi rtog paʼi bloʼi yul thams cad las ʼdas
paʼo // spros pa med pa yaṅ yin te / sgra daṅ rnam par rtog paʼi źen paʼi
yul ma yin paʼi phyir ro // … chos gaṅ gis kyaṅ mtshon par byar myed pa
mtshan ñid med pa ñid don dam pa yin pas / bloʼi spyod yul min źes byaʼ
baʼo / … blo thams cad ʼkhrul pa kun rdzob pa yin paʼi phyir bloʼi yul
mthaʼ dag kun rdzob kyi bden pa yin pas / kun rdzob kyi bden pa las bzlog
pa don dam paʼi bden pa bloʼi yul ma yin par grub po /
292 Kevin Vose

kind of) consciousness but would be beyond the scope of concep-

tual and non-conceptual awareness. rGya dmar pa, unfortunately,
leaves the discussion unresolved.
For his part, rGya dmar pa introduces a further distinction when
explaining the relationship between the two truths (in technical ter-
minology, the “meaning of the division into two truths” [bden pa
gñis kyi dbye baʼi don]). He relates, “[The two truths] are inex-
pressible as the same or other due to not being one and not being
different,” a position that he identifies as the view of the “utterly
non-abiding system (rab du mi gnas paʼi lugs).”18 He provides an
alternative view, that of the “illusionists” (sgyu ma lta bu) who hold
that “the two truths are two qualitative divisions in a single en-
tity, just like product and impermanence;” however, he notes that
Śāntideva’s position is the former.19 This discussion does not an-
swer the previous conundrum of how the ultimate is an object of
knowledge but not a referent of awareness. However, we can detect
some amount of harmony between these two stances and the two
Madhyamaka viewpoints rGya dmar pa here mentions: The “il-
lusionist” position holds that the two truths are “qualitative divi-
sions” (chos kyi dbyeʼ ba) of an entity (dṅos po), with “entity” being
(in some presentations) equated with “object of knowledge.” This
suggests a stronger status for ultimate truth, making it a knowable
phenomenon, accessible to analysis. In contrast, the “non-abiding”
position states that the relationship between the two truths is sim-
ply ineffable, transcending speech in the same way that Śāntideva
declared that ultimate truth transcends awareness.20

18
rGya dmar pa, Tshig don gsal bar bśad pa, 126,3–4 (59b3–4): gcig
pa yaṅ ma yin tha dad pa yaṅ ma yin pas de ñid daṅ gźan du brjod du med
pa … dbyeʼ baʼi don ʼdi ni rab du mi gnas paʼi lugs la ltos ste rnam par
bźag paʼo /
19
rGya dmar pa, Tshig don gsal bar bśad pa, 126,4–5 (59b4–5): sgyuʼ
ma lta buʼ ni byas pa daṅ mi rtag pa bźin du bden pa gñis dṅos po cig la
chos kyi dbyeʼ ba gñis su ʼdod mod kyi / slob dpon ʼdiʼi lugs ma yin pas lṅa
{read: sṅa} ma ltar yin no /
20
Unless, à la rṄgog, we pair “ineffable” with “beyond conceptuality,”
rather than “beyond awareness.”
 Making and remaking the ultimate 293

Looking beyond rGya dmar pa’s work, we see that this divi-
sion of Madhyamaka is widely referred to – and widely rejected
– in early bKaʼ gdams pa literature.21 rṄog, his student Gro luṅ pa
Blo gros ʼbyuṅ gnas (c. 1040–1120),22 and Phya pa all refer to this
as a potential bifurcation of Madhyamaka according to positions
on ultimate truth. They all argue against it, but for different rea-
sons. José Cabezón points out that rṄog and Gro luṅ pa see both
of these potential divisions of Madhyamaka as, instead, deviations
from the middle way.23 “Illusionists” assert (in this portrayal) that
the illusory appearance of phenomena is ultimate truth, which in
rṄog’s and Gro luṅ pa’s estimation amounts to an extreme of real-
ism, as this illusory nature would stand through reasoned analysis.
Additionally, Gro luṅ pa is keen to point out that the “illusionist”
view is not that of Śāntarakṣita, although he does not say who held
it.24 The “non-abiding” position carries several possible valences in
rṄog’s and Gro luṅ pa’s work: rṄog seems to equate the position
both with the meditational practice of “no[thing] abiding” in the
mind (yid la mi gnas pa) and with the ontological claim that no
phenomenon abides (chos kun mi gnas); these would be extremes
of quietism (from a practice standpoint) and nihilism (from an on-

21
This bifurcation of Madhyamaka is also rejected by Tsoṅ kha pa and
his dGe lugs pa followers; see Napper, Dependent-Arising and Emptiness,
403–440 (Appendix I: “The Division of Mādhyamikas Into Reason-
Established Illusionists and Proponents of Thorough Non-Abiding”) and
Ruegg, Three Studies, 96–101.
22
José Cabezón discusses the problems attendant upon Gro luṅ pa’s
dates, as well as this provisional estimate, in “The Madhyamaka in Gro
lung pa’s Bstan Rim chen mo.”
23
My discussion here is based on that in Cabezón, “The Madhyamaka
in Gro lung pa’s Bstan Rim chen mo,” which relies on rṄog Blo ldan śes
rab, sPriṅ yig bdud rtsi thig le and Gro luṅ pa Blo gros ʼbyuṅ gnas, bsTan
rim chen mo.
24
Gro luṅ pa criticizes the view that “Śāntarakṣita and others posit il-
lusion as the ultimate;” Gro luṅ pa, bsTan rim chen mo, 805: slob dpon źi
ba ʼtsho la sogs pa ni sgyu ma don dam pa bźed pa ste / While he does not
tell us who the “others” are, they likely include Kamalaśīla.
294 Kevin Vose

tological perspective).25 Gro luṅ pa, instead, states that the “non-
abiding” view holds that the non-abiding of phenomena is itself
established by a “reasoning consciousness,” which, similar to al-
lowing that an illusory nature withstands reasoning, suggests that
something stands out of emptiness – albeit, in this case, that some-
thing would be non-existence.26 This perspective would then repre-
sent an overly reified ultimate. In all interpretations, rṄog and Gro
luṅ pa reject both the “illusionist” and “non-abiding” positions and
so discount their use for dividing Madhyamaka.
In his recently published doxography, Phya pa takes a quite differ-
ent approach to this division, claiming that since all Mādhyamikas
assert both the utter non-abiding and the illusory appearance of all
phenomena (not neither), there can be no division of Madhyamaka
along these lines. The important issue, instead, is just what “with-
stands analysis” (dpyad bzod pa). If one holds that appearances
withstand analysis – that they are not utterly non-abiding – one
would not claim that they are like illusions; this, then, would make
one a realist. If one does not hold that appearances are “mere af-
firming negatives” (ma yin dgag pa tsam), one deprecates the con-
ventional and becomes a nihilist.27 For Phya pa, then, to say that
phenomena are “utterly non-abiding” and to say that they are “like
illusions” come back to the same point: appearances do not with-
stand analysis and so exist only illusorily.
Further, Phya pa states, “All Mādhyamikas, due to not asserting
that illusory [phenomena] withstand analysis, assert that only utter
non-establishment withstands analysis; there is [thus] no difference

25
rṄog’s summary criticism reads (sPriṅ yig bdud rtsi thig le, 708):
sgyu ma gñis med chos kun mi gnas dbu ma yi lugs gñis rnam ʼbyed de yaṅ
rmoṅs pa mtshar bskyed yin; “Distinguishing two Madhyamaka systems
[according to those who hold that] illusion is non-duality and [those who
hold that] all phenomena do not abide amazes [only] fools.”
26
This position would also make emptiness an “affirming negative”
(ma yin dgag), as the existence of “non-abiding” would be affirmed. This
view is unacceptable to Gro luṅ pa and to all the bKaʼ gdams pa authors
treated here.
27
Phya pa, bDe bar gśegs pa, 65,6ff.
 Making and remaking the ultimate 295

[among Mādhyamikas] at all.”28 “Utter non-establishment” (cir yaṅ

ma grub pa) appears to be synonymous with “utterly non-abiding”
in Phya pa’s usage, an equation not reflected in rṄog’s or Gro luṅ
pa’s discussions. Holding that “utter non-establishment withstands
analysis” saves this portrayal of the ultimate from the extremes
of quietism and nihilism that rṄog argued against, as Phya pa’s
ultimate would be reached by a “reasoning consciousness” – it
would “withstand” the analysis of that consciousness – and would
be counted as an existent “object of knowledge.” Where Gro luṅ pa
criticized the “non-abiding” ultimate for affirming the existence of
“non-abiding,” Phya pa draws a fine line: the ultimate withstands
analysis but is not established by reasoning; it is still “utterly not
established.”29
While Phya pa’s portrayal might seem to solve some of the
problems in the “illusionist” and “non-abiding” positions adduced
by rṄog and Gro luṅ pa (by redefining those positions) and, fur-
ther, might seem to ameliorate divisions within Madhyamaka, two
points bear consideration. First, he introduced this discussion by
noting “someone claims there are two systems of assertions on ulti-
mate truth.”30 However, Phya pa’s discussion of the “illusory” status
of phenomena centers on the conventional world. The real problem
would not arise from claiming that conventional phenomena are
illusory but would arise from the claim that an “illusory” nature
is ultimate truth, which is the position rṄog and Gro luṅ pa both
rejected. Elsewhere, Phya pa too argues against this position at
length, which he characterizes as the view that the union of appear-
ances and emptiness (snaṅ stoṅ gñis tshogs) is the ultimate; to this
claim, he points out that each of the components would per force be

28
Phya pa, bDe bar gśegs pa, 67,2: dbu ma ba thams cad kyis sgyu ma
lta bu ñid ni spyad par [dpyad bzod par?] mi ʼdod pas / cir yaṅ ma grub
pa kho na dpyad bzod par ʼdod pa la tha dad gtan med pa yin no //
29
For a more complete discussion of Phya pa’s views on the ultimate,
see Vose, Resurrecting Candrakīrti, 92–99.
30
Phya pa, bDe bar gśegs pa, 65,6–7: don dam paʼi bden pa la yaṅ kha
cig bden pas stoṅ paʼi snaṅ pa sgyu ma lta bur smra ba daṅ / ma yin dgag
du bden pa gaṅ du yaṅ rab tu mi gnas par smra baʼi lugs gñis yod zer ba.
296 Kevin Vose

the ultimate, making ordinary appearances ultimately true.31 Phya

pa, then, follows his gSaṅ phu forebears in rejecting the ultimate
validity of the “illusionist” view.32 The frequency and vehemence
of this refutation in early bKaʼ gdams pa literature suggests that it
was a powerful current outside of bKaʼ gdams pa circles in twelfth-
century Tibet.33
The second point to consider is that, as Phya pa was well aware,
not all those who call themselves Mādhyamikas assert that the ul-
timate bears analysis. Such a position would be labeled “realist”
by other Tibetan Mādhyamikas, and constituted one of the chief
points of contention between Phya pa and twelfth-century support-

31
Phya pa emphasizes the distinction between the absolute negative,
emptiness, that a reasoning consciousness realizes and the affirming
negative, the collection of appearance and emptiness (the illusion-like
nature), that inference comprehends. The inference that proves “entities
are empty of a true nature” must realize an affirming negative, because
it associates the inferential subject, “entity,” with the predicate, “emp-
tiness.” However, a reasoning consciousness realizes only the absolute
negative, “empty of a true nature,” based on this inference. Phya pa, dBu
ma śar gsum, 93,14–97,14; especially 94,14–18.
32
Phya pa adopts a position very close to the “illusionist” view on “the
meaning of the division into two truths.” Above, we saw rGya dmar pa
report the “illusionist” position to hold that “the two truths are two quali-
tative divisions (chos kyi dbyeʼ ba) in a single entity.” Phya pa held that
“the two truths are only different isolates in a single entity;” Phya pa, dBu
ma śar gsum, 10,12: bden pa gñis ṅo bog cig la ldog pa tha dad pa kho
naʼo //
33
The distinction between “illusionist” Madhyamaka and “utterly
non-abiding” Madhyamaka may have arisen in proto-bKaʼ brgyud pa
sources. The two terms are used to denote types of Madhyamaka in
Maitrīpāda/Advayavajra’s Tattvaratnāvalī, where the terms are given as
māyopamādvayavādin and sarvadharmāpratiṣṭhānavādin. On Maitrī-
pāda’s views, see Mathes, “Blending the Sūtras with the Tantras,” 201–
227. Ruegg reports that the distinction is also found in sGam po pa bSod
nams rin chen’s work, Tshogs chos legs mdzes ma, in volume one of his
gSuṅ ʼbum (Delhi: Khasdub Gyatsho Shashin, 1975), ca, folio 85a; see
Ruegg, Three Studies, 32–35.
 Making and remaking the ultimate 297

ers of Candrakīrti.34 His claim, then, that there is no division within

Madhyamaka is not a claim that there is no dispute within Madhya-
maka. Instead, his is the highly polemical claim that those who
disagree with him are not even Mādhyamikas.
Phya pa’s reconfiguration of the “illusionist” and “non-abiding”
positions allows him to accept his teacher rGya dmar pa’s claim
that Śāntideva held the “non-abiding” view, without subscribing to
the transcendent ultimate that such a position may have entailed.
Additionally, like his gSaṅ phu predecessors, Phya pa can reject the
notion that Śāntarakṣita and Kamalaśīla hold an “illusionist” ulti-
mate, while at the same time accepting the conventional validity
of “illusion-like” phenomena.35 These steps are important for pre-
serving the gSaṅ phu interpretation of Madhyamaka, which cham-
pioned the works of all of these authors: Śāntideva, Śāntarakṣita,
and Kamalaśīla. With the rise of Candrakīrti’s importance in Tibet
during Phya pa’s lifetime, preserving Śāntideva’s affiliation with
views that now came to be labeled “Svātantrika” – and thereby
dissociating Śāntideva from the new Prāsaṅgika – required inter-
pretive finesse.36 Furthermore, in Phya pa’s treatment, the “non-
abiding” view comes to represent a well-acceptable Madhyamaka
stance; in Phya pa’s view, all Mādhyamikas espouse it. This inter-
pretation would make it possible for Phya pa’s students to stand be-
hind the “non-abiding” label and, as we will see in the next section,
to subdivide the position further.

34
Twelfth-century Prāsaṅgika views on ultimate truth are treated in
Vose, Resurrecting Candrakīrti, 88–92.
35
As noted above (n. 32), Phya pa’s position on “the meaning of the
division into two truths” is nearly identical with the “illusionist” position
that rGya dmar pa reports.
36
The Indian commentaries on the Bodhicaryāvatāra that treated
the text from a Yogācāra-Madhyamaka standpoint, discussed by Saito
(“Śāntideva in the History of Mādhyamika,” 259), are important prec-
edents for the early gSaṅ phu interpretation. Śāntarakṣita’s citation of the
Bodhicaryāvatāra in the Tattvasiddhi may also be taken as indicating a
compatibility between Śāntarakṣita’s and Śāntideva’s views, although it
seems doubtful that the Tattvasiddhi’s author is the same Śāntarakṣita
who wrote the Madhyamakālaṃkāra.
298 Kevin Vose

One of those students who did not take up the “non-abiding”

label and who would reject his teacher’s Madhyamaka views in fa-
vor of the new Prāsaṅgika was gTsaṅ nag pa brTson ʼgrus seṅ ge.
gTsaṅ nag pa’s discussion of Śāntideva’s ultimate evinces some of
the same tensions we saw in rGya dmar pa’s explication. gTsaṅ nag
pa holds that ultimate truth is an “object of knowledge” but, “an
instance of the ultimate is passed beyond referents of awareness.”37
A significant difference from rGya dmar pa’s presentation is that
gTsaṅ nag pa credits his position to Candrakīrti, noting that ulti-
mate truth is “found in the perspective of a final reasoning con-
sciousness, … just as the Master Candrakīrti said.”38 Unlike his
gSaṅ phu forebears, gTsaṅ nag pa takes Prajñākaramati’s lead in
drawing Candrakīrti’s exposition of the two truths into his own
explanation of Śāntideva’s stanza. While gTsaṅ nag pa’s position
on the ultimate may appear identical to rGya dmar pa’s stance two
generations prior, Candrakīrti’s presence is crucial, as an examina-
tion of gTsaṅ nag pa’s division of Madhyamaka reveals.
gTsaṅ nag pa interprets Śāntideva’s statement that higher yogis
harm lower yogis as implying three successively higher levels of
Madhyamaka.39 The first type holds that “an affirming negation,
a composite of appearance and emptiness” is the ultimate. This,

37
gTsaṅ nag pa, sPyod ʼjug gi rnam bśad, 557,2–3 (36a2–3): śes byaʼi
lus bden gñis su rnam par gźag pa and 557,7–8 (36a7–8): don dam paʼi
mtshan gźiʼ ni bloʼi yul las ʼdas pa.
38
gTsaṅ nag pa, sPyod ʼjug gi rnam bśad, 557,5 (36a5): mthar thug rigs
paʼi śes ṅor rñed pa.
39
gTsaṅ nag pa initially speaks of two types of Mādhyamikas, but then
points out a third Madhyamaka view that invalidates these two (sPyod
ʼjug gi rnam bśad, 561,2–3 [38a2–3]): dbuʼ maʼi rnal ʼbyor ñid la stoṅ pa
mtshan mar lta ba gñis yod de / bden paʼi dṅos pos stoṅ paʼi śes bya ma
yin dgag don dam par smra ba daṅ / stoṅ ñid med dgag don dam par smra
baʼo // de dag la yaṅ dbuʼ ma chen poʼi rigs pas gnod te / “Mādhyamika
yogis have two views on the mark of emptiness: (1) those who hold that an
affirming negative, an object of knowledge that is empty of true entity, is
the ultimate; and (2) those who hold that an absolute negative, emptiness,
is the ultimate. (3) The reasoning of Great Madhyamaka harms those
also.”
 Making and remaking the ultimate 299

of course, is the “illusionist” position already critiqued by rṄog,

Gro luṅ pa, and Phya pa. This position is invalidated (or “harmed”)
by the second type of Mādhyamika, “those who hold that an ab-
solute negative (med dgag), emptiness, is the ultimate.” As noted,
this view appears to have been a very common position among
early bKaʼ gdams pa authors, as it continued to be in the develop-
ment of Tibetan Madhyamaka. This position, then, would seem to
be unassailable; yet gTsaṅ nag pa states, “The reasoning of Great
Madhyamaka (dbu ma chen po) harms those, also.” What is wrong
with holding emptiness to be the ultimate? And, what does “Great
Madhyamaka” signify? gTsaṅ nag pa tells us
Since the entity to be negated is not established, the negation too is
not established, like the death of a barren woman’s child. In that way,
it is realized that ultimately phenomena – entites, non-entities, and so
forth – do not exist at all. … [This] also dispels the idea that the ulti-
mate is established as an analytical referent of awareness.40
These points suggest that “Great Madhyamaka” upholds a strong
interpretation of Śāntideva’s transcendent ultimate: the reasoning
process that negates “true entity” (bden paʼi dṅos po) itself has no
standing and “ultimately phenomena do not exist at all.” In this in-
terpretation, the seemingly unproblematic position that “emptiness
is the ultimate,” while invalidating the “illusionist” position, still
assigns too strong a status to negation.41 Something still stands out
of emptiness. The position is further criticized for a perceived waf-
fling from the ultimate’s transcendence; only “Great Madhyamaka”
understands that the ultimate passes beyond the scope of human
intellect.

40
gTsaṅ nag pa, sPyod ʼjug gi rnam bśad, 561,3–4 (38a3–4): dgag byaʼi
dṅos po ma grub pas de bkag pa yaṅ mi ʼgrub ste / mo śam gyi bu śi ba
bźin źes byaʼo / de ltar na don dam par dṅos po daṅ dṅos med lasogs paʼi
chos ʼgaʼ yaṅ med par rtogs par ʼgyur ro // … don dam bloʼi spyod yul du
grub par ʼgyur sñam paʼi rtog pa ʼaṅ bsal ba yin no //
41
Pa tshab, too, points out that since the object of negation is not es-
tablished, the negation of it lacks status (dBuʼ ma rtsa baʼi śes rab kyi
ti ka, 49: dgag byaʼi raṅ bźin ma grub pas na bkag pa yaṅ mi ʼthad de);
see Georges Dreyfus and Drongbu Tsering, “Pa tshab and the origin of
Prāsaṅgika,” in this volume.
300 Kevin Vose

gTsaṅ nag pa’s invocation of Candrakīrti (to make essentially

the same point that rGya dmar pa made) can be understood in the
context of his Madhyamaka rankings. “Great Madhyamaka” aligns
Śāntideva’s ultimate with Candrakīrti’s and gives both a transcend-
ent reading. In showing how Great Madhyamaka trumps those who
hold emptiness to be the ultimate, gTsaṅ nag pa attacks a position
very much like the one Phya pa argued for and attempts to answer
Phya pa’s critique of Candrakīrti and his newfound followers. Phya
pa holds that an absolute negation, emptiness, is the ultimate and
that emptiness withstands analysis, a position that gTsaṅ nag pa
understands as an overly reified view. Where Phya pa held that the
ultimate is an object of knowledge, gTsaṅ nag pa criticizes those
who “think that the ultimate is established as an analytical ref-
erent of awareness.” We can recall that Phya pa emphasizes that
the ultimate is “utterly non-established.” gTsaṅ nag pa may here
misrepresent his former teacher’s position in order to draw a clear
separation from his own view, which itself offers a significant sof-
tening from both Candrakīrti’s and Śāntideva’s proclamations.
Jayānanda, for one, took Candrakīrti very literally and claimed
that the ultimate was not even an object of knowledge. Phya pa
pointed out a number of problems with this position, the most sig-
nificant being that the ultimate – realization of which is soteriologi-
cally necessary – could not be known.42 Here, we see gTsaṅ nag
pa arguing that the ultimate is an “object of knowledge” but is not
“established as an analytical referent of awareness,” a position he
arrives at in response to Phya pa’s critique. Among gTsaṅ nag pa’s
chief concerns are defending and championing Candrakīrti’s inter-
pretation of Madhyamaka, here reading that interpretation (rightly
or wrongly) into Śāntideva’s text. “Great Madhayamaka,” then, is
gTsaṅ nag pa’s term for what others in this time period began to
call “Prāsaṅgika.”43

42
For the debates between Jayānanda and Phya pa over the status of the
ultimate, see Vose, Resurrecting Candrakīrti, 88–99.
43
In his contribution to the present volume, Thomas Doctor shows that
rMa bya, likewise, calls Candrakīrti’s views “Great Madhyamaka.”
 Making and remaking the ultimate 301

Buddha perception: Prāsaṅgika v. Svātantrika

A second set of divisions of Madhyamaka found in these early

commentaries takes as its chief criterion the status of those trans-
formed by their realization of the ultimate: Buddhas. These divi-
sions are found in exegeses of Śāntideva’s answer to the objection
that the Madhyamaka understanding of emptiness renders the pas-
sage from saṃsāra to nirvāṇa pointless. Paul Williams noted that
where the received Sanskrit of Śāntideva’s text here offers preci-
sion, the Tibetan translation of it offers ample room for interpreta-
tion.44 Translating from the Sanskrit, we can read the objection as
“If what is ceased ultimately is cycling in saṃsāra conventionally,
then the Buddha, too, would be cycling in saṃsāra; thus, what use
would the practices of enlightenment be?”45 In the Tibetan transla-
tion, “what is ceased” (nirvṛtaḥ) is rendered mya ṅan ʼdas, allowing
the possibility that “nirvāṇa” is at issue. Additionally, the verbal
sense of saṃsāra is lost, as are the adverbial usages of “ultimate”
and “conventional.” The Tibetan, then, offers several possibilities,
including the very straightforward rendering, “If the ultimate is
nirvāṇa and saṃsāra is the conventional, the Buddha too would
be saṃsāra; thus, what use would the practices of enlightenment
be?”46
The ambiguity of the Tibetan translation allows for a number of
interpretations, not all of which would seem coherent when reading
the received Sanskrit. Williams pointed out that many Indian and

44
Williams, “On Prakṛtinirvāṇa/Prakṛtinirvṛta in the Bodhicaryā-
vatāra,” 522–523.
45
The Sanskrit of stanzas IX.13cd and 14ab reads (La Vallée Poussin,
Prajñākaramati’s Commentary [1907], 385): nirvṛtaḥ paramārthena
saṃvṛtyā yadi saṃsaret // buddho ʼpi saṃsared evaṃ tataḥ kiṃ bodhi-
caryayā / While redundant, my “cycling in saṃsāra” conveys the Sanskrit
verbal use of the word.
46
The Tibetan, edited from the canonical versions, is found in
Oldmeadow, A Study of the Wisdom Chapter, vol. 2, 34: gal te don dam
mya ṅan ʼdas // ʼkhor ba kun rdzob de lta na // saṅs rgyas kyaṅ ni ʼkhor
ʼgyur bas // byaṅ chub spyod pas ci źig bya // The Dunhuang version does
not differ substantially (Saito, A Study of the Dun-huang Recension, 50).
302 Kevin Vose

Tibetan commentators read this objection as embodying a misun-

derstanding of the difference between “natural nirvāṇa” (or “natu-
ral cessation;” prakṛtinirvāṇa/prakṛtinirvṛta) and the attainment of
nirvāṇa.47 Prajñākaramati reads the passage in just this way, having
the objector state that “a Buddha, who has cessation due to aban-
doning all defilements, also would be cycling in saṃsāra.”48 The
identity between natural nirvāṇa and saṃsāra that Mādhyamikas
accept leads the objector to contend (mistakenly) that the attain-
ment of nirvāṇa would leave one still in saṃsāra. Prajñākaramati’s
reading of the objection makes for an easy Madhyamaka answer:
despite the “natural cessation” of all phenomena, the attain-
ment of nirvāṇa depends on attaining the cessation of ignorance
(avidyānirodha), at the root of the twelve links of dependent aris-
ing, as explained in a lengthy quote from the Śālistambasūtra.49
Upon the cessation of ignorance and the remaining links, “there
would be no cycling in saṃsāra.”50
rṄog Blo ldan śes rab, bSod nams rtse mo tells us, took a simi-
lar approach to explicating Śāntideva’s question and answer. On
rṄog’s reading, the objector claims that since all things are natu-
rally nirvāṇa, which is in the end no different from saṃsāra, those

47
Williams, “On Prakṛtinirvāṇa/Prakṛtinirvṛta in the Bodhicaryā-
vatāra,” 522ff., especially 525–526 where Williams notes, “The distinc-
tion between innate ‘enlightenment’ and that attained through following
the path means that the prakṛtinirvāṇa is almost universally employed
in Tibet to explain the opponent’s objection and its solution.” As will be
seen, the early bKaʼ gdams pa materials present an important counter to
this claim.
48
La Vallée Poussin, Prajñākaramati’s Commentary (1907), 385,17–
18: buddho ʼpi sarvāvaraṇaprahāṇato nirvṛto ʼpi saṃsāret / The Tibetan
translation here renders nirvṛta as mya ṅan las ʼdas pa, consistent with
the handling of Śāntideva’s stanza. However, in Prajñākaramati’s intro-
duction to this passage, his usage of prakṛtinirvṛta (385,7) is rendered
raṅ bźin gyis ldog pa. For the edited Tibetan of the complete passage, see
Oldmeadow, A Study of the Wisdom Chapter, vol. 2, 34–35.
49
La Vallée Poussin, Prajñākaramati’s Commentary (1907), 386,12ff.
50
La Vallée Poussin, Prajñākaramati’s Commentary (1907), 389,12:
saṃsāraṇaṃ na syāt /
 Making and remaking the ultimate 303

who attain nirvāṇa are in actuality still in saṃsāra; so, the objector
wonders, what is the point of the practices of enlightenment? rṄog
explains (according to bSod nams rtse mo) that while all things are
ultimately naturally nirvāṇa, conventionally there is a difference
between the attainment of nirvāṇa and saṃsāra according to “the
extinction or non-extinction of the adventitious causes that gener-
ate afflictions.”51 rṄog’s explanation, then, may have relied on that
of Prajñākaramati, or at minimum was drawn from consultation of
a Sanskrit version of Śāntideva’s text related to that preserved in
Prajñākaramati’s commentary, a version that facilitates a “natural
nirvāṇa versus attained nirvāṇa” reading of the objection and an-
swer.
Surprisingly, Prajñākaramati’s and rṄog’s explication of this
passage was not adopted by other early bKaʼ gdams pa authors.
One reason for this departure is the above-noted ambiguity of the
Tibetan translation of Śāntideva’s question. A second reason is
Śāntideva’s somewhat cryptic answer: “If the causes do not have
their continuum cut, illusion also is not stopped; upon the causes
having their continuum cut, it does not arise even conventionally.”52
Following Prajñākaramati, “the causes” would be ignorance, de-
sire, and hatred – the causes of saṃsāra; the cutting of their con-
tinuum would refer to the reversal of the twelve links of dependent
arising. The parallel construction of this passage implies that what
“does not arise even conventionally” is “illusion.” However, the va-
riety of interpretations left open by the objector’s question allows
Tibetan authors to develop various readings of just what has its
“continuum cut” and just what Śāntideva claims “does not arise,”
readings that give rise to distinct classifications of Madhyamaka.

51
bSod nams rtse mo, sPyod pa la ʼjug paʼi ʼgrel pa, 495.3,4 (295b4):
ñon moṅs pa skyed paʼi rgyu glo bur ba zad ma zad kyis ʼbyed do.
52
La Vallée Poussin, Prajñākaramati’s Commentary (1907), 386,6–
7: pratyayānām anucchede māyāpy ucchidyate na hi / pratyayānāṃ
tu vicchedāt saṃvṛtyāpi na saṃbhavaḥ // Oldmeadow, A Study of the
Wisdom Chapter, vol. 2, 35: rkyen nams rgyun ni ma chad na // sgyu
maʼang ldog par mi ʼgyur gyi // rkyen nams rgyun ni chad pas na // kun
rdzob tu yaṅ mi ʼbyuṅ ṅo // While the original has no pronoun in the final
pāda, I insert “it” as a placeholder for the interpretations discussed below.
304 Kevin Vose

The clearest statement of this alternate interpretation is found

in bSod nams rtse mo’s commentary, our source for the previous
presentation of rṄog’s reading. Having given rṄog’s position, bSod
nams rtse mo notes, “That explanation is not very good. Regarding
this, the great commentator explains in this way.”53 Following this
unnamed “great commentator” (likely to be Phya pa), he claims
that rṄog has misunderstood the objection, for in Sautrāntika, “the
intentional objects of nirvāṇa are ultimate and thoroughly afflicted
intentional objects are conventional.” Given this, the Sautrāntika
opponent here asks
Would the Buddha also have awareness of appearances or would he
not? If not, it follows that the Buddha is a non-entity or is matter. If
Buddha has awareness of appearances, is that conventional or ulti-
mate? If ultimate, [your] thesis stating that awareness is conventional
deteriorates. If conventional, it follows that [the Buddha] is saṃsāra
due to having conventional, thoroughly afflicted proliferations. If you
accept that, what is the point of the bodhisattva practices?54
Rather than charging Madhyamaka with a suffering Buddha –
through the conflation of natural nirvāṇa and the attainment of
nirvāṇa – in this interpretation the objector is concerned about
what a Buddha can perceive. The equations that the Tibetan trans-
lation allows between nirvāṇa and the ultimate, on one hand, and

53
bSod nams rtse mo, sPyod pa la ʼjug paʼi ʼgrel pa, 495.3,5 (295b5):
bśad pa de ha caṅ legs pa ma yin no // ʼdi la ʼgrel chen byed pas ʼdi ltar
ʼchad de /
54
bSod nams rtse mo, sPyod pa la ʼjug paʼi ʼgrel pa, 495.3,5–495.4,2
(295b5–296a2): spyir mdo sde pa raṅ gi grub mthaʼ mya ṅan las ʼdas paʼi
dmigs pa ni don dam pa yin la / kun nas ñon moṅs paʼi dmigs pa ni kun
rdzob ces bsams nas rgol ba ni / gal te don dam mya ṅan ʼdas // ʼkhor ba
kun rdzob de lta na // saṅs rgyas kyaṅ ni ʼkhor ʼgyur bas // byaṅ chub
spyod pas ci źig bya // źes smos te / gal te don dam pa ni mya ṅan las ʼdas
pa yin na ʼkhor ba ni kun rdzob kyi spros pa yin na saṅs rgyas laʼaṅ snaṅ
bcas kyi blo yod dam med / med na saṅs rgyas dṅos med dam bems por
thal la / yod na de kun rdzob yin nam don dam yin / don dam yin na blo
ni kun rdzob yin par brjod kyi dam bcaʼ ñams la / kun rdzob yin na kun
rdzob kun nas ñon moṅs spros pa yod pas ʼkhor bar thal lo / de ʼdod na
byaṅ chub spyod pas ci źig bya źes rgol ba na /
 Making and remaking the ultimate 305

saṃsāra and the conventional, on the other, leaves the Buddha per-
ceiving only an ultimate emptiness, without the ability to perceive
conventional appearances. If a Buddha perceives conventional ap-
pearances, the objector reasons, that Buddha must have “conven-
tional, thoroughly afflicted proliferations,” which the bodhisattva
path was to have removed.
bSod nams rtse mo’s interpretation of the issue is not unique to
him but is first found in rGya dmar pa’s commentary. rGya dmar pa
further reads the objector to equate “mistake” (ʼkhrul pa, bhrānta)
and awareness (blo, buddhi); a Buddha’s perception of appearances
would entail a mistaken Buddha.55 rGya dmar pa and bSod nams
rtse mo thus see a debate on just what connection a Buddha has
with the conventional world. Specifically, if nirvāṇa is the ultimate,
do those who reach nirvāṇa perceive the conventional, which has
(in this interpretation) been declared to be saṃsāra? And if they
do, do these perceptions entail on the part of Buddhas ignorance,
the root of saṃsāra?
rGya dmar pa notes that Mādhyamikas have two possible an-
swers to this objection, which divides them into two camps: “those
who assert that wisdom has its continuum cut” and “those [who
assert] that wisdom does not have its continuum cut.” rGya dmar
pa clearly reads Śāntideva’s answer (“upon the causes having their
continuum cut, it does not arise even conventionally”) in a new
way: here, the issue is whether awareness continues through the
transformation to buddhahood or, alternatively, whether it is a
Buddha’s wisdom that “does not arise.” On the first group’s answer,
rGya dmar pa tells us
Some Mādhyamikas assert that since all awareness is mistaken, when
mistake is extinguished awareness itself does not exist and thus wis-
dom has its continuum cut; “even conventionally” wisdom does not
exist. These assertions are not reasonable. … Even though mistake is
extinguished, wisdom is not stopped. … Since [Buddhas] see (gzigs
pa) illusory dependent arising as just illusion without the capacity to
apprehend it as true, they are not mistaken.56

55
rGya dmar pa, Tshig don gsal bar bśad pa, 133,3–4 (63a3–4).
56
rGya dmar pa, Tshig don gsal bar bśad pa, 134,6–135,1 (63b6–64a1):
306 Kevin Vose

Accepting a strong correlation between nirvāṇa and perceiving the

ultimate, emptiness, the first camp holds that the transformation to
a Buddha eradicates all awareness, which is necessarily mistaken.
Without awareness – the instrument that perceives conventional
appearances – wisdom (which is here understood to be a type of
awareness) cannot arise. With mistaken awareness extinguished,
Buddhas are left in emptiness.
This notion is unacceptable to rGya dmar pa, a partisan of
the second camp, who instead holds that mistake and awareness
are separable. Rather than read Śāntideva’s answer as entailing a
wisdom-less Buddha, rGya dmar pa nuances his response, writing
“When, at buddhahood, ‘the causes’ that are collected in saṃsāra
‘have their continuum cut,’ that [awareness] collected in saṃsāra
‘even conventionally does not arise.’”57 Whereas the first group
understands Śāntideva to claim that “even conventionally, wisdom
does not arise” due to all awareness being extirpated along with
ignorance, this interpretation limits the negation to those states
of mind “collected in saṃsāra.” While “mistake” characterizes
awareness in saṃsāra, it is not a necessary quality of awareness.
Wisdom is a Buddha’s awareness, stripped of mistake by way of
a Buddha’s overcoming the conception that ordinary appearances
are true. rGya dmar pa notes a further significant difficulty in the
first interpretation: “If wisdom did not exist, the Buddha would not
exist; thus, the Buddha’s teachings themselves would not exist and
the sangha who realize them would not exist.”58

ʼon kyaṅ blo thams cad ʼkhrul pa yin pas ʼkhrul pa zad paʼi tshe blo ñid
med pas ye śes rgyun chad do źes kun rdzob du yaṅ ye śes med do źes dbuʼ
ma pa kha cig ʼdod pa ni mi rigs ste / … ʼkhrul pa zad kyaṅ ye śes de mi
ldog go / … rten ʼbrel sgyuʼ ma la bden par ʼdzin pa mi mṅaʼ bar sgyuʼ ma
ñid du gzigs pas ma ʼkhrul źes byaʼo /
57
rGya dmar pa, Tshig don gsal bar bśad pa, 133,7–8 (63a7–8): saṅs
rgyas paʼi tsheʼ ʼkhor bas bsdus paʼi rkyen rnams rgyun ni chad pas na /
ʼkhor bas bsdus pa kun rdzob du yaṅ mi ʼbyuṅ ṅo /
58
rGya dmar pa, Tshig don gsal bar bśad pa, 134,3 (63b3): ye śes myed
na saṅs rgyas ñid med pas / deʼi bstan pa ñid med ciṅ de la bsgrub paʼi
dge ʼdun med pas /
 Making and remaking the ultimate 307

rGya dmar pa’s portrayal of these two Mādhyamika explanations

appears in similar form in bSod nams rtse mo’s explication. Having
already declared that the “natural nirvāṇa v. attained nirvāṇa” in-
terpretation of rṄog misses the point and reoriented the problem
around the issue of what a Buddha perceives, bSod nams rtse mo
gives two possible answers to this problem, which he further iden-
tifies as the Prāsaṅgika and Svātantrika positions. He begins with
the Prāsaṅgika answer, which reads Śāntideva’s “causes” as “the
causes of awareness of appearances:”
‘Upon the causes having their continuum cut:’ The diamond-like
meditative equipoise cuts the continuum of all signs and conceptual-
ity, thereby cutting the continuum of afflictions. Cutting that [further]
cuts the continuum of actions. Cutting the continuum of that [further]
cuts the continuum of awareness of appearances, whereby saṃsāra,
‘even conventionally, does not arise.’59
This equation of saṃsāra and perceiving conventionalities works
well with the Tibetan translation of Śāntideva’s stanza. However,
bSod nams rtse mo cannot accept this equation, as it either leaves
Buddhas in the dark or, as the objector has claimed, leaves all man-
ner of realized beings in saṃsāra.
Instead, bSod nams rtse mo states “We do not assert that all con-
ventionalities are saṃsāra, nor do we assert that all nirvāṇas are
ultimate; we posit [them] as ultimate or conventional through bear-
ing or not bearing analysis.”60 He endorses the Svātantrika read-
ing of Śāntideva’s answer, that “‘Upon causes,’ that is, actions and
afflictions, ‘having their continua cut, even conventionally’ that is,

59
bSod nams rtse mo, sPyod pa la ʼjug paʼi ʼgrel pa, 495.4,4–495.4,5
(296a4–296a5): rkyen rnams rgyun ni chad pa na* ste rdo rje lta buʼi tiṅ
ṅe ʼdzin gyis mtshan ma daṅ rnam par rtog pa thams cad rgyun bcad pas
ñon moṅs pa rgyun bcad / de bcad pas las rgyun bcad / de rgyun bcad
pas snaṅ bcas kyi blo rgyun bcad pas ʼkhor ba kun rdzob tuʼaṅ ṅo źes bya
baʼo / * The text incorrectly reads ma chad pa for chad pa na.
60
bSod nams rtse mo, sPyod pa la ʼjug paʼi ʼgrel pa, 496.1,1 (296b1):
kho bo cag kun rdzob thams cad ʼkhor bar khas mi len / mya ṅan ʼdas pa
thams cad don dam du khas mi len te / dpyad bzod mi bzod kyis don dam
kun rdzob tu ʼjog go /
308 Kevin Vose

even to Buddhas’ conventional wisdom, [illusion] ‘does not arise’

as saṃsāra.”61 Appearances themselves are not cut off by the trans-
formation to buddhahood, afflictions are. As with rGya dmar pa,
appearances have a somewhat neutral status: illusory appearances
can “arise as saṃsāra” to the afflicted or they can be perceived by a
Buddha’s conventional wisdom without these appearances serving
as a cause of suffering. As for rGya dmar pa, afflictions and aware-
ness are separable. There is a point to the practices of enlighten-
ment, after all.
Before considering what we can glean from these divisions of
Madhyamaka, we must ask a more basic question: was this alternate
interpretation of Śāntideva’s objection and answer simply a misun-
derstanding of the original passage? The issue of a Buddha’s ability
to perceive ordinary appearances would be nearly impossible to
derive from the Sanskrit versions that we have of Śāntideva’s text.
Prajñākaramati’s lengthy commentary takes a straightforward ap-
proach to this passage, explaining the distinction between saṃsāra
and nirvāṇa with recourse to the twelve links of dependent arising.
While my investigation of Indian commentaries is far from com-
plete, we can note that Vibhūticandra’s interpretation of this pas-
sage suggests a concern with the relation between “mistake” and
“appearances,” and whether a Buddha’s ability to perceive the latter
entails possession of the former. He writes, “Ultimately, saṃsāra
just does not exist; the appearances of saṃsāra due to mistake
[lasts] for as long as ignorance exists. Having dispelled ignorance
through the ārya path, those [appearances of saṃsāra] also do not
exist.”62 Vibhūticandra’s reading echoes the Prāsaṅgika interpreta-

61
bSod nams rtse mo, sPyod pa la ʼjug paʼi ʼgrel pa, 496.1,2–3 (296b2–
3): rkyen las daṅ ñon moṅs pa rgyun chad na kun rdzob tuʼaṅ ste saṅs
rgyas la ye śes kun rdzob yod kyaṅ ʼkhor bar mi ʼbyuṅ źes bya ba dbu ma
raṅ rgyud pa rnams bśad pa byed do / In reading “even conventionally”
as “even to Buddhas’ conventional wisdom,” bSod nams rtse mo must
read Śāntideva as assuming that Buddhas’ meditative absorption (mñam
bźag ye śes) contains no appearances at all; the point here, as bSod nams
rtse mo sees it, is that appearances, while occurring in Buddhas’ conven-
tional wisdom, are not the afflicted appearances of saṃsāra.
62
Bodhicaryāvatāratātparyapañjikā Viśeṣadyotanī, 260b7–261a1: don
 Making and remaking the ultimate 309

tion that bSod nams rtse mo sketched and rejected. Vibhūticandra’s

commentary, however, was composed later than all of the Tibetan
commentaries discussed herein and so could not serve as bSod
nams rtse mo’s referent.63 Vibhūticandra may be repeating an in-
terpretation from an earlier Indian commentary on Śāntideva’s text
(which would likely require an alternate version of the Sanskrit that
would allow for this interpretation), suggesting that Tibetan com-
mentators may have had some Indian precedent for reading this
passage as concerning a Buddha’s perceptive abilities. At the very
least, we cannot dismiss this Tibetan interpretation on the grounds
of poor philology.
Returning to the issue of Madhyamaka classifications drawn
from this passage, it is clear that rGya dmar pa and bSod nams
rtse mo refer to the same debate, portraying the two Madhyamaka
positions in very similar terms and themselves taking the same side
in the dispute. What are we to make of bSod nams rtse mo’s addi-
tion of labeling the two positions “Prāsaṅgika” and “Svātantrika?”
In the generation between rGya dmar pa and bSod nams rtse mo,
Candrakīrti’s philosophy began to attract a strong following. Among
the issues that Phya pa, who was bSod nams rtse mo’s teacher of the
Bodhicaryāvatāra, found the most distressing about this develop-
ment were the “Prāsaṅgika” (although he did not use the term) por-
trayals of the ultimate and of buddhahood. We see Jayānanda, the
twelfth-century commentator on the Madhyamakāvatāra, interpret
Candrakīrti’s claims that the ultimate “is ineffable and just not a

dam par ʼkhor ba med pa ñid ʼkhrul pas ʼkhor baʼi snaṅ ba ni gti mug yod
pa ji srid duʼo // ʼphags paʼi lam gyis gti mug spaṅs nas de ʼaṅ med par
ʼgyur ro /
63
In his colophon (Bodhicaryāvatāratātparyapañjikā, 285a7), Vibhū-
ticandra states that he translated his text into Tibetan himself at ʼBriṅ
mtshams, making it highly unlikely that it would have been known to
bSod nams rtse mo, as Vibhūticandra made the first of his three trips to
Tibet in 1204, in the company of Śākyaśrībhadra, twenty-two years after
bSod nams rtse mo died. On Vibhūticandra’s travels, see Stearns, “The
Life and Tibetan Legacy,” 127–146. It remains possible that Vibhūticandra
wrote his commentary in Tibet and in this passage offered something of
a response to bSod nams rtse mo.
310 Kevin Vose

referent of consciousness” and that Buddhas have no mind very lit-

erally.64 Since the ultimate is beyond awareness, Jayānanda tells us,
realization of the ultimate must entail the cessation of awareness
and with it the ability to perceive appearances.65 Phya pa argued
at length against these positions, for reasons quite like bSod nams
rtse mo’s: Buddhas must have wisdom, wisdom is a type of aware-
ness, and so awareness must continue in a purified form through
the transformation to buddhahood; to hold otherwise, Phya pa says,
one’s position would resemble the nihilism of the Cārvākas.66 Phya
pa, in turn, was criticized for his “reified ultimate,” an ultimate that
bears analysis and is accessible to awareness (although, as noted

64
La Vallée Poussin, Madhyamakāvatāra, 108,9–11: saṅs rgyas rnams
la ni chos thams cad rnam pa thams cad du mṅon par rdzogs par byaṅ
chub paʼi phyir / sems daṅ sems las byuṅ baʼi rgyu ba gtan log par ʼdod
pa yin no / “We assert that for Buddhas, due to being manifestly and
completely enlightened to all phenomena in all aspects, the movement of
mind and mental factors has entirely ceased.” Candrakīrti makes similar
statements in his discussion of buddhahood, particularly in his autocom-
mentary to stanzas XII.8–9.
65
Jayānanda, Madhyamakāvatāraṭīkā, 146a7–146b1: ci yaṅ thugs su
chud pa med paʼi sgo nas byaṅ chub paʼi phyir sems daṅ sems las byuṅ
baʼi rgyu ba gtan log par ʼdod pa yin te ñams su myoṅ baʼi mtshan ñid
can gyi sems daṅ tshor ba la sogs pa sems las byuṅ ba rnams kyi kun
du spyod pa ste ʼjug pa log par ʼdod pa yin te / sems daṅ sems las byuṅ
ba rnams ʼjug pa ma yin no źes paʼo // des na ci yaṅ snaṅ ba med pa yin
no źes paʼi tha tshig ste / rnam par rtog pa thams cad ʼgag paʼi phyir
ro / “Since enlightenment is by way of not knowing at all, we assert that
the activities of mind and mental factors – feeling and so forth – hav-
ing the character of experiencing, have ceased their engagement; there
is no engagement of mind and mental factors. Therefore, there is no ap-
pearance at all, because all conceptuality has been blocked.” Jayānanda’s
views on ultimate truth and buddhahood are discussed in detail in Vose,
Resurrecting Candrakīrti, 88–92 and 112–122.
66
Phya pa chos kyi seṅ ge, dBu ma śar gsum, 72,16–76,5. This passage
is translated in Vose, Resurrecting Candrakīrti, 159–164 and analyzed at
122–131.
 Making and remaking the ultimate 311

above, Phya pa holds the ultimate to be “utterly not established”), a

position that Jayānanda rejects as “Svātantrika.”67
bSod nams rtse mo remains the earliest known author to pair the
terms “Prāsaṅgika” and “Svātantrika,” here using them to distin-
guish alternative visions of ultimate truth and buddhahood. His us-
age – and the absence of these terms in rGya dmar pa’s discussion –
holds two significant consequences for our understanding of how
Tibetans read Śāntideva. Prior to the widespread dissemination of
Candrakīrti’s texts in the early twelfth century, Śāntideva was not
a Prāsaṅgika. While rGya dmar pa refers to variant interpretations
of Śāntideva (without calling these interpretations “Prāsaṅgika” or
“Svātantrika”), he reads Śāntideva’s views as consistent with those
of Śrīgupta and Jñānagarbha; and he seems to have had Indian
precedent for doing so.68 Further, even as Candrakīrti’s fame grew,
Prāsaṅgika became only one option for reading Śāntideva; as
bSod nams rtse mo’s work evinces, Śāntideva could be read as a
Svātantrika. Śāntideva became a Prāsaṅgika only after generations
of Tibetan debate.
We can recall that gTsaṅ nag pa, writing in the same generation
as bSod nams rtse mo, invoked Candrakīrti’s transcendent ulti-
mate to explain Śāntideva’s presentation; his “Great Madhyamaka”
unites Candrakīrti and Śāntideva around a transcendent reading of
ultimate truth. Intriguingly, gTsaṅ nag pa says nothing about the
Prāsaṅgika-Svātantrika divide over a Buddha’s ability to perceive
appearances. Rather, his explanation of this passage in Śāntideva
returns to a portrayal similar to Prajñākaramati’s: at issue is a mis-
understanding on the part of the objector between natural nirvāṇa
and the attainment of nirvāṇa. He has the objector ask,

67
Jayānanda, Madhyamakāvatāraṭīkā, 120a.
68
rGya dmar pa reads Śāntideva’s portrayal of the two truths as consist-
ent with that of Jñānagarbha (Tshig don gsal bar bśad pa, 127 [60a]) and
refers to Śrīgupta to explain a Buddha’s perception of ordinary appear-
ances (Tshig don gsal bar bśad pa, 135 [64a]). Saito notes the Yogācāra-
Madhyamaka interpretation of the Bodhicaryāvatāra in two early Indian
commentaries (“Śāntideva in the History of Mādhyamika,” 259).
312 Kevin Vose

If it is suitable that even natural nirvāṇa is saṃsāra conventionally,

it is not contradictory that the attainment of buddhahood through
making effort on the path also is saṃsāra conventionally; thus, what
would be the point of attaining a buddhahood that does not cast off the
sufferings of saṃsāra?69
gTsaṅ nag pa quickly answers that Śāntideva’s discussion teaches
that upon the attainment of buddhahood, the causes of saṃsāra
are cut off.70 This brief answer and the failure to discuss the is-
sue of a Buddha’s perception – following three generations (if we
include Phya pa) of commentaries that discuss the issue – suggests
that gTsaṅ nag pa’s portrayal of a transcendent ultimate does not
provide him with a good answer to rGya dmar pa’s and bSod nams
rtse mo’s criticisms. An ultimate beyond the ken of the intellect,
realization of which entails the cessation of awareness, provides no
nuanced way of presenting how a Buddha’s “conventional wisdom”
arises. Rather than answering bSod nams rtse mo’s critique of the
Prāsaṅgika “blind Buddha,” gTsaṅ nag pa changes the subject. One
senses that the early Tibetan tradition following Candrakīrti had to
set aside the thorny issue of where an utterly transcendent ultimate
left a Buddha’s perception of ordinary appearances.

The early evolution of Madhyamaka categories

We have seen how Śāntideva’s statements on ultimate truth and

the distinction between saṃsāra and nirvāṇa gave rise to compet-
ing interpretations, which in turn served as criteria for classify-
ing Mādhyamikas. Śāntideva’s ultimate could be “like an illusion”
or “utterly non-abiding;” his buddhahood could be Prāsaṅgika or
Svātantrika. One further discussion suggests how we might trace
an evolution among these twelfth-century Madhyamaka catego-
ries. Rather than a division between “illusionists” and “utterly
non-abiding-ists,” Grags pa rgyal mtshan, bSod nams rtse mo’s

69
gTsaṅ nag pa, sPyod ʼjug gi rnam bśad, 562,7–8 (38b7–8): raṅ bźin
gyis mya ṅan las ʼdas kyaṅ kun rdzob du ʼkhor bar ruṅ na lam la ʼbad pas
saṅs rgyas thob kyaṅ kun rdzob du ʼkhor ba mi ʼgal bas ʼkhor baʼi sdug
bsṅal mi ʼdor baʼi saṅs rgyas ʼthob pas ci bya źes pa la /
70
gTsaṅ nag pa, sPyod ʼjug gi rnam bśad, 562,8 (38b8).
 Making and remaking the ultimate 313

younger brother, writes of a division within the “utterly non-abid-

ing” Madhyamaka position between “Continuum Cutting Utterly
Non-Abiding [Mādhyamikas]” and “Union Utterly Non-Abiding
[Mādhyamikas].”71 In his discussion of these two groups, instead
of these unwieldy names Grags pa rgyal mtshan calls the former
“Prāsaṅgika” and the latter “Svātantrika.” His explanation of the
two categories maps very closely onto the divisions we have seen
rGya dmar pa and bSod nams rtse mo make: Prāsaṅgikas hold
that “Since Buddhas are always in meditative absorption, they
only [have] ultimate [awareness];”72 ordinary awareness perceiv-
ing appearances has had its continuum cut. The Svātantrika po-
sition, which Grags pa rgyal mtshan endorses, instead holds that
“Buddhas’ non-conceptual minds are ultimate; as [their] pure
worldly wisdom is supported by [non-conceptual] wisdom, it ac-
cords with the conventional.”73 Grags pa rgyal mtshan’s notion of

71
Grags pa rgyal mtshan, Rin po cheʼi ljon śiṅ, 21.4,1 (42b1): rgyun
chad rab tu mi gnas paʼi ʼdod tshul daṅ / zuṅ ʼjug rab tu mi gnas paʼi ʼdod
tshul.
72
Grags pa rgyal mtshan, Rin po cheʼi ljon shiṅ, 21.4,5–6 (42b5–6):
dbu ma thal ʼgyur pa dag ni … saṅs rgyas ni dus rtag tu mñam par gźag
pa yin pas don dam pa ʼbaʼ źig go źes zer ro / This passage is set within
a discussion of what kinds of awareness ordinary and enlightened beings
possess (21.3,6–21.4,1 [42a6–42b1]: blo thams cad bsdu na blo kun rdzob
pa daṅ / blo don dam pa źes bya ba gñis yin pas na / blo gñis po de gaṅ
zag gaṅ gi rgyud la ldan źe na / “When encompassing all awareness, there
are two: conventional awareness and ultimate awareness. Thus, one might
ask, ‘Which persons have these two awarenesses in their continua?’”).
Grags pa rgyal mtshan goes on to list four positions (the “Hearer” posi-
tion, the Yogācāra position, the Prāsaṅgika position, and the Svātantrika
position), detailing where each school stands on the mental states of “or-
dinary beings” (so so skye bo), āryas, and Buddhas.
73
Grags pa rgyal mtshan, Rin po cheʼi ljon śiṅ, 22.1,1–3 (43a1–3): dbu
ma raṅ rgyud pa dag … saṅs rgyas kyi thugs rnam par mi rtog pa ni
don dam pa yin la / dag pa ʼjig rten paʼi ye śes ni ye śes la dmigs nas
rnam graṅs kyi kun rdzob yin no / For a more substantial discussion of
this passage that discusses Grags pa rgyal mtshan’s views on “concordant
ultimates” and “concordant conventionals,” see Vose, Resurrecting Can-
drakīrti, 104–107.
314 Kevin Vose

“pure worldly wisdom” mirrors his brother’s “conventional wis-

dom;” both show how Buddhas can continue to perceive conven-
tional appearances along with their realization of emptiness. Grags
pa rgyal mtshan further notes that both kinds of wisdom are neces-
sary to explain non-abiding nirvāṇa (apratiṣṭhitanirvāṇa, mi gnas
paʼi mya ṅan las ʼdas pa), in which Buddhas are both fully realized
and fully able to aid sentient beings.74
Grags pa rgyal mtshan’s discussion suggests that, at least in one
reckoning, Prāsaṅgika and Svātantrika evolved out of the “utterly
non-abiding position” and that neither could be mapped onto the
“illusionist” position, which was widely rejected in bKaʼ gdams pa
circles. Having cast aside the “illusionist” view, early bKaʼ gdams
pa authors (and their Sa skya pa students) found significant enough
philosophical differences to split further the Madhyamaka view.
Despite a mutual adherence to the “non-abiding” of all phenomena,
disputes over buddhahood produced, at least in part, the Prāsaṅgika-
Svātantrika divide.75 The only outlier category is gTsaṅ nag pa’s
“Great Madhyamaka,” interestingly posited by the one Candrakīrti
supporter in this group of commentators. Above, we saw that gTsaṅ
nag pa’s “Great Madhyamaka” refers to Candrakīrti’s revivalists,
whom others in this time period would call “Prāsaṅgikas.” The
similarities between gTsaṅ nag pa’s views and those of Jayānanda,
as well as the positions that bSod nams rtse mo and Grags pa rgyal
mtshan criticize (and label “Prāsaṅgika”), strengthens this identi-

74
After stating the Prāsaṅgika position, Grags pa rgyal mtshan notes,
“That is not correct; it incurs the fault that it would absurdly follow that
Buddhas would not enter non-abiding nirvāṇa” (Rin po cheʼi ljon śiṅ,
21.4,6 [42b6]: deʼaṅ yaṅ dag pa ma yin te / saṅs rgyas rnams ni mi gnas
paʼi mya ṅan las ʼdas pa la bźugs pa ma yin par thal ʼgyur baʼi skyon
yod do /). Following his statement of the Svātantrika position, he notes,
“Thus, there is non-abiding nirvāṇa” (Rin po cheʼi ljon śiṅ, 22.1,3–4
[43a3–4]: des na mi gnas paʼi mya ṅan las ʼdas paʼaṅ yin la /).
75
This, of course, discounts those like rṄog and Gro luṅ pa who re-
jected the “non-abiding” position out of hand. However, as noted above,
both understood the “non-abiding” view quite differently from how it is
portrayed by rGya dmar pa and Phya pa.
 Making and remaking the ultimate 315

fication. Why then doesn’t gTsaṅ nag pa take up the “Prāsaṅgika”

moniker?
To sketch a tentative answer, we can note that bSod nams rtse
mo’s discussion of Prāsaṅgika and Svātantrika, when explicating
Śāntideva’s distinction between saṃsāra and nirvāṇa, skews the
playing field against the former, equating it with the clearly prob-
lematic stance that awareness is cut off upon buddhahood. Rather
than take up the “Prāsaṅgika” side of the debate on this issue –
and thereby attempt to defend a blind Buddha – gTsaṅ nag pa ig-
nored the dispute altogether. At least on this issue, the Svātantrika
position had a clear advantage over Prāsaṅgika in twelfth century
Tibet.76 In contrast, “Great Madhyamaka” would represent gTsaṅ
nag pa’s attempt to divide up Madhyamaka with Candrakīrti on
top. Rather than make a case for Candrakīrti’s superiority on the
issue of a Buddha’s perception, gTsaṅ nag pa uses Candrakīrti’s
portrayal of a transcendent ultimate, utterly distinct from mundane
objects of knowledge, to valorize the “Great Madhyamaka” view.
In the context of early Tibetan Bodhicaryāvatāra commentaries,
these evolving Madhyamaka categories attempt to give the author’s
system top billing and to claim Śāntideva as their own. gTsaṅ nag
pa aligns Śāntideva’s ultimate with Candrakīrti’s, adopting him
into Great Madhyamaka. The host of Svātantrika-leaning com-
mentators instead nuanced Śāntideva’s stanzas on ultimate truth
and the saṃsāra-nirvāṇa distinction in order to forge a more co-
herent model of ultimate truth and those transformed by realization
of it. rGya dmar pa and bSod nams rtse mo would have us believe
that Śāntideva’s “true thought” aligns with what would come to be
called “Svātantrika.”
Śāntideva, then, was called many things by early Tibetan com-
mentators. Fidelity to Śāntideva is difficult to judge given the poetic
nature of his work. It may well be easier to square Śāntideva’s state-
ment that “the ultimate is not a referent of awareness” with a tran-

76
This is not to say that no Candrakīrti partisans adopted the label
“Prāsaṅgika” in this period. Georges Dreyfus and Drongbu Tsering’s ar-
ticle in this present volume shows that Pa tshab Ñi ma grags indeed used
the term “Prāsaṅgika” for his Candrakīrti-inspired Madhyamaka.
316 Kevin Vose

scendent reading, consistent with how Candrakīrti’s twelfth cen-

tury revivalists portrayed his views. However, reading Śāntideva
as advancing views similar to those of Śrīgupta, Jñānagarbha,
Śāntarakṣita, or Kamalaśīla held historical credibility and re-
mained a viable option for early Tibetan commentators. We see
in these commentaries a variety of ways to reconcile Śāntideva’s
seemingly transcendent ultimate with his broader project of map-
ping the practices of enlightenment. Harmonizing the ultimate
with the path leading to its realization is an issue at the very core of
Mahāyāna Buddhism; harmonizing realization of the ultimate with
the kind of epistemological program that gSaṅ phu Neʼu thog was
known for would remain a defining feature of Tibetan Buddhist
scholasticism. Early Tibetan Bodhicaryāvatāra commentaries pro-
vide a glimpse of how these processes played out in the making and
remaking of one of the preeminent figures of Buddhist India.

References

Sanskrit and Tibetan works cited

Grags pa rgyal mtshan. rGyud kyi mṅon par rtogs pa rin po cheʼi ljon śiṅ. In
Sa skya paʼi bkaʼ ʼbum, volume 3. Tokyo: Toyo Bunko, 1968.
Gro luṅ pa Blo gros ʼbyuṅ gnas. bsTan rim chen mo. Mundgod: Library of
His Eminence Trijang Rinpoche, 2001. Also, bKaʼ gdams gsuṅ ʼbum, vol-
umes 4 and 5.
rGya dmar pa Byaṅ chub grags. Byaṅ chub sems dpaʼi spyod pa la ʼjug paʼi
tshig don gsal bar bśad pa. In bKaʼ gdams gsuṅ ʼbum, volume 6, 11–175.
Jayānanda. Madhyamakāvatāraṭīkā (dBu ma la ʼjug paʼi ʼgrel bśad). bsTan
ʼgyur, sDe dge edition, Tohoku 3870; dbu ma, volume ya.
bKaʼ gdams gsuṅ ʼbum phyogs bsgrigs. Chengdu: dPal brtsegs bod yig dpe
rñiṅ źib ʼjug khaṅ, 2006.
rṄog Blo ldan śes rab. sPriṅ yig bdud rtsi thig le. In bKaʼ gdams gsuṅ ʼbum,
volume 1, 707–710.
Phya pa Chos kyi seṅ ge. dBu ma śar gsum gyi stoṅ thun, edited by Helmut
Tauscher. Vienna: Arbeitskreis für Tibetische und Buddhistische Studien
Universität Wien, 1999.
— bDe bar gśegs pa daṅ phyi rol paʼi gźung rnam par ʼbyed pa. In bKaʼ
gdams gsuṅ ʼbum, volume 9, 7–73.
 Making and remaking the ultimate 317

— sPyod ʼjug bsdus don. In bKaʼ gdams gsuṅ ʼbum, volume 7, 131–145.
Prajñākaramati. Bodhicaryāvatārapañjikā. See La Vallée Poussin.
Śāntideva. Bodhicaryāvatāra (Byaṅ chub sems dpaʼi spyod pa la ʼjug pa).
bsTan ʼgyur, sDe dge edition, Tohoku 3871; dbu ma, volume ya.
bSod nams rtse mo. sPyod pa la ʼjug paʼi ʼgrel pa. In Sa skya paʼi bkaʼ ʼbum,
volume 2. Tokyo: Toyo Bunko, 1968.
gTsaṅ nag pa brTson ʼgrus seṅ ge. sPyod ʼjug gi rnam bśad. In bKaʼ gdams
gsuṅ ʼbum, volume 13, 487–581.
Vibhūticandra. Bodhicaryāvatāratātparyapañjikā Viśeṣadyotanī (Byaṅ chub
kyi spyod pa la ʼjug paʼi dgoṅs paʼi ʼgrel pa khyad par gsal byed). bsTan
ʼgyur, sDe dge edition, Tohoku 3880; dbu ma, volume śa.

Western language works cited

Cabezón, José Ignacio. “The Madhyamaka in Gro lung pa’s Bstan Rim chen
mo.” Proceedings of the Eleventh Seminar of the International Association
of Tibetan Studies. Forthcoming.
Kano, Kazuo. rNgog Blo-ldan-shes-rab’s Summary of the Ratnagotravibhāga:
The First Tibetan Commentary on a Crucial Source for the Buddha-
nature Doctrine. Ph.D. Dissertation, Hamburg University, 2006.
Kramer, Ralf. The Great Translator: Life and Works of rNgog Blo ldan shes
rab (1059–1109). Munich: Indus Verlag, 2007.
La Vallée Poussin, Louis de. Prajñākaramati’s Commentary to the Bodhi-
caryāvatāra of Çāntideva, Bibliotheca Indica. Calcutta: Asiatic Society
of Bengal, 1905–1914.
— Madhyamakāvatāra par Candrakīrti, Bibliotheca Buddhica 9. Osnabrück:
Biblio Verlag, 1970.
Mathes, Klaus-Dieter. “Blending the Sūtras with the Tantras: The Influence
of Maitrīpa and his Circle on the Formation of Sūtra Mahāmudrā in the
Kagyu Schools.” In Tibetan Buddhist Literature and Praxis: Studies in
its Formative Period 900–1400, Proceedings of the Tenth Seminar of the
International Association of Tibetan Studies, Oxford 2003, volume 4,
edited by Ronald M. Davidson and Christian K. Wedemeyer, 201–227.
Leiden: Brill, 2006.
Napper, Elizabeth. Dependent-Arising and Emptiness. Boston: Wisdom
Publications, 1989.
Oldmeadow, Peter R. A Study of the Wisdom Chapter (Prajñāpāramitā Pa-
riccheda) of the Bodhicaryāvatārapañjikā of Prajñākaramati. Ph.D.
Dissertation, Australian National University, 1994.
Ruegg, David Seyfort. Three Studies in the History of Indian and Tibetan
Madhyamaka Philosophy, Studies in Indian and Tibetan Madhyamaka
318 Kevin Vose

Thought, Part One. Vienna: Arbeitskreis für Tibetische und Buddhistische

Studien Universität Wien, 2000.
— Two Prolegomena to Madhyamaka Philosophy, Candrakīrti’s Prasanna-
padā Madhyamakavṛttiḥ on Madhyamakakārikā I.1 and Tsoṅ kha pa Blo
bzaṅ grags pa / rGyal tshab Dar ma rin chen’s dKaʼ gnad/gnas brgyad
kyi zin bris, Annotated Translations, Studies in Indian and Tibetan Ma-
dhyamaka Thought, Part 2. Vienna: Arbeitskreis für Tibetische und
Buddhistische Studien Universität Wien, 2002.
Saito, Akira. “Śāntideva in the History of Mādhyamika.” In Buddhism
in India and Abroad, edited by Kalpakam Sankarnarayan, Motohiro
Yoritomi, and Shubhada A. Joshi, 257–263. Mumbai and New Delhi:
Somaiya Publications, 1996.
— “Bu ston on the sPyod ʼjug (Bodhisattvacaryāvatāra).” In Tibetan
Studies: Proceedings of the 7th Seminar of the International Association
for Tibetan Studies, edited by Helmut Krasser, Michael Torsten Much,
Ernst Steinkellner, and Helmut Tauscher, 79–85. Vienna: Verlag der
Österreichischen Akademie der Wissenschaften, 1997.
— A Study of the Dun-huang Recension of the Bodhisattvacaryāvatāra. Mie:
Mie University, 2000.
Stearns, Cyrus. “The Life and Tibetan Legacy of the Indian Mahāpaṇḍita
Vibhūticandra.” Journal of the International Association of Buddhist
Studies 19, no. 1 (1996): 127–161.
Vose, Kevin. Resurrecting Candrakīrti: Disputes in the Tibetan Creation of
Prāsaṅgika. Boston: Wisdom Publications, 2009.
Williams, Paul. “On Prakṛtinirvāṇa/Prakṛtinirvṛta in the Bodhicaryāvatāra:
A Study in the Indo-Tibetan Commentarial Tradition.” Études Asiatiques
46, no. 1 (1992): 516–550.