Can The Veda Speak - Dharmakirti

Contents
Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
The Place of PVSV 164,24–176,16 in the Work of Dharmakīrti . . . . . . . . 7
A Synopsis of PV(SV) 1.312–340 . . . . . . . . . . . . . . . . . . . . . . . . . 23
Part 1 – Translation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Appendix: Philological Remarks . . . . . . . . . . . . . . . . . . . . . . . . . 79
Part 2 – Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Logic in a Religious Context: Dharmakīrti in Defence of āgama . . . . . . 83
Dharmakīrti and the Mīmāṃsakas in Conflict . . . . . . . . . . . . . . . . . 119
Abbreviations and Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . 151
Indices
Index locorum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Sanskrit texts and genres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
Place names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
Sanskrit and other names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Gods and saints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Schools and religious groups . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Sanskrit words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Tibetan words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Modern authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

VINCENT ELTSCHINGER, HELMUT KRASSER, JOHN TABER
CAN THE VEDA SPEAK?

DHARMAKĪRTI AGAINST MĪMĀṂSĀ EXEGETICS AND VEDIC
AUTHORITY. AN ANNOTATED TRANSLATION OF PVSV 164,24–176,16
ÖSTERREICHISCHE AKADEMIE DER WISSENSCHAFTEN
PHILOSOPHISCH-HISTORISCHE KLASSE
SITZUNGSBERICHTE, 831. BAND
BEITRÄGE ZUR KULTUR- UND GEISTESGESCHICHTE ASIENS

NR. 74
Herausgegeben von Helmut Krasser

ÖSTERREICHISCHE AKADEMIE DER WISSENSCHAFTEN
PHILOSOPHISCH-HISTORISCHE KLASSE
SITZUNGSBERICHTE, 831. BAND
Can the Veda speak?

Dharmakīrti against Mīmāṃsā exegetics and
Vedic authority. An annotated translation of
PVSV 164,24–176,16
Vincent Eltschinger, Helmut Krasser, John Taber

Vorgelegt von k. M. HELMUT KRASSER
in der Sitzung am 17. Juni 2011
British Library Cataloguing in Publication data

A Catalogue record of this book is available from the British Library
Die verwendete Papiersorte ist aus chlorfrei gebleichtem Zellstoff hergestellt,

frei von säurebildenden Bestandteilen und alterungsbeständig.
Alle Rechte vorbehalten

ISBN 978-3-7001-7129-4
Copyright © 2012 by
Österreichische Akademie der Wissenschaften
Wien
Druck und Bindung: Ferdinand Berger & Söhne Ges.m.b.H., A-3580 Horn
Printed and bound in the EU
http://hw.oeaw.ac.at/7129-4
http://verlag.oeaw.ac.at
Foreword

The project presented in this volume began in fall of 2006, when John
Taber was Visiting Professor of Buddhist Studies at the Institute for
South Asian, Tibetan and Buddhist Studies of the University of Vienna
for winter term, 2006–07, replacing Prof. Ernst Steinkellner, who had
just retired. Dr. Vincent Eltschinger of the Institute for the Cultural and
Intellectual History of Asia (IKGA) of the Austrian Academy of Sciences,
who had been working for some years on Dharmakīrti’s engagement
with Mīmāṃsā and whose translation and study of Pramāṇavārtti‐
ka(svavṛtti) 1.213–268 was about to appear, proposed to Prof. Taber
that they read some text of common interest together while he was in
Vienna, preferably one relating to the Buddhist‐Mīmāṃsaka debates.
They chose PV(SV) 1.312–340. This passage is the culmination of
Dharmakīrti’s critique of the Mīmāṃsā theory of the authorlessness of
the Veda as the foundation of its authority. It is closely related to the
extensive portions of the Mīmāṃsā section of PV 1, which begins with
v. 213, that Eltschinger had already translated; although there is a Jap‐
anese translation of the passage,1 no translation of it into a European
language exists. After selecting this text Taber and Eltschinger invited
Dr. Helmut Krasser, soon to become Director of the IKGA, who had
been working on issues relating to Dharmakīrti’s understanding of
scripture, to participate. Then, to everyone’s delight, Steinkellner also
expressed interest in joining the group.
Starting in November, 2006, the group began meeting once a week,
usually on Tuesday afternoons, in the seminar room of the IKGA, then
located in Strohgasse. The sessions were conducted in English, since
the intention was to produce an English translation. Eltschinger brave‐
ly provided a draft translation of the section to be covered each day.
The sessions were marked by lively exchanges (including challenges to
Taber to justify his intuitions of English usage by looking up words in
Webster’s Dictionary!). On several occasions – as he has done over the
1 See Wakahara 1990.

2 FOREWORD

years in many collaborations he has organized at the IKGA – when the
group was stumped, Steinkellner succeeded in pulling a rabbit out of a
hat. The group met approximately 7–8 times, until Taber’s departure at
the end of January, 2007. By then, they had reached p. 167, l. 8. All were
pleased with the results thus far and felt that every effort should be
made to continue the collaboration.
It was not until summer, 2008, however, that Taber was able to
break free from his duties as Chair of the Philosophy Department at the
University of New Mexico (UNM) and return to Vienna. For three
weeks, from July 21 to August 16, the team of Eltschinger, Krasser, and
Taber met nearly every afternoon for four to five hours, in the new fa‐
cilities of the IKGA in Prinz‐Eugen‐Strasse, to hammer out a draft of the
remaining ten pages of the text. (Taber’s trip to Vienna was financed by
the College of Arts and Sciences of the University of New Mexico and
the IKGA.) Steinkellner was occupied with other projects by this time.
Eltschinger, as before, prepared the initial draft translation; the ses‐
sions were this time conducted in German. By Taber’s departure the
team had produced a translation of the entire passage.
Over the ensuing months Eltschinger prepared the notes to the
translation. Then the project lay mostly dormant until fall, 2009, when
Taber, thanks to a teaching release from UNM, was able to copy‐edit
the entire translation. The results of the copy‐editing and remaining
problems with the translation were discussed over e‐mail that fall, con‐
tinuing into January, 2010, when Taber also wrote the synopsis. In
spring of 2010 attention shifted to the notes, as Taber and Krasser also
worked on their essays. (Eltschinger’s essay had been ready for some
time.) Steinkellner also kindly read the translation and offered com‐
ments and suggestions, which were taken into account in a final
proofing of the manuscript in fall, 2010. Electronic formatting and
other technical tasks were in the hands of Krasser throughout the pro‐
ject.
From the outset, the team endeavoured to provide a readable Eng‐
lish translation. At the same time, they sought to base their translation
on a close reading of Śākyabuddhi’s and Karṇakagomin’s commen‐
taries, which proved, however, to be too long to be quoted and trans‐
lated in full in the footnotes (many of the footnotes summarize these
commentaries’ insights). The translators drew occasional inspiration
from Manorathanandin’s commentary and also took into consideration
FOREWORD 3

Dharmottara’s commentary on those stanzas of PV 1 that recur in PVin
2. Three essays of varying length and function round off the transla‐
tion. Eltschinger’s introduction locates the translated section in the
overall economy of the PV(SV); Krasser’s essay focuses on Dharma‐
kīrti’s views on āgama, which serve as an important part of the back‐
ground of the text; Taber’s contribution is meant to do justice to the
system of Mīmāṃsā exegesis that is being attacked by Dharmakīrti.
Since Krasser’s and Taber’s essays can be considered as contributions
in their own right, they are treated here as appendices to the transla‐
tion. Needless to say, the present study does not presume to provide
the last word on this extremely rich final section of Dharmakīrti’s PV 1.
It is, rather, meant to draw attention to the deep involvement of exe‐
getical and apologetic issues in the Buddhist master’s overall philo‐
sophical project.
It is the authors’ pleasure to acknowledge their gratitude to Eli
Franco, Elisa Freschi, Kunio Harikai, Kei Kataoka, Shoryu Katsura, Isa‐
belle Ratié, François Voegeli and, last but not least, Kiyotaka Yoshimizu
for their helpful comments and suggestions.
The funds for printing this book were generously provided by the
Austrian Science Fund (FWF), for which we would like to express our
gratitude.

Albuquerque and Vienna,
Spring 2011

Introduction

VINCENT ELTSCHINGER
The Place of PVSV 164,24–176,16 in the Work

of Dharmakīrti

1.1. Dharmakīrti is likely to have belonged to one of the first gener‐
ations of those early medieval Buddhist intellectuals who attempted
to answer an ever‐increasing Brahmanical hostility towards Bud‐
dhism and other non‐ or anti‐Vedic movements. The first among
these thinkers was most certainly Dignāga (480–540, or slightly
earlier), the author of several (now lost) tracts against the Nyāya,
the Vaiśeṣika, and the Sāṅkhya.1 Together with the Mīmāṃsā and
the author of the Vādavidhi, these schools were again the target of
Dignāga’s last and major work, the Pramāṇasamuccaya (PS). During
more than a century Dignāga’s version of the hetuvidyā remained
the standard system studied at Nālandā. Dignāga, however, seems to
have limited the scope of his answer to issues of “logic” and “epis‐
temology.” Dharmakīrti was the one who provided Dignāga’s
thought with an ontology, a psychology, and a religious doctrine,
thus building a genuinely Buddhist system able both to respond to
Brahmanical criticism and to defend Buddhism as a salvational
path.2 As a representative of the most uncompromising Brahmanical
orthopraxy,3 the Mīmāṃsā and its doctrine of the authorlessness
(apauruṣeyatva) of the Veda had recently become Buddhism’s mor‐
tal enemy. It is thus hardly surprising that Dharmakīrti’s main op‐
ponent was Mīmāṃsā, quite probably in the person of its most bril‐
liant classical exponent, Kumārila Bhaṭṭa (Dharmakīrti’s senior con‐
temporary).

1 On Dignāga, see Frauwallner 1959 and Hattori 1968.
2 General introductions to Dharmakīrti’s thought include Steinkellner 1971,

Stcherbatsky 1984, Katsura 1984, Dreyfus 1997, Dunne 2004, Eltschinger 2010.
3 Rather than “orthodoxy.” See McCrea 2009: 55.
8 VINCENT ELTSCHINGER
1.2. We know next to nothing of Dharmakīrti’s life, though he is

likely to have been born to a Brahmin family and have spent at least
part of his career in Nālandā. His dates are still a matter of contro‐
versy. According to Frauwallner’s chronology, Dharmakīrti’s dates
must be around 600–660 CE. However, on the basis of very different
arguments, these dates have been challenged by Lindtner, Kimura,
and Krasser, who are inclined to locate Dharmakīrti in the sixth cen‐
tury CE.4 Whatever the case may be, the first non‐philosophical
testimony regarding Dharmakīrti is Yijing’s remark to the effect that
Wuxing had studied his system in Telāḍhaka (near Nālandā) under
the guidance of Jñānacandra. Dharmakīrti’s works are traditionally
held to amount to seven.5 Among them, the Pramāṇavārttika (PV) is
the first and by far the most important in that it shows us Dhar‐
makīrti in his philosophical workshop. His later works mainly con‐
sisting of didactic digests or treatises dealing with specific topics,
the PV is the fountainhead of Dharmakīrti’s thought, the work to
which any serious account of his system should primarily refer. As
demonstrated by Frauwallner, the PV is a composite work compo‐
sed of at least two layers.6 Its first chapter (PV 1 = Svārthānumāna‐
pariccheda), dedicated to the three valid logical reasons and accom‐
panied by a prose auto‐commentary (PVSV), is likely to have origi‐
nally formed an independent treatise (entitled Hetuprakaraṇa ac‐
cording to Frauwallner’s – groundless – hypothesis). The last three
chapters, entirely in verse, present themselves as a very loose and
independent commentary on Dignāga’s PS (PV 2 = Pramāṇasiddhi‐
pariccheda on the maṅgalaśloka of the PS, PV 3 = Pratyakṣapa‐
riccheda on PS 1, and PV 4 = Parārthānumānapariccheda on PS 3).
The present study consists in an annotated translation of the con‐
cluding section of PV 1, viz., PV 1.312–340 and PVSV 164,24–176,16.

1.3. Although PV 1/PVSV deals with the three kinds of logical rea‐
sons (hetu, viz., essential property, effect, and non‐perception),
4 See Krasser 2012. For a summary of research before Krasser, see Eltschinger
2007a: 25–29. I accept Krasser’s new chronology.

5 On Dharmakīrti’s works, see Frauwallner 1954 and Steinkellner/Much 1995:
23–44.
6 See Frauwallner 1954: 142–151.
THE PLACE OF PVSV 164,24–176,16 9
hence with private inferential judgment (svārthānumāna), this topic

is overshadowed by two lengthy, seventy page‐long digressions.7
The first one (= PV 1.40–185/PVSV 24,16–93,5) focuses on the so‐
called apoha theory (Dharmakīrti’s innovative account of the two
truths), which presents Dharmakīrti’s views on the genesis of con‐
cepts, language, error, and practice.8 The second one deals with
Dharmakīrti’s understanding of scriptural authority (āgamaprā‐
māṇya) and mainly consists in his polemics against the Mīmāṃsaka
doctrine of the authorlessness of the Veda (PV 1.213–338/PVSV
107,14–176,4).9 One should be wary of considering Dharmakīrti’s
polemics against authorlessness as a mere appendix to PV 1. To put
it briefly, this doctrine claims that human judgment (conditioned by
moral laxity) is the source of error and that human beings lack any
cognitive access to the supersensible realm. As a consequence, only
an authorless scripture – the Veda – can provide unbiased access to
the invisible Dharma. And this claim is nothing but an attack on
Buddhism, Jainism, Sāṅkhya, and other emerging religious move‐
ments (like Pāñcarātra and Pāśupata) deriving the authority of their
scriptures from the trustworthiness (āptatva) of their human or di‐
vine authors.10 But the Veda is far from being the only point at stake.
For according to Mīmāṃsā, all that can be proven to be rooted in
this authorless Veda (vedamūla) is authoritative, including those
7 Gnoli’s edition of the PVSV has 176 pages. The two excursus amounting to about
140 pages, the systematic treatment of the three kinds of valid reasons does not
exceed 35 pages.
8 The most thorough account of Dharmakīrti’s apoha theory remains Frauwall‐
ner’s (see Frauwallner 1932 and 1933). See also Dunne 2004: 113–144.
9 For a more precise analysis of this digression, see below, pp. 10–12. This digres‐
sion on scriptural authority occurs towards the end of Dharmakīrti’s treatment

of non‐perception as a logical reason and returns to it at the very end of our pas‐
sage, viz., PV 1.339/PVSV 176,6–12. The question raised by Dharmakīrti’s oppo‐
nent can be summarized as follows: Since there is no object left out of considera‐
tion by scripture (āgama), if scripture keeps silent on a certain object, this ob‐
ject can be considered as non‐existent. Non‐perception can thus be defined as
the silence (nivṛtti) of the three means of valid cognition (pramāṇa). As Dharma‐
kīrti makes clear at PV 1.339, the aim of this lengthy excursus is to prove that
the silence of scripture doesn’t allow one to ascertain the non‐existence of a cer‐
tain entity.
10 See McCrea 2009.
Smṛtis and Purāṇas that are replete with an ostensible hostility to‐
wards Buddhism and other “heretical” (pāṣaṇḍa) denominations. In
other words, authorlessness is nothing but an attempt to “natural‐
ize” Vedic ritual and Brahmanocentric society, to legitimize the hos‐
tility of these “secondary” scriptures to movements challenging the
Brahmanical order. This is something Dharmakīrti was most cer‐
tainly aware of, something which, if accepted, represented a serious
danger for Buddhism.

1.4. After a comparatively short theoretical elaboration on scrip‐
tural authority and its criteria (PVSV 107,14–112,6),11 Dharmakīrti
starts criticizing the doctrine of authorlessness by attacking the
Mīmāṃsakas’ and the Grammarians’ account of the relation be‐
tween word and meaning (śabdārthasambandha; PVSV 113,8–
120,7), which these schools take to be permanent (nitya, but in the
sense of sāṃvyavahārikanitya, whereas Dharmakīrti’s portrayal of
their position interprets it as kūṭasthanitya). Dharmakīrti then turns
to the Mīmāṃsaka arguments in favour of the eternality (anāditā,
viz., beginninglessness; PVSV 120,8–126,15) of the Veda and Vedic
memorization‐cum‐recitation (adhyayana). His discussion returns
to linguistic issues as he sets about criticizing the two main classical
accounts of authorlessness or, at least, permanent speech: first, the
Grammarians’ sphoṭa (according to which meaningful linguistic
units are “transphonetic” but revealed by ephemeral articulated
sounds; PVSV 126,24–134,25), and second, the Mīmāṃsā’s (more
precisely, Kumārila’s) varṇavāda (according to which language con‐
sists of phonemes that are one and permanent [kūṭasthanitya], ap‐
pear in permanent [sāṃvyavahārikanitya] series and are revealed
11 Properly speaking, Dharmakīrti’s own doctrine is expounded in PVSV 107,14–
109,22 (see Yaita 1987; Dunne 2004: 361–366; Eltschinger 2007a: 217–227; see
also below, pp. 83–118). It is followed by a short critique of (mainly) Naiyāyika
views on āptavāda (PVSV 109,23–110,15), by a rebuttal of the Mīmāṃsaka con‐
tention that a morally immaculate āpta is impossible (PVSV 110,15–111,11),
and by Dharmakīrti’s own views on nescience as personalistic belief (PVSV
111,11–112,5). On this, see Yaita 1988, Dunne 2004: 366–373, Eltschinger
2007a: 227–239. Note that the present outline of PVSV 107,14–176,16 is but a
very cursory one leaving several aspects out of consideration.
by ephemeral articulated sounds; PVSV 134,26–141,14).12 In all this,

Dharmakīrti takes his adversaries’ claims about permanence for
granted and uses them in unceasing reductiones ad absurdum. There
follows, then, a passage commonly known as the vināśitvānumāna
(“inference of perishability”13) where Dharmakīrti improves on
Vasubandhu’s arguments to the effect that all entities perish as soon
as they are produced (PVSV 141,14–150,5). After a few elaborations
of a more logical character, Dharmakīrti returns to a topic already14
dealt with while criticizing the beginninglessness of the Veda, viz.,
that of mantras and their efficacy (PVSV 155,18–164,24).15 His argu‐
ment, here, is as follows: If, as the Mīmāṃsā claims, the Vedic man‐
tras were authorless, then the Brahmins couldn’t arbitrarily restrict
their efficacy to the twice‐born, excluding women and śūdras. In
other words, authorless mantras ought to yield their results for
every possible user independently of his or her social background.16
Either the Vedic mantras are authorless (and then their efficacy is
the same for everyone), or they are of human origin, hence arbitrar‐
ily controllable by their “author” (kartṛ) as to their fruitfulness and
the conditions (social, ritual, behavioural, psychological, devotional)
that have to be fulfilled for them to bring about their results (and
then the Mīmāṃsaka doctrine falls to the ground). Then comes the
concluding part of the chapter (PVSV 164,24–176,16), the one
translated in the present book.17
12 On all this, see Eltschinger 2007a: 239–385.
13 See Sakai 2011.
14 See
PVSV 123,14–124,28 and Eltschinger 2007a: 299–307. On Dharmakīrti’s
position on mantras, see Eltschinger 2001 and 2008. See also Wakahara 1988.
15 Mantras proper do not exhaust PVSV 155,18–164,24. This passage also contains
Dharmakīrti’s own doctrine concerning phonetic series (PVSV 157,29–162,11).

See Ōmae 1999 and Eltschinger 2007a: 204–212.
16 This conclusion is only legitimate once it has been proven that “castes” are noth‐
ing but arbitrary conventions, which Dharmakīrti duly demonstrates in PVSV

157,10–18. See Eltschinger 2000: 103–115.
17 Wakahara 1990, which provides a Japanese translation of the same passage, and
Eltschinger 2001: 94–101, are the only secondary literature I am aware of. But
since Dharmakīrti quotes PV 1.317–329 in his PVin 2 (70,6–72,8) and briefly
deals there (PVin 2 72,10–11) with the ekadeśāvisaṃvāda argument (see PV
1.330–335 and PVSV 173,14–175,10, below, pp. 61–73; see also below, pp. 18–
21), Steinkellner 1979: 73–78 is to be considered as an important contribution

to our passage. Note, however, that PVin 2.39, 43 and 44 do not belong to the PV
(see below, p. 53 n. 89 and p. 55 n. 97). Dharmottara’s general introduction to
PVin 2.35–50 provides us with a fascinating interpretation of the meaning of our
passage (PVinṬms 71a4–b3/PVinṬTib Dze 227b4–228a2): na vāstavaḥ kaścit sam‐
bandhaḥ kiṃ tu sāmayikaḥ / icchāto ’rthābhidhānād iti / ayaṃ ca vāstavasam‐
bandhavādaḥ pramāṇavārttike vistareṇa niṣiddha ity atra na vistāryate / iha ca
vedāpauruṣeyatve na varṇānāṃ padānāṃ vā nityatvam upakāri / loke vede ca
teṣāṃ viśeṣābhāvād rathyāpuruṣavākyasyāpi prāmāṇyaprasaṅgāt / kevalaṃ vā‐
kyam evāpauruṣeyaṃ pratipādayitavyaṃ vākyārthaś ca / ācāryasya tu padarū‐
paṃ pauruṣeyaṃ pratipādayato ’yam abhiprāyaḥ / padād vākyaṃ na vyatiricya‐
te / tasya pauruṣeyatve vākyasyāpi pauruṣeyatvam / vākyarūpe ca pauruṣeye
’rtho ’pi pauruṣeyaḥ sidhyati / ato vedaḥ pauruṣeyaḥ siddho bhavati / athāpi
pauruṣeyāṇāṃ padānāṃ samāhārātmakaṃ vākyam apauruṣeyaṃ padāny eva vā
nityāni syuḥ / tathāpy arthasyāpauruṣeyatve vedaprāmāṇyaṃ na rūpasyeti / pa‐
dānāṃ nityatvaṃ tadracanātmakasya ca vākyasyāpuruṣaracitatvaṃ vārttike
nirākṛtam apy abhyupagamya vārttikakārikābhir vāstavārthasambandhaniṣe‐
dhaṃ darśayati / vaidikānāṃ śabdānāṃ vākyabhūtānāṃ padabhūtānāṃ vā ’rtho
na kenacid api śakyo niścetuṃ puruṣopadeśaṃ vineti darśayitum āha. “There is
no real relation [between word and meaning], but a [purely] conventional [one],
because the expression of a meaning is according to [a certain speaker’s] inten‐
tion (icchā). But since the doctrine according to which the relation [between
word and meaning] is real has [already] been refuted extensively in the
Pramāṇavārttika, [Dharmakīrti] does not develop [it] here [in the Pramāṇaviniś‐
caya]. But [neither] is the [alleged] permanence of the phonemes or of the
words of any help to the authorlessness [and, hence, to the authority] of the
Veda, because, since these [phonemes or words] do not differ in ordinary [lan‐
guage] and in the Veda, it would follow that even the statement of a man in the
street would be authoritative. Rather, it is the sentence and the meaning of the
sentence that must be shown to be authorless. But the following is the intention
of Ācārya [Dharmakīrti] in showing that the nature of the word is of human
origin (pauruṣeya): [Since] the sentence is not distinct from the word, if the
[word] is of human origin, [then] the sentence is of human origin also. And if the
nature of the sentence is of human origin, [then its] meaning too is established
as of human origin, [and] therefore, the Veda is established as of human origin.
But even if the sentence consisting of an aggregation of words of human origin
were authorless, or [if] the words themselves were permanent, the Veda would
[only] be authoritative if [its] meaning [, and] not [its verbal] form, were
authorless. [Now,] although [Dharmakīrti] has refuted, in the [Pramāṇa]vārttika,
that words are permanent and that a sentence consisting of an arrangement of
these [words] is not created by humans, [he now provisionally] accepts [it] and
presents, by [resorting to] the stanzas of the [Pramāṇa]vārttika, a refutation of a
real relation with the meaning. [And] in order to show that no one is able to as‐

1.5 Up to this point, Dharmakīrti has demonstrated the possibility,
even the necessity, of atīndriyadarśana. That mantras yield results
is due to the fact that certain beings (including humans, gods, bodhi‐
sattvas, etc.) are able to discern, hence to perceive, which phonetic
series is efficient and which is not; that these superior beings (pu‐
ruṣātiśaya, puruṣaviśeṣa) prescribe, as a covenant (samaya; glossed
pratijñā), ethical and behavioural conditions for the fruition of man‐
tras; and, finally, to the fact that they empower (adhi√ṣṭhā) these
mantras so that they benefit ordinary persons (prākṛtapuruṣa).
Such is the conclusion of his elaborations on mantras:
It is this [demonstration that the creators of mantras possess a supe‐
riority in cognition and power] that dismisses the [Mīmāṃsaka] proof
that humans are incapable [of perceiving supersensible things]. In‐
deed, [we have] shown that the mantras are made by humans and
yield results due to their [the humans’] empowerment. Therefore,
since there is someone who possesses a [cognitive] superiority, the
[Mīmāṃsaka] proofs that [supposedly] negate this [cognitively supe‐
rior person] are also refuted. As for [that] which is regarded as a proof
[by the Mīmāṃsaka, viz., reasons] like cognition, sense‐faculties, lan‐
guage, and humanity, [all this is nothing but] an [inconclusive]
pseudo‐argument (pramāṇābha),18 for there is no correct cognition
[of a state of affairs] from [a reason] possessing a remainder. As for
the proof that [supposedly] negates [the existence of] a superior per‐
son, namely, ‘Because, since he possesses cognition, sense‐faculties,
and language, [the alleged creator of a mantra] is a human being [, like
every other human being],’ this [proof] is strictly unconvincing, be‐
cause there can be proofs neither of the negation nor of the sameness.
[That is to say,] in the case of supersensible [things], indeed, a [per‐
son] who [can]not perceive [the said things] cannot [validly] negate
certain the meaning of the Vedic sounds – be they sentences or words – inde‐
pendently of a [purely] human instruction, [Dharmakīrti] says [the following].”
18 Note PVSVṬ 583,16–17: kiṃ kāraṇam / vipakṣavṛtteḥ sandehena sarvasya śeṣa‐
vattvāt /. “Why [are these reasons inconclusive]? – Because, since one can sus‐
pect [their] occurrence in the counter‐instances [, viz., in human beings who are
superior], all [of them] possess a remainder.” On the śeṣavadanumāna, see PV
1.331/PVSV 173,22–26 below, pp. 62–65, pp. 62–63 nn. 126–127, and pp. 83–
118.
[them, and this for three reasons: first,] because [this person of lim‐
ited cognition can]not know them even though they exist; [second,]
because, for the very reason [that they are supersensible, both kinds
of] incompatibility [between them are] not established; and [third,]
because it is not incompatible that [something co]exists with [some‐
thing] that is not incompatible [with it]. This too [has already been]
said [above]. [But] the sameness [of the alleged superior person] with
another is not established either [, and this again for three reasons:
first,] because one cannot know that there is no difference [between
the two; second,] because non‐perception has been dismissed as be‐
ing a [valid logical] reason in cases such [as the qualities of another’s
mind]; and [third,] because one sees a certain difference [in wisdom
or intelligence between two persons] in spite of the fact that [their]
humanity, etc., is the same, and that if a difference exists, [then] same‐
ness is not established. [But all this has already been] said [earlier].
Therefore, this inference possessing a remainder is incapable [of
proving that a superior person does not exist], because even though
one does not notice the presence [of the logical reason] in the dissim‐
ilar instances, one [can] suspect the contrary.19
In PV 1.312–329/PVSV 164,24–173,13, Dharmakīrti turns to the ab‐

surd consequences which a Mīmāṃsaka exposes himself to as long
as he rejects atīndriyadarśana. As Dharmakīrti strongly insists upon
in an earlier passage,20 by denying the Veda any human agency and
19 PVSV 164,9–24 (together with PV 1.310cd–311): aśaktisādhanaṃ puṃsām ane‐
naiva nirākṛtam // pratipāditā hi puruṣakṛtās tadadhiṣṭhānāc ca phaladā man‐

trāḥ / tad asti kaścid atiśayavān iti tatpratikṣepasādhanāny api prativyūḍhāni /
buddhīndriyoktipuṃstvādi sādhanaṃ yat tu varṇyate / pramāṇābhaṃ yathārthā‐
sti na hi śeṣavato gatiḥ // yat tu buddhīndriyavacanayogāt puṃstvād iti puruṣāti‐
śayapratikṣepasādhanaṃ tat tv agamakam eva / pratikṣepasāmānyasādhanayor
asambhavāt / na hy atīndriyeṣv ataddarśinaḥ pratikṣepaḥ sambhavati / satām
apy eṣām ajñānāt / ata eva virodhāsiddheḥ / avirodhinā ca saha sambhavāvi‐
rodhād ity apy uktam / nāpītarasāmānyasiddhir viśeṣāsambhavasya jñātum aśak‐
yatvāt / īdṛśeṣu cānupalabdher hetutvapratikṣepāt / puṃstvādisāmye ’pi kasya‐
cid viśeṣasya darśanāt / sambhavadviśeṣe ca sāmyāsiddh[e]ra ity uktam / tasmāc
cheṣavad anumānam etad vyatirekasya sandehād asamartham adarśane ’pi vipa‐
kṣavṛtteḥ /. a PVSVṬ 584,17: sāmānyāsiddher; PVSVTib D358a7/P525a7 = PVṬ Ñe
D45a6/P51b3: spyi ma grub pa’i phyir ro, against PVSV 164,22–23: sāmyāsid‐
dhir. On this passage, see Eltschinger 2001: 101–114 and 2008: 281–286.
20 See especially PVSV 112,16–27 and Eltschinger 2007a: 240–243.
hence intentionality, the Mīmāṃsaka deprives it of any meaning, for

meaningfulness depends on conventions (saṅketa, samaya) that are
nothing but shared semantic intentions (vivakṣā, vaktur icchā/abhi‐
prāyaḥ).21 An authorless scripture could only be unintelligible and
devoid of truth value. But there is yet another reason why the
meaning of the Veda, granted it exists, cannot be arrived at by the
Mīmāṃsaka unless he contradicts his own claim that humans, qua
humans, cannot perceive supersensible things. Since Vedic words
have an invisible relation to invisible things, no one can pretend to
ascertain what they really refer to. In other words, nobody can rule
out the possibility that the paradigmatic Vedic injunction: agniho‐
traṃ juhuyāt svargakāmaḥ, may actually mean khādec chvamāṃsaṃ
svargakāmaḥ.22 As it turns out, every interpretation of the Veda re‐
flects nothing but the interpretor’s own arbitrary conceptions. The
Veda is, then, a mutus liber, sealed by its original, unfabricated na‐
ture as well as by the cognitive limitations of humans. As a conse‐
quence, if the Mīmāṃsaka pretends, as he does, to provide an
authoritative exegesis of the Veda,23 he has either to attribute hu‐
man agency and intentionality to the Veda (together with an unbro‐
ken explanatory tradition), or he has to admit that its most re‐
spected interpreters, Jaimini and Śabarasvāmin,24 were endowed
with extraordinary perceptual abilities.

1.6. More precisely, insofar as Vedic words refer to supersensible
realities, their relation to these things is supersensible too. As Dhar‐
makīrti insists, their being limited ([prati]niyata) to a specific thing
21 See PV 1.327/PVSV 172,15–24 below, pp. 58–59; see also Eltschinger 2007a:
134–143.
22 See PV 1.318/PVSV 167,11–14 below, pp. 40–41 and n. 35; see also below, pp.
126–127 and 135–140.

23 For a detailed account of Mīmāṃsa’s exegetical methods, see below, pp. 119–
149.
24 In PVṬ Ñe D49a4/P56b2 ≈ PVSVṬ 589,16, Mīmāṃsakas such as Jaimini are pre‐
sented as vedārtham ākhyātāraḥ. In PVṬ Ñe D47a6/P54a1–2, these are termed

*pramāṇabhūtapuruṣas (skyes bu tshad mar gyur pa). Śabarasvāmin is mentio‐
ned at PVṬ Ñe D49b5/P57a6 = PVSVṬ 589,31–590,12.
is beyond the reach of ordinary experience.25 In other words, the

Mīmāṃsaka exegete may well believe that Vedic words refer to the
things he expects them to point to, such as the daily morning and
evening oblations he is acquainted with in the case of “agnihotra;”
he may well believe that they have the desired meaning (abhimatār‐
tha) or conform to his own wishes (svecchānurūpa).26 Yet this re‐
mains just wishful thinking. As an ordinary (prākṛta),27 hence cogni‐
tively limited (arvāgdarśin, arvāgdarśana as a bahuvrīhi compound)
person,28 the Mīmāṃsaka cannot discern (vivecana, viveka) what
these words really refer to29 and is simply ignorant of the true na‐
ture (tattva) of their meaning.30 And as Dharmakīrti repeatedly
points out, the Mīmāṃsaka can resort neither to ordinary linguistic
usage (prasiddhi, lokavāda), nor reason(ing) (yukti, nyāya),31 nor
tradition (sampradāya).32 The only thing he can do, and which Dhar‐
makīrti actually accuses him of doing, is to ascribe extraordinary
cognitive abilities to his authorities. And such is Dharmakīrti’s in‐
troduction to the passage translated below: “Moreover, in holding
such a view, the Jaiminīyas compromise their own position with
25 See PVṬ Ñe D46a7/P52b8 = PVSVṬ 585,27–28 and PVṬ Ñe D50a1–2/P57b3–4
≈ PVSVṬ 590,16–17.
26 See PVṬ Ñe D46a4/P52b3–4 ≈ PVSVṬ 585,21.
27 See
PVṬ Ñe D46b4/P53a5 = PVSVṬ 586,10, PVṬ Ñe D63b7/P76b2 = PVSVṬ
610,12, PVṬ Ñe D56a3/P66a2–3 = PVSVṬ 598,21 (D with no equivalent of prā‐
kṛta).
28 See PVSVṬ 598,15, PVṬ Ñe D61a1/P72b2 = PVSVṬ 605,24–25, PVSVṬ 585,28–
30 (to be compared with PVṬ Ñe D46b1/P52b8–53a1).

29 See PVṬ Ñe D47a2–3/P53b4–5 ≈ PVSVṬ 586,21–22, PVṬ Ñe D52b2–3/P61a5–6
= PVSVṬ 593,27, PVṬ Ñe D46b7–47a1/P53b2–3 ≈ PVSVṬ 586,18–19.

30 See PVṬ Ñe D45b5/P52a3–4 = PVSVṬ 585,8.
31 PVṬ Ñe D51a6/P59a8–b1 = PVSVṬ 592,7, PVṬ Ñe D51a6–b1/P59b1–4 ≈ PVSVṬ
592,7–13. These passages contain suggestive claims, on the part of the Bud‐
dhists, that they alone are rational. See below, p. 42 n. 40.
32 See, e.g., Dharmakīrti’s conclusion in PVSV 168,3–4, below, p. 44. Note also PVSV
169,4–13, below, pp. 47–48, where Dharmakīrti demonstrates that the Vedic
schools or recensions (śākhā), in being notoriously liable to decay, charismatic
revival, and innovation, are far from handing down unaltered scriptures.
their own assertion.”33 The Mīmāṃsakas gain access to the invisible

meaning of the Veda due to persons (like Jaimini) who must ipso
facto be somehow superior to ordinary people. Therefore, they de
facto recognize epistemically superior beings able to discern the
meaning of Veda,34 for this cannot be achieved without such per‐
sons.35 But this conflicts with the Mīmāṃsakas’ ever‐repeated doc‐
trine (their very “complexion,” Dharmakīrti says) according to
which there can be no person able to perceive supersensible
things.36 Furthermore, if the Mīmāṃsaka acknowledges that his own
authorities somehow access the supersensible realm, all his argu‐
ments against the existence of atīndriyadarśipuruṣas such as those
recognized by the Buddhists (bauddhādyabhimata) fall to the
ground.37 As a consequence, he will be forced to admit that either all
or a few human beings, and certainly not only Mīmāṃsakas, are en‐
dowed with extraordinary abilities.38
33 PVSV 164,24–25: api caivaṃvādino jaiminīyāḥ svam eva vādaṃ svavācā vidhu‐
rayanti /.
34 According to PVṬ Ñe D45b1–2/P51b6–7 = PVSVṬ 584,22–23: svam eva vādam
iti kathaṃcid atiśayavato jaiminyādeḥ sakāśād vedārthagatir bhavatīti puruṣāti‐

śayābhyupagamavādam.
35 According to PVṬ Ñe D45b2–3/P51b8–52a1 = PVSVṬ 584,24–25: vidhurayanti
bādhante ’tiśayavatpuruṣapratikṣepeṇa vedārthagater asambhavāt /.

36 According to PVṬ Ñe D45b1/P51b6 = PVSVṬ 584,21: evaṃvādina iti nāsty atī‐
ndriyārthadarśī puruṣa ity evaṃvādino jaiminīyāḥ, and PVṬ Ñe D45b2/P51b7–8

= PVSVṬ 584,23–24: nāsty atīndriyārthajñaḥ puruṣaḥ kaścid ity anayā svavācā.
37 Most regularly mentioned among these puruṣātiśayapratikṣepasādhanas or a‐
tīndriyārthadarśipuruṣapratikṣepasādhanas is puruṣatva or puṃstva. See PVṬ Ñe

D46b6–7/P53b1–2 ≈ PVSVṬ 586,17–18, PVṬ Ñe D47a2/P53b4 ≈ PVSVṬ 586,22–
23, PVṬ Ñe D47a1–2/P53b3 = PVSVṬ 586,19–20, PVṬ Ñe D45b6–7/P52a5–6
≈ PVSVṬ 585,11–13. Note PVṬ Ñe D45b6–7/P52a5–6 ≈ PVSVṬ 585,11–13: atha
puruṣatvādisāmye ’py asādhāraṇaśaktiyukto vaidikānāṃ śabdānām atīndriyair
arthaiḥ saha sambandhasya vettā kaścij jaiminyādiḥ kalpyate / tadā tadvaj jaimi‐
nyādivat. “But if one postulates that a certain [person] such as Jaimini, possessed
with a special ability in spite of his being [allegedly] the same [as any other per‐
son] regarding humanity, knows the relation of the Vedic words with [their] su‐
persensible meanings, then, just as Jaimini, etc.” See above, pp. 13–15.
38 See
PVṬ Ñe D46a1/P52a8 ≈ PVSVṬ 585,14–15, PVṬ Ñe D46b5/P53a6–7
= PVSVṬ 586,13–14, PVṬ Ñe D47a3/P53b5–6 ≈ PVSVṬ 586,23–24, PVṬ Ñe
D45b6–7/P52a5–6 ≈ PVSVṬ 585,11–13.

1.7. Strictly speaking, Dharmakīrti’s critique of authorlessness
comes to a close with PV 1.329/PVSV 173,13. But before returning
to the questions that concerned him prior to his excursus on scrip‐
tural authority, namely, the relation between (scriptural) words and
their meanings (PV 1.336–338/PVSV 175,10–176,4) and non‐per‐
ception (PV 1.339/PVSV 176,5–12), and then concluding with anti‐
Hindu sarcasms (PV 1.340), Dharmakīrti attacks yet another at‐
tempt to legitimize scriptural authority (PV 1.330–335/PVSV
173,14–175,10), viz., ekadeśāvisaṃvādana. According to his oppo‐
nent, the Veda as a whole, including those of its loci that bear on su‐
persensible (hence inverifiable/infalsifiable) matters, can be proved
to be true provided one of its parts (ekadeśa) bearing on empirical
(hence verifiable/falsifiable) matters can be proved to be veracious.
This polemic raises two distinct problems. The first one concerns
the identity of Dharmakīrti’s opponent. For, as it is obvious, the doc‐
trine he criticizes comes quite close to that of the Nyāya.39 Accord‐
ing to his commentators, however, Dharmakīrti’s opponent is
(an)other Mīmāṃsaka(s) (dpyod pa pa gźan dag, Śākyabuddhi), and
more specifically an “old Mīmāṃsaka” (vṛddhamīmāṃsaka, Karṇa‐
kagomin, Manorathanandin).40 Taking this seriously, who can this
39 See NSū 2.1.68 and NBh 96,11–97,16. Note especially NBh 97,8–9: dṛṣṭārthenā‐
ptopadeśenāyurvedenādṛṣṭārtho vedabhāgo ’numātavyaḥ pramāṇam iti, āptaprā‐

māṇyasya hetoḥ samānatvād iti /. “One can infer that one Vedic passage whose
object is invisible is a pramāṇa from the [fact that the] Āyurveda, the teaching of
an āpta the object of which is visible [, is a pramāṇa; it is so] because [their]
cause, the [personal] authority of the āpta, is the same.” And further, NBh 97,15–
16: draṣṭṛpravaktṛsāmānyāc cānumānam / ya evāptā vedārthānāṃ draṣṭāraḥ
pravaktāraś ca ta evāyurvedprabhṛtīnām apīty āyurvedaprāmāṇyavad vedaprā‐
māṇyam anumātavyam iti /. “And [such an] inference is [made possible] by the
[fact that these scriptures’] seers and expounders are the same. Since those very
āptas who saw and expounded the objects of the Veda are exactly those of the
Āyurveda, etc., one can infer that, as the Āyurveda is authoritative, the Veda is au‐
thoritative.” Note, however, that Dharmakīrti’s portrayal of his adversary does
not allude to the underlying principle of draṣṭṛsāmānya. See Oberhammer 1974
and Steinkellner 1979: 78 n. 272.
40 PVṬ
Ñe D62b2/P74b4–5, PVṬ Ñe D62b2/P74b5 = PVSVṬ 608,17: anyas tu
mīmāṃsakaḥ; PVV 409,13: aparo mīmāṃsakaḥ; PVṬ Ñe D62b7/P75a5, PVṬ Ñe
D66b7–67a5/P80b2–81a1: dpyod pa pa. PVSVṬ 608,15–16: aparo vṛddhamī‐
māṃsakaḥ, PVV 409,8: vṛddhamīmāṃsakānām, PVV 411,4: jaranmīmāṃsakaḥ.
rather unorthodox Mīmāṃsaka be? Should we think of Bhavadāsa,

who is criticized by Kumārila and is likely to be one among the vṛd‐
dhamīmāṃsakas, i.e., the “pre‐Śabarasvāmin” Mīmāṃsakas alluded
to by Jayantabhaṭṭa?41 Whoever he may be, what we would have to
do with is a Mīmāṃsaka who does not (yet?) acknowledge the au‐
thorlessness of the Veda as a criterion of its reliability, or at least
one who puts forward addititional criteria for its authority.42 How‐
ever, since Dharmakīrti does not provide any hints as to the identity
of his opponent,43 one may feel justified in hypothesizing that Dhar‐
makīrti also targets the Nyāya.44 Though I am strongly inclined to
favor the hypothesis of a Mīmāṃsaka opponent, I would like to
postpone any conclusion until further research has been carried out
Vibhūticandra seems to regard this designation as sarcastic (Vibh 409 n. 8): vṛd‐
dhaś cakṣurdoṣopahatatvāt. Against this interpretation, see below, n. 41. See also
PVin 2 72,10–11 (Steinkellner 1979: 77–78 and n. 252), TSK 2775/TSŚ 2774 and
TSPK 736,3–18/TSPŚ 892,13–893,10.
41 See Bronkhorst 1994: 383–385. NM I.664,6–7 is all the more interesting that it
contrasts the vṛddhamīmāṃsakas (jarajjaiminīyas in NM I.664,16) and the fol‐

lowers of Śabara (śābara), thus allowing us to understand vṛddhamīmāṃsaka as
“pre‐Śabarasvāmin Mīmāṃsaka”: vṛddhamīmāṃsakā yāgādikarmanirvartyam
apūrvaṃ nāma dharmam abhivadanti yāgādikarmaiva śābarā bruvate /. “The
old Mīmāṃsakas declare that dharma is what is called apūrva, which is to be
brought about by ritual activities such as sacrifices. The followers of Śabara say
that the ritual activities such as sacrifices themselves are [dharma].” Translation
Bronkhorst 1994: 384. On vṛddha‐ or jaranmīmāṃsakas, see also Kunjunni Raja
1963: 199 and Eltschinger 2007a: 161 n. 7.
42 Though it is more likely to refer to (a) Buddhist(s) than to (a) heterodox Mīmā‐
ṃsaka(s), ŚV codanā 121 testifies to Kumārila’s familiarity with such a strategy:

yo ’pīndriyārthasambandhaviṣaye satyavāditām / dṛṣṭvā tadvacanatvena śrad‐
dheye ’rthe ’pi kalpayet //. “The one who has observed [a certain person’s] truth‐
fulness regarding a matter related to the sense objects might postulate [that it
is] also [veracious] regarding something to be trusted, since it is a statement of
this [same person].” For Sucaritamiśra’s and Pārthasārathimiśra’s explanations,
see Eltschinger 2007a: 99 n. 98, and below, p. 115. For an alternative hypothesis
regarding the vṛddhamīmāṃsaka, see Krasser 2012: 567–568 with n. 79.
43 Unless one understands parityajya very literally (“giving up, abandoning, disre‐
garding”) in the following statement (PVSV 173,16–17): anyas tv apauruṣeyam

āgamalakṣaṇaṃ parityajyānyathā prāmāṇyaṃ vedasya sādhayitukāmaḥ prāha.
44 As PVSVṬ 610,12 and PVṬ Ñe D64a2/P76b5–6 ≈ PVSVṬ 610,15–16 (mīmāṃsa‐
kādiḥ) might testify to. See also Steinkellner 1979: 78 n. 252.

on this passage. The second problem is of a philosophical nature.

Dharmakīrti criticizes his opponent’s apologetic strategy (i.e.,
ekadeśāvisaṃvādana) by pointing out that to infer the authority of
the Veda in this way amounts to a śeṣavat‐inference, an inference
“with a remainder.”45 What does this pseudo‐inference consist of?
To put it briefly, this kind of inconclusive argument characteristi‐
cally doesn’t rule out the possibility of a counterexample. In other
words, that one empirically verifiable Vedic locus is true in fact does
not exclude (an)other Vedic statement(s) being false. But Dharma‐
kīrti is well aware of the fact that his own apologetic strategy (ulti‐
mately, the one he ascribes to Dignāga), which amounts to nothing
but a sophisticated version of ekadeśāvisaṃvādana, is open to the
same charge. But is it really so? According to Dharmakīrti, one is en‐
titled to infer a certain scripture’s reliability (avisaṃvāda) in regard
to non‐empirical (adṛṣṭa, atīndriya, atyakṣa, etc.) things if all that
this scripture claims to be perceptible is indeed perceptible (con‐
trary to most of the Vaiśeṣika categories and to the Sāṅkhya
constituents), all that it holds to be “objectively” inferable is indeed
(successfully) inferable (contrary to īśvara and the ātman), and if
one cannot detect any inconsistency or internal contradiction
(pūrvāparavirodha) in it (contrary to the prescription of physical
ablutions in order to annihilate moral defilements).46 Now, as Dhar‐
makīrti insists, the difference between his own and his opponent’s
strategies lies in the scope of the verification process. Whereas his
opponent restricts it to one (trivial, atyantaprasiddha) point, Dhar‐
makīrti requires that it should bear on all the empirically verifia‐
ble/falsifiable matters alluded to in the treatise under scrutiny.47
What Dharmakīrti doesn’t say in this context, however, is that the
verification strategy according to which he proves the Buddhist
scriptures to be reliable is an alternative one. As he announces in PV
1.217/PVSV 109,11–19 and makes clear in PV 2.145–279, the relia‐
bility of the Buddhist scriptures can be inferred from the fact that
their principal tenet (pradhānārtha), the four Noble(’s) Truths,
45 See above, pp. 13–15 and n. 18, and below, pp. 62–63 nn. 126–127.
46 See below, pp. 86–87 and 107–111; see also below, p. 64 n. 130, pp. 69–70 n.
142, p. 77 n. 172 and p. 78 n. 174.

47 See PVSV 173,26–174,6 below, pp. 63–65.
withstands critical analysis. In other words, the first strategy (PV

1.215–216/PVSV 108,16–109,11) is of a “heresiological” character,
whereas the second (PV 1.217/PVSV 109,11–19) is apologetic in na‐
ture. And it is of the first strategy as applied to the Veda that PV
1.332–334 and PVSV 174,14–175,4 provide a fascinating example.

JOHN TABER
A Synopsis of PV(SV) 1.312–340

1. The validity of the Veda cannot be based on “authorlessness”

(apauruṢeyatva).
1.1. If the Veda were authorless, we could never know what it means.
1.1.1. The Veda itself does not establish its meaning; nor, according
to Mīmāṃsā principles, can some authoritative expositor do so.
Dharmakīrti begins his final assault on the Mīmāṃsā teaching of
the authorlessness of the Veda with the statement, “Moreover, in
holding such a view” – namely, that humans are unable to cognize
supersensible things, which is the basis for the Mīmāṃsaka’s claim‐
ing that the mantras of the Veda, in particular, could not possibly
have been composed by humans – “the Jaiminīyas compromise
their own theory with their own assertion.”
If humans cannot cognize the supersensible, as Mīmāṃsā main‐
tains, then how can one ascertain the meaning of the Veda? The
Veda does not itself declare what it means. Humans must surmise
what it means. (312) If you think some esteemed person – Jaimini
himself, for instance – is able to ascertain its meaning, then you
must accept that humans are capable of cognizing supersensible
things. But why should only Jaimini be capable of this? In short, if
you deny that humans are capable of cognizing the supersensible
because they are afflicted by defects like desire, then you must
deny that Jaimini is able to do so – in which case there is no way the
meaning of the Veda can be known. If, on the other hand, you ac‐
cept that Jaimini, at least, can know such things, then other humans
ought to be able to know them as well. In that case, the question
becomes why we should place our confidence in one person as
knowing the meaning of the Veda rather than another. (313)
The Mīmāṃsaka at this point proposes that one should accept as
an authority that person whose interpretations are supported by
other pramāṇas. But, Dharmakīrti responds, it is hard to see how
other pramāṇas could support his interpretations, since percep‐
24 JOHN TABER
tion, etc., do not apprehend the supersensible. And if it were acces‐

sible to perception, one wouldn’t need scripture to cognize it. If the
Mīmāṃsaka were to hold that one cannot cognize the supersensible
solely through other pramāṇas – the other pramāṇas are only effec‐
tive, even in regard to their own objects, in dependence on scrip‐
ture – that would entail the absurd consequence that one couldn’t
even infer fire from smoke unless it were supported by scripture. If,
finally, the Mīmāṃsaka were to say that we have to rely on the
other pramāṇas in determining when scripture applies, then one
could once again ask why we need to resort to scripture at all when
it comes to supersensible matters. (314)
If, moreover, the Mīmāṃsaka insists that only a Vedic statement
explained by someone who is a reliable expositor, insofar as other
statements of his are confirmed by other pramāṇas, is authoritative
– or else, as the Vṛtti puts it, a Vedic statement that is consistent
with other pramāṇas – then that should be the definition of scrip‐
ture – “consistency with [other] pramāṇas (pramāṇasaṃvāda)” –
not “authorlessness” (apauruṣeyatva). (315)
Dharmakīrti returns to the point that if Jaimini is able to ascer‐
tain the supersensible meaning of the Veda, then other humans,
even though afflicted by various defects, ought to as well. The Mī‐
māṃsaka’s acceptance of Jaimini as an authority means he believes
humans are capable of cognizing supersensible things. Yet that is
precisely what he denies! (316)
Or, again, if he insists that people, afflicted with moral defects,
cannot cognize these things, then how is the meaning of the Veda
known? The supposedly authoritative person who expounds it –
this Jaimini – is incapable of knowing it on his own. Nor could he
have learned it from someone else; for the same problem would
arise for that other person. And of course the Veda does not declare
its meaning itself. (317)
Thus, we really have no idea what statements like agnihotraṃ
juhuyāt svargakāmaḥ, which is usually understood to mean “One
who desires heaven should perform the Agnihotra,” really mean.
For all we know, it could mean, “One should eat dog meat”! This is
not the case for all traditions of practice. In the case of a human
tradition, founded by an original Preceptor, such as the Buddhist
tradition, the meanings of the statements one follows, i.e., the Pre‐
PVSV 164,24–176,16 – SYNOPSIS 25
ceptor’s intentions, are transmitted along with them. And in the

case of more rational traditions like Buddhism, Dharmakīrti main‐
tains, the learned tend to following reason (nyāya) when determin‐
ing what is to be done and avoided; they do not just rely on authori‐
tative teachers. Here the Mīmāṃsaka is allowed to object: But don’t
you Buddhists also accept certain teachings from supposedly relia‐
ble persons, such as those concerning the arrangement of the world
(i.e., cosmology)? Dharmakīrti responds, such statements are not to
be considered pramāṇas; just because their authors have been
shown reliable in some matters we are able to confirm does not
mean they are reliable about everything they talk about. For practi‐
cal purposes, when it comes to supersensible matters (in particular,
matters of merit and demerit), one follows scripture because one
has no other recourse. But, strictly speaking, scripture is not a
pramāṇa. (318)
1.1.2. Common usage (prasiddhi) is not a means of establishing the

meaning of the Veda, either.
Now the Mīmāṃsaka proposes: ordinary parlance (lokavāda)
determines the meaning of the Veda. This may be an allusion to the
principle declared in MīSū 1.3.30 that the words of the Veda and
those of ordinary language have the same meanings; in any case,
Mīmāṃsakas routinely cite prasiddhi, common usage, as a reason
for interpreting Vedic passages in certain ways.1 Yet, Dharmakīrti
points out, even if we construe Vedic words as ordinary ones, they
will still in many instances be polysemic – as indeed ordinary
words often are. Who will be able to determine which of many pos‐
sible meanings of a word in the Veda is the right one? The “ordi‐
nary” meaning, moreover, is established by convention, which is
accessible to instruction, while the Veda is supposedly eternal. How
could someone, in the case of words which supposedly have a be‐
ginningless, authorless relation with supersensible meanings, tell
us what they mean? If one held that there is an explanation of the
meaning of the Vedic word that is beginningless and authorless as
well – that is to say, an exegetical tradition that extends forever
back into the past – then the problem is just moved back a step.
1 See below, pp. 135–140.

26 JOHN TABER
How is the meaning of this beginningless explanation known?

Moreover, how do we know the explanation has never been corrup‐
ted? We are aware of various factors that introduce errors into tra‐
ditions – enmity, pride, and so forth. And why would the Mīmāṃ‐
saka, of all people, put confidence in a supposedly unbroken tradi‐
tion of explanation, since he is the one who emphasizes that hu‐
mans afflicted with moral faults are not to be trusted?2 That, in fact,
is his most characteristic point, says Dharmakīrti. In his Svavṛtti to
3.319, Dharmakīrti goes further into how we hear of Vedic schools
recovering after nearly dying out – even today some have only a
few reciters – so that one might suspect that even those schools
that have many adherents today could have been nearly extinct at
one time but were restored, and that in the process of restoration
errors could have crept into the recitation of the Veda in various
ways. And the same could be the case for any “beginningless” tradi‐
tion of Vedic interpretation. In summary, one cannot establish the
meaning of the Veda either through an authorless explanation or
ordinary linguistic practice. Even if the relation of word and
meaning were eternal, ordinary parlance teaches us that words are
polysemic, so that, in the case of Vedic words, only someone who
knows supersensible things could know which meanings are the
correct ones. And also according to ordinary parlance, the mean‐
ings of words are conventional; in themselves they are indifferent
as to what they mean, it is only convention that assigns them cer‐
tain meanings. Thus, a person capable of knowing supersensible
things would again be required in order to tell us the supersensible
things to which Vedic words are assigned. (319)
But even the Mīmāṃsaka, continues Dharmakīrti, does not re‐
ally follow common usage when it comes to explaining the meaning
of Vedic words. He offers as examples the words svarga and urvaśī.
svarga commonly means “heaven,” but the Mīmāṃsaka construes it
as “delight.”3 urvaśī, meanwhile, is usually the name of a nymph
who resides in heaven, but in general the Mīmāṃsaka does not
interpret proper names in the Veda as referring to individuals –
which would impugn the eternality of the Veda – but offers etymo‐
2 Thus, Kumārila’s famous statement in ŚV codanā 144ab. See below, p. 125 n. 15.
3 See ŚBh V.72,6–7 ad MīSū 4.3.15 and below, pp. 139–140 n. 78.
PVSV 164,24–176,16 – SYNOPSIS 27
logical analyses of them instead.4 At the same time, ritualists re‐

ferred to the two kindling sticks that are rubbed together to start
the fire in the Soma sacrifice as Urvaśī (the lower) and Purūravas
(the upper one), who according to a widespread myth was her con‐
sort.5 The Mīmāṃsaka cannot claim that one must sometimes re‐
sort to an uncommon meaning because the common one is blocked,
i.e., it does not fit the context, for how could one ever know in the
case of a Vedic statement, which refers to a supersensible state of
affairs, that the common meaning is blocked? (320ab) And if we ac‐
cept uncommon meanings in the case of such words as svarga and
urvaśī, how do we know we shouldn’t do so for agnihotraṃ juhuyāt
svargakāmaḥ (i.e., “One should eat dog meat”)? The meaning of this
sentence cannot be resolved by other Vedic passages, because the
meaning of those other passages are in doubt as well – for all we
know, the occurrence of agnihotraṃ juhuyāt svargakāmaḥ there
could also mean “One should eat dog meat.” (320cd)
The Mīmāṃsaka has made it clear that he puts little faith in what
humans say. Yet now he takes ordinary parlance as the criterion of
the meaning of the Veda. He is being egregiously inconsistent.
(321) Indeed, by calling common usage into question he disquali‐
fies it as a pramāṇa that could in any way be applied in interpreting
the Veda, i.e., when the common meaning is not blocked. The un‐
derstanding of the Veda by means of common usage is “a haphaz‐
ard grasping” which follows no rule at all. (322) Finally, it is com‐
mon usage itself – which we know to be ambiguous so much of the
time – which has raised doubts about the meaning of the Veda in
the first place. (323) Therefore, if one were to resort to common
usage, it is difficult to see how one could ever be certain of the
meaning of the Veda. (324)
1.2. An authorless Veda could not have an inherent meaning.
Thus, a teacher like Jaimini, who presumes to tell us what the Veda
means without really knowing it himself, is simply foisting upon his
followers “his own conception” in the guise of the Veda, and is really
no different from any other religious teacher (insofar as he is simply
4 See below, p. 126 n. 17.
5 See below, pp. 49–50 n. 78.

28 JOHN TABER
inventing doctrines about what one must do in order to attain salva‐

tion, etc.). He is like the man who, when asked the way to Pāṭalipu‐
tra, says, “The post says it is this way,” instead of correctly saying
(upon reading the post), “I say it is this way” – for the post cannot say
anything! (325)
The idea that the Veda of itself is restricted to a certain meaning,
a meaning which Jaimini is supposedly able to divine, is also cer‐
tainly mistaken. Words in general are not restricted to meanings by
nature; rather, they are suited to mean anything, for we use them
arbitrarily to refer to whatever we want. (326ab) And if a word were
by nature restricted to only one meaning, how could one ever know
it in the case of a Vedic word, whose meaning is something super‐
sensible? (326cd) In fact, however, a word is restricted to a certain
meaning only by the intention of the one who employs it; the con‐
vention (saṅketa), by which the meaning of a word is learned, is the
statement of this intention. If a word were authorless, as is suppos‐
edly the case for a Vedic word, then it would not be restricted to any
particular meaning – it would be in effect meaningless; moreover,
there being no convention expressing an author’s intention to con‐
vey its meaning, a particular meaning for it could never be learned.
(327) If, to the contrary, a Vedic word were naturally restricted to a
certain meaning, then it could not be used arbitrarily to mean what‐
ever one wants, nor would one even require the statement of a con‐
vention to cognize what it means; for if its meaning were something
sensible, its restriction to that would be evident to anyone. (If, on the
other hand, it were something supersensible, no one would ever be
able to cognize it.) Nor would a convention, which arbitrarily assigns
a meaning according to the speaker’s intention, necessarily reveal
just the natural meaning of a word. (328) When one has to do with
conventions, one can no longer really speak of the restriction of a
word to a certain meaning, natural or not. (329)
2. Nor can the validity of the Veda be based on an inference from the
confirmation of a few of its statements.
Because of the futility of attempting to ground the authority of the
Veda on its authorlessness, then – which Dharmakīrti has been con‐
sidering since PVSV 1.224 – there are those who might resort to es‐
tablishing its validity on the basis of an inference from the truth of
PVSV 164,24–176,16 – SYNOPSIS 29
some of its statements which we are able to confirm. Just as those

statements are true, so are all of its statements – even those which the
Buddhists find most questionable – “because they are part of the same
[Veda].” (330)
Dharmakīrti rejects this inference as a śeṣavat inference, a falla‐
cious type of reasoning which he discusses in the opening section of
PV(SV) 1, which treats the three types of valid reason (PV[SV] 1.11). It
would be like concluding, from the fact that the few grains one has
sampled are cooked, that all of the grains in a pot of rice are cooked.
And although Dharmakīrti himself, following Dignāga, has proposed a
similar account of scripture at PV(SV) 1.216, namely that the validity
of its statements is inferred from the fact that they are made by a
trustworthy person (āpta) whose other statements have been shown
to be reliable, he stresses that such an inference can be made only if
all of the statements of scripture that can be confirmed by other pra‐
māṇas are correct; and even then, such an inference is made only out
of necessity, because there is no other means of knowing what per‐
tains to the achievement of a higher state of life and, eventually, salva‐
tion. (331) But to proclaim that the whole Veda is true because it con‐
tains a true statement here and there, while it makes so many other
assertions that are patently false – that there is a soul, that some
things are eternal, and so forth; in his Vṛtti Dharmakīrti meticulously
characterizes a variety of fallacious Vedic teachings – such a person’s
audacity exceeds that of the woman who brazenly denies committing
adultery even when caught in the very act! (332–334) Indeed, if one
considered a person an authority just because one of his statements
turned out to be true, then there would be no one we would not con‐
sider an authority. (335)
3. Conclusion.
Returning finally to the question of non‐perception, specifically, whe‐
ther the silence of scripture, along with that of the other pramāṇas
perception and inference, can establish the non‐existence of some‐
thing (PV[SV] 1.199), which precipitated his discussion of āgama,
Dharmakīrti reiterates the point made at 1.213 that there is no in‐
variable concomitance between sounds and the things signified by
them – speech is not of the nature of the things signified by it, nor
their effect. Thus, we can never be sure from an utterance of the exis‐
30 JOHN TABER
tence of that which it is supposed to signify. (336) Should one object

that words are in fact the indirect effects of what they signify insofar
as they are uttered upon perceiving them, then the response would be
that if that were the case, we would not find words having opposite
meanings in different teachings, e.g., the word “sound,” which refers
to something permanent in the writings of one school, but to some‐
thing impermanent in the writings of others. (337) Thus, there must
always be doubt about scripture, whose relation to what it means is
never fixed. Those who rely on scripture in the conduct of their lives,
putting into practice its teachings, cannot be said to be acting on the
basis of true knowledge of anything. (338) Scripture cannot even reli‐
ably inform us as to what does not exist, i.e., insofar as it is silent
about it. For even if everything falls within the scope of scripture, one
cannot be certain that the reason scripture doesn’t mention some‐
thing is because it doesn’t exist. (Here, Dharmakīrti’s argument is not
explicit. It seems to be as follows: Given that scripture does not have a
fixed relation to any meaning, we can no more be certain, from the
fact that it does not mention something, that it does not exist, than we
can be certain, from the fact that it does mention something, that it
does.) (339ab) Thus, one cannot conclude, from the non‐cognition of a
thing by means of perception, inference, and scripture, that it does not
exist. (339cd)
In sum, anyone who believes in the Veda and follows the Vedic‐
sanctioned practices of pious Brahmins – ablutions to remove sin, and
so forth – is a pathetic creature completely devoid of discrimination.
(340)

Part 1 – Translation

Moreover, in holding such a view, the Jaiminīyas compromise their
own position with their own assertion,1 for
[Vedic] words do not [themselves] declare: “This is our mean‐
ing, not this.” This meaning [which Vedic words have] must be
postulated by humans.2 The latter are possessed, however, of
[moral defects] like desire.3 4Besides, how is the distinction
made that among [such humans] one knows what [the meaning
of the Veda] really is, the other doesn’t?5 Why isn’t there for
you in some way some [other person who is] knowledgeable [of
the meaning,] given that he is as human as the [other] [i.e., Jai‐
mini, etc.]? (PV 1.312–313)
These Vedic words surely do not cry out like this: “Come you revered
Brahmins, this is to be grasped as our meaning, not something else.”
[Such Vedic words] simply fall upon the ear, without an association
with a particular meaning being manifest. For these [words] one per‐
son postulates a certain meaning, another another.6 7But words have
no natural restriction by which they favor one meaning, not another.
They are simply seen as referring to this or that due to convention. It is
not right that one8 totally ignorant [person] afflicted by [moral] defects
can explain what [the meaning] of these [Vedic words] really is, their
1 See above, pp. 15–17 and nn. 33–36.
2 I.e., by the Mīmāṃsakas (PVṬ Ñe D45b4/P52a2 = PVSVṬ 584,27).
3 As a consequence, the meaning postulated (kalpita) by these human beings is not a
pramāṇa (PVṬ Ñe D45b4/P52a2–3 ≈ PVSVṬ 584,28; Vibh 404 n. 1).

4 The Mīmāṃsaka opponent now objects that only such persons as Jaimini know (vet‐
tṛ) the meaning of the Veda (PVṬ Ñe D45b4–5/P52a3 ≈ PVSVṬ 584,29).

5 Explanation, PVṬ Ñe D45b5–6/P52a4–5 = PVSVṬ 585,8–10: nātra kiṃcit kāraṇam
asti mīmāṃsakasya / puruṣatvāviśeṣāt / sarvo vā vetti na vā kaścid iti bhāvaḥ /. “The

Mīmāṃsaka lacks any criterion (kāraṇa) for this [distinction], because [these per‐
sons] do not differ as to [their] humanity: either all [of them would] know [the
meaning], or none [of them]. Such is the sense.”
6 As a consequence, there can be no ascertainment (nirṇaya) of the meaning of the
Veda (*vedārtha; PVṬ Ñe D46a4/P52b4 ≈ PVSVṬ 585,21–22).

7 The Mīmāṃsaka opponent now objects that words and their meanings have a natu‐
ral (svābhāvika) relation, so that Vedic words are restricted (pratiniyama) to one
meaning (ekārtha; PVṬ Ñe D46a4–5/P52b4–5 ≈ PVSVṬ 585,23–24).
8 I.e., Jaimini or Śabarasvāmin (PVṬ Ñe D46b2/P53a2 = PVSVṬ 585,30–31).
34 PVSV 164,24–176,16
restriction to a meaning being [in general] unknown because they refer

to [something] supersensible, [and] not another. But if this [person
and] no other knows [the meaning of the Vedic words] due to some su‐
periority, e.g.,9 of the intellect or the sensory faculties, where does his
superior knowledge of the supersensible come from? 10Likewise, why
is some other [person]11 who can perceive things beyond the reach [of
ordinary cognition] in [terms of] space, time, and nature considered
impossible? 12[Among] the “proofs” [brought forward] to discard these
[other superior beings],13 there is indeed none which does not apply to
this one [too, whom you accept]. 14Just as the [latter] is distinguished
[through his ability to know the supersensible] in spite of there being
these “proofs,” so should all others be [distinguished]. Therefore [in
this matter], non‐commitment [to a particular person as the authority
regarding the meaning of the Veda] alone is justified.
15If [you hold that] that one knows the meaning [of the Vedic
words] whose [explanatory] statements are consistent with
9 Besides buddhi and indriya, the commentators add, as an explanation of °ādi, re‐
peated practice (abhyāsa; PVṬ Ñe D46b3/P53a4 = PVSVṬ 586,9–10; PVV 404,9).

10 Introduction, PVṬ Ñe D46b4–5/P53a6 = PVSVṬ 586,13: bhavatu vā jaiminiprabhṛtiḥ
puruṣo ’tīndriyārthasya vettā /. “Or, let there be a human being, such as Jaimini, who
knows something supersensible.”
11 According to PVṬ Ñe D46b5/P53a6–7 ≈ PVSVṬ 586,13–14, “likewise” (tathā) should
be understood as, “like Jaimini, etc.” (jaiminyādivat), whereas “some other [person]”
(anyo ’pi) should be interpreted as “the superior person(s) recognized by [followers
of other traditions] such as the Buddhists (puruṣātiśayo bauddhādyabhimataḥ).”
12 Introduction, PVṬ Ñe D46b6/P53a8–b1 ≈ PVSVṬ 586,15–16: so ’py atīndriyārtha‐
darśy astv itīṣyatām / na ced abhimato ’pi jaiminyādir mā bhūt / yataḥ… “[You]
ought to admit that the one [accepted by another tradition] must also perceive the
supersensible. If not, even the [one] recognized [by you as superior, i.e.], Jaimini,
etc., cannot be [held to perceive the supersensible], because…”
13 See p. 17 n. 37.
14 The following is said on the hypothesis that the Mīmāṃsaka, going against his own
arguments (sādhana, so‐called proofs, such as puruṣatva), accepts that people such
as Jaimini perceive what the meaning of the Veda really is (vedārthatattva)PVṬ or are
different from, say, Buddhist authoritiesPVSVṬ (compare PVṬ Ñe D47a1–2/P53b3–4
and PVSVṬ 586,21).
15 The Mīmāṃsaka now turns from atīndriyadarśana to another criterion in order to
argue for Jaimini’s authoritative vedārthajñāna. Jaimini and others discriminate (vi‐
vecana) the meaning of the Veda only when they make statements (vacana) that are
TRANSLATION 35
valid cognition, [we would reply that this is not the case,] for
there can be no valid cognition of transcendent [things]. (PV
1.314)
Suppose the following [be urged]: We do not commit ourselves to
someone’s explanation [of the Veda] on the ground of his authority as a
person, but because [his explanation] is consistent with other valid
cognitions. Among all the numerous expositors [of the Veda,] we assent
to the one who makes [the meaning] coincide with valid cognitions
such as perception. [Answer:] This is not [the case], because the other
means of valid cognition do not function with regard to supersensible
[things] such as adṛṣṭa [i.e., merit and demerit]. It is indeed just be‐
cause of the unavailability of those [other means when it comes to su‐
persensible matters] that scripture is required in order to cognize
them; for otherwise,16 one would not cognize [such things] when other
means of valid cognition do not function, even though this [scripture] is
available. And since one would cognize [these] things through these
[other means of valid cognition] alone, scripture would not be a means
at all. Objection: One does not cognize supersensible [things] through
any of the other [means of valid cognition] alone.17 18[Answer:] How
consistent with valid cognition (pramāṇasaṃvādin; PVSVṬ 586,26, to be compared

with PVṬ Ñe D47a4/P53b6–7; PVṬ Ñe D47a4–5/P53b7–8 = PVSVṬ 586,28–29).
16 I.e., if, even with regard to something which is accessible to scripture (āgamaga‐
mya), it were only on the grounds of a statement’s consistency with other valid co‐
gnitions that one ascertained (niścaya) the meaning, then… (PVṬ Ñe D47b3–4/
P54a8–b1 ≈ PVSVṬ 587,19).
17 I.e., independently of scripture (PVṬ Ñe D47b6/P54b4 = PVSVṬ 587,23), but rather,
through perception, etc., as accompanied by scripture (āgamasahita; compare PVṬ

Ñe D47b6–7/P54b4–5 and PVSVṬ 587,24). This is of course reminiscent of Bhartṛ‐
hari’s reservations about inference. VPR 1.30–32, 34: nāgamād ṛte dharmas tarkeṇa
vyavatiṣṭhate / ṛṣīṇām api yaj jñānaṃ tad apy āgamapūrvakam // dharmasya cāvy‐
avacchinnāḥ panthāno ye vyavasthitāḥ / na tāṃl lokaprasiddhatvāt kaś cit tarkeṇa
bādhate / avasthādeśakālānāṃ bhedād bhinnāsu śaktiṣu / bhāvānām anumānena
prasiddhir atidurlabhā // … yatnenānumito ’py arthaḥ kuśalair anumātṛbhiḥ / abhi‐
yuktatarair anyair anyathaivopapādyate //. “And without āgama, reasoning cannot
determine merit; even the knowledge of seers presupposes āgama. And no one can
by means of reasoning invalidate the paths of merit which have been determined
without interruption, because they belong to ordinary knowledge. Since things can
have different powers due to differences regarding conditions, places, and times,
their knowledge (prasiddhi) is extremely difficult to obtain through inference …
36 PVSV 164,24–176,16
[can something like heaven be] called “supersensible” and [at the same
time be] the object of [means of valid cognition] such as perception?19
Objection: Yet even with regard to their own objects, these [means of
valid cognition] are effective only in dependence on scripture.20 [An‐
swer:] There would [then] be no [inferential] cognition of fire, etc., on
the basis of a [logical reason] such as smoke without scripture [as its
support]! Objection: One certainly does not seek after another means
Even something that has been inferred with [great] effort by skillful experts in in‐
ference is demonstrated [to be] otherwise by other more skillful ones.” Note also
VPV 89,2–6 ad VPR 1.32 (quoted in Akamatsu 2010: 187): tatra rūpasāmānyād apa‐
hṛtabuddhiḥ parokṣaviśeṣo durjñānaṃ bhedam arvāgdarśano darśanamātreṇāgam‐
yam āgamenaiva prapadyate / kālabhedād api / grīṣmahemantādiṣu kūpajalādīnām
atyantabhinnāḥ sparśādayo dṛśyante / tatra sūkṣmam avasthānaviśeṣaṃ prākṛtam
aprākṛtagamyam āgamacakṣur antareṇāpratyakṣam anumānamātreṇāniścitaṃ kaḥ
sādhayitum asammūḍhaḥ prayatate /. “The ordinary man (arvāgdarśana), misled by
external resemblance, is unable to see the difference and can see it only with the
help of tradition. Similarly, properties of things change with time. The temperature
of the waters of a well and the like is very different in summer and winter. Which in‐
telligent man would try to demonstrate, by mere reasoning, this subtle difference in
nature, imperceptible to the ordinary man, unascertainable by inference and in‐
comprehensible except through knowledge derived from tradition?” Translation
Iyer 1965: 44. See also Akamatsu’s (2010: 188) comments on this passage. On Bhar‐
tṛhari’s views on inference, see e.g. Iyer 1992: 84–86, Aklujkar 1989, Houben 1997:
322–327 and Akamatsu 2010. Note also the following passage of Pakṣilasvāmin’s
NBh (3,11–12): kaḥ punar ayaṃ nyāyaḥ? pramāṇair arthaparīkṣaṇam / pratyakṣā‐
gamāśritaṃ cānumānam / sānvīkṣā /. “But what is this ‘reasoning’? The examining
of an object through the means of knowledge; and inference depends upon percep‐
tion and scripture. This is [what is meant by] ‘analysis.’” Translation Perry 1997:
451. On Pakṣilasvāmin’s problematic assertion and its ideological background, see
Perry 1997: 450–452; on Uddyotakara’s, Vācaspati’s, and Udayana’s comments on
NBh 3,11–12, see Perry 1997: 452–457.
18 Introduction, PVṬ Ñe D47b7/P54b5: ji skad du brjod pa na śin tu lkog tu gyur pa’i
don la yaṅ mṅon sum la sogs pa’i ’jug par ’dod pa ma yin nam /. “[But] in so speaking,
don’t [you] accept that [pramāṇas] such as perception also function with regard to
something supersensible?”
19 I.e., either something is the object (*viṣaya) of perception, etc., and it is not transcen‐
dent (*atyantaparokṣa), or it is transcendent, and pramāṇas such as perception do

not apply to it. As a consequence, there can be no consistency with other pramāṇas
in the case of something transcendent (PVṬ Ñe D48a1–2/P54b6–7).
20 As a consequence, scripture is the main epistemic requirement (*prādhānya; PVṬ
Ñe D48a3/P55a1).
TRANSLATION 37
of valid cognition when scripture applies;21 however, whether scrip‐

ture applies is in itself not known.22 [Answer: If that is so,] how does
the addition of this scripture enhance these [means of valid cognition]
which are capable of establishing [supersensible things] by them‐
selves? [If] on the other hand, a [means of valid cognition is by itself]
incapable [of establishing them], it will surely not prove that scripture
applies [to supersensible things] either. How indeed does another
[means of valid cognition] establish the fact that scripture applies,
which, as it relates to supersensible things, is supersensible [itself]? In
this way, however, something other [than authorlessness] should be
the definition of scripture,23 for
[from all this] it follows that a [Vedic] statement that has been
sanctioned24 by a [person] whose statements are [otherwise]
consistent with valid cognitions is scripture. [In that case] the
[Veda’s] authorlessness [turns out to be] useless. (PV 1.315)
If, even though the account of scripture [as authorless] should remain
the same, it is due [only] to valid cognitions that a scripture is scripture
with regard to a certain [meaning], [then] the consistency of [its]
statements with valid cognitions should be the definition of scripture,
not not being composed by a human being (apuruṣakriyā). [This is so]
because, though the [authorlessness of the Veda] is the same for all the
meanings [taught by human interpreters], one [still] cognizes [the cor‐
rect meaning only] on the ground that it is not negated by valid cogni‐
tions; for even though this [authorlessness] is given, you do not accept
21 For otherwise, scripture would lose its independent effectiveness with regard to its
own object (*svaviṣaya; compare PVṬ Ñe D48a5/P55a4–5 and PVSVṬ 588,11–12).

22 I.e., one does not know whether its application to something supersensible is reli‐
able or not; therefore, one seeks after another means of valid cognition in order to
evaluate (*parīkṣā, *vicāra) a particular application (PVṬ Ñe D48a5–6/P55a5–7).
23 For
those who claim that one should adopt (√grah), among many interpreters
(vyākhyātṛ), the teaching (bhāṣita) of the one who agrees (saṃsyandayati) with pra‐
māṇas such as perception (PVṬ Ñe D48b3–4/P55b5–6 ≈ PVSVṬ 588,21–22).
24 kṛta is explained as saṃskṛta (PVṬ Ñe D48b5/P55b7–8 = PVSVṬ 588,23–24). PVṬ
Ñe D48b5/P55b8 ≈ PVSVṬ 588,24–25: vacanasya ca saṃskāras tadarthasya pramā‐

ṇānugṛhītatvakhyāpanam /. “And to sanction a [Vedic] statement is to declare that
its meaning is supported by the pramāṇas.” See also Vibh 404 n. 5.
38 PVSV 164,24–176,16
[that one cognizes the correct meaning25] in another case, [where the
meaning is] inconsistent with valid cognitions.
Furthermore,
if [in fact] there were knowledge of something transcendent
without scripture [then] it would be admitted [by you] that
there is someone who knows supersensible things. (PV 1.316)
If it is accepted that [one] human being [such as Jaimini] has true
knowledge of a transcendent thing (parokṣa) without depending on
scripture, [then] it would be admitted that there are [other] human
beings who [, just like Jaimini,] perceive supersensible things, 26becau‐
se [any] valid cognition, insofar as it is dependent on perception, is im‐
possible without a [prior] perceptual cognition of those [things].27 It is
indeed because other means of valid cognition are impossible, since
perception does not function with regard to those [supersensible
things], that scripture is a means of valid cognition for cognizing
[things] which are not the objects (ālambana) of these [other means of
valid cognition]. But since the functioning of other means of valid cog‐
nition [with regard to supersensible things] implies (anvākarṣati) a
[prior] perception, a superior person should not be excluded. There‐
fore, [you have to agree that] other means of valid cognition do not
function with regard to supersensible [things. And] because he does
25 Or: “you do not accept [that the scripture makes something known correctly]” (PVṬ
Ñe D49a2/P56a6).
26 The Mīmāṃsaka opponent now claims that interpreters such as Jaimini do not teach
the meaning of the Veda on the basis of their perception of it, but on the basis of an
inference (anumāna; compare PVṬ Ñe D49a5–6/P56b4–5 and PVSVṬ 589,19–20).
According to Karṇakagomin, PVSV 166,21–22 is an answer to this objection (tan
na / yataḥ). But according to Śākyabuddhi, PVSV 166,21–22 completes the first part
of the answer, which must be supplied as follows (PVṬ Ñe D49a6–7/P56b5–6): de ni
ma yin te / gal te dbaṅ po las ’das pa’i don de ni mṅon sum gyis mthoṅ bar ’gyur ba
de’i tshe de sṅon du soṅ ba can gyi rjes su dpag pa de la ’jug par ’gyur ro // ci’i phyir
źe na. “This is not [the case, for only] if he had seen this supersensible meaning
through perception would inference, which presupposes this [perception] (tatpūr‐
vaka), apply to it. – Why?”
27 According to PVṬ Ñe D49a7/P56b6–7, °tad° = atīndriya (in the genitive case: tasyā‐
tīndriyasya); but according to PVSVṬ 589,21, °tad° = pratyakṣa (in the instrumental
case: tena pratyakṣeṇa).
TRANSLATION 39
not, for this very reason,28 comprehend which particular meaning the
[Vedic] scripture refers to, this Jaimini [whom you are always talking
about] or another [person, such as Śabara],
filled with [moral defects] such as desire, does not know the
meaning of the Veda himself, nor [does he know it] from an‐
other [person], nor does the Veda [itself] make [its meaning]
known.29 [So] how is there comprehension of the meaning of
the Veda? (PV 1.317 = PVin 2.35)
Since no human being has overcome the confusion which is due to
[moral] defects, as an expositor [of the Veda] he does not know the su‐
persensible restriction [of Vedic words] to a particular meaning by
himself. Nor does another [person] teach him [this restriction], either,
for there would be the same [undesirable] consequence for this [other
person] too [, namely, he too would be incapable of knowing supersen‐
sible things, due to moral defects]. Indeed, a blind [person] does not
find the way when led by [another] blind [person]! Nor does the Veda
disclose its own meaning itself,30 for it would follow that [Jaimini’s or
someone else’s] instruction would be useless.31 Therefore, this verbal
goitre32 [that we call the Veda] whose meaning has not been compre‐
28 I.e., because the other means of valid cognition do not function with regard to su‐
persensible things (PVṬ Ñe D49b4/P57a4–5 ≈ PVSVṬ 589,30).

29 I.e.,
when one listens to it (śrūyamāṇa) independently of someone’s instruction
(upadeśam antareṇa; PVinṬms 71b4).
30 I.e.,
independently of someone’s instruction (upadeśanirapekṣa; PVṬ Ñe D50a4/
P57b7 = PVSVṬ 590,19–20).
31 I.e., all humans would grasp the meaning of the Veda merely upon hearing it (*śra‐
vaṇamātreṇa; PVṬ Ñe D50a4–5/P57b8–58a1).

32 As a lump of flesh (māṃsapiṇḍa) located between the neck and the head (ghāṭāmas‐
takayor madhye), is called an excrescence or goitre (gaḍu), because it is useless (niṣ‐

phala), the Vedic word too is like a goitre, because it is useless, since it is of un‐
known meaning (aparijñātārtha; PVṬ Ñe D50a5–6/P58a2–3 ≈ PVSVṬ 590,22–24).
According to the Bhāskarī (I.98,9), however, gaḍu (in fact: antargaḍu) refers to a
kubjapṛṣṭhastho māṃsapiṇḍaḥ, a lump of flesh located on the back of a hunchback;
in the ĪPV (I.64,10) a Buddhist opponent compares the permanent self (ātman) to a
hunchback’s hump in that it only results in speculative weariness (kalpanāyāsamā‐
traphala) but brings nothing. See Ratié 2006: 51 n. 31. The (generally pejorative)
expression śabdagaḍumātra is commonplace in Indian philosophy. See, e.g., ĪPVV
I.62,16, AJP I.323,11 (accusing a statement very similar to PVSV 39,6–8 of being śab‐
40 PVSV 164,24–176,16
hended, is in this way33 a veritable splinter. Fixed by the ligaments of

bad doctrine and [extremely] difficult to remove, it causes pain.34
Therefore, what valid cognition is there that the [Vedic] state‐
ment [which is ordinarily taken to mean] “One who desires
heaven should perform the Agnihotra” doesn’t mean “One
should eat dog meat”?35 (PV 1.318 = PVin 2.36)
A [Vedic] sentence such as “One who desires heaven should perform
the Agnihotra” is neither [inherently] close to nor remote from any
dagaḍumātra); see also MBh II.37,17. Note that gaḍu may also refer to a tumor. In
this sense, it occurs in the famous maxim (nyāya): gaḍupraveśe ’kṣitārakavinirga‐
maḥ (“going out of the pupil of the eye when the tumor enters,” Franco 1987: 392–
393 n. 137), which, according to Franco, is an equivalent of “falling from the frying
pan into the fire.” See, e.g., AJP I.56,6, TUS 148,12 (Franco 1987: 149), HBṬĀ 368,1.
33 I.e.,
because one doubts (saṃśaya) whether its meaning is this or that (PVṬ Ñe
D50a6/P58a3–4 ≈ PVSVṬ 590,24).
34 Difficult to remove even for a compassionate being ([mahā]kāruṇika) who would be
eager to do so (PVṬ Ñe D50a7–b1/P58a4–6 ≈ PVSVṬ 590,25–26), it causes pain to

those who put their faith in it (tadabhiprasanna; PVṬ Ñe D50a7/P58a4 = PVSVṬ
590,24–25), or who adopt the Veda (aṅgīkṛtaveda; PVṬ Ñe D50a7/P58a4 = PVSVṬ
590,25).
35 PV 1.317–318 are quoted in NM II.220,4–5 and 10–11. PV 1.318 recurs in a slightly
different form (pāda c: jinaḥ sarvajña ity evaṃ) in TSK 3527/TSŚ 3526. agnihotraṃ
juhuyāt svargakāmaḥ is featured by Śabarasvāmin as an example when discussing
the problem of how the meaning of a Vedic sentence is derived from the meanings
of its words (ŚBh I.110,5 ad MīSū 1.1.24); here, it is not clear that he actually intends
to refer to a particular Vedic passage. When discussing the injunction to perform the
agnihotra, however, ad MīSū 1.4.4, he cites agnihotraṃ juhoti svargakāmaḥ (ŚBh
II.285,1). The Vedic passages that come closest to this are TaitS 1.5.9.1 (agnihotraṃ
juhoti), KāṭhS 6.3 (payasāgnihotraṃ juhoti), and MaitS 1.8.6 (yad evaṃ vidvān agni‐
hotraṃ juhoti). (Garge 1952: 102 considers the Maitrāyaṇī passage, not TaitS
1.5.9.1, as providing the originative injunction, because the latter does not deal pri‐
marily with the agnihotra; KāṭhS 6.3 is evidently a guṇavidhi. For a helpful discus‐
sion of Śabara’s rather imprecise method of citing Vedic passages in general, see
Garge 1952: 39–45.) See also ŚBh III.497,2–509,6 ad MīSū 2.2.13–16. Various other
authors, both Mīmāṃsaka and non‐Mīmāṃsaka, cite agnihotraṃ juhuyāt (svarga‐
kāmaḥ) as the paradigm of a Vedic injunction; see MNP 16,3–17,6; AS 32,1–7; NV
254,4 ad NSū 2.1.59 (agnihotraṃ juhuyāt); etc. On this injunction and its interpreta‐
tion in Mīmāṃsā, see below, pp. 135–140.
TRANSLATION 41
[particular] meaning;36 [hence] we do not see any distinctive property

[of the sentence which would determine] that its meaning is that one
should pour ghee, etc., into a certain element [i.e., fire] in a way that is
admissible [to Brahmins], but not that one should eat dog meat.
37Objection: This consequence is similar for any [scriptural statement,
whether it is authorless or not]: Do people nowadays follow the mean‐

ing of statements whose [original] expounders are no longer living (pa‐
rokṣa) according to the [original] intention, or rather a contrary [mean‐
ing]? [Answer:] No [, we do not have the same consequence in every
case], for a tradition is made possible by [the fact that] a teacher may
reveal his own intention [to his contemporaries, and they in turn re‐
veal it to others, etc.]. Indeed, this is not possible in the case of words
that lack an [original] expounder. And speaking with the intention of
having people understand, [the teacher] follows the common usage of
ordinary conventions. For this reason also, the meaning of this [kind of
scripture] is established, [but] not [that] of authorless words, for in
their case, there is no desire of anyone [to communicate].38 Moreover,
learned people apply themselves to [things which are] to be obtained
and avoided and their causes39 by following reasoning alone, not by
36 Because the hypothesis of a real (vāstava) relation between words and meanings
has been refuted earlier (see above pp. 10–12, and PVSV 172,15–16 below, p. 58;
PVṬ Ñe D50b3/P58b1 = PVSVṬ 591,10).
37 For
a parallel passage, see TSK 2774/TSŚ 2773 and TSPK 735,19–736,3/TSPŚ
891,20–892,12.
38 In PVSV 167,16–21, Dharmakīrti has argued that the consequence is not the same
for any scriptural statement, i.e., that it is possible to know the meaning of a scrip‐
ture of human origin (pauruṣeya): (1a) because of the possibility of a tradition (sam‐
pradāya) and (1b) because of the teacher’s compliance with ordinary linguistic con‐
ventions (saṅketānupālana; PVṬ Ñe D51a4–5/P59a7–8 ≈ PVSVṬ 591,29–31). PVSV
167,21–23 provides a second argument to the same effect: (2) one can ascertain
(niścaya) the meaning of a scripture of human origin by following reasoning (nyāyā‐
nusāreṇa; compare PVṬ Ñe D51a5/P59a8 and PVSVṬ 592,6).
39 I.e., what is to be avoided (heya) is suffering (duḥkha), and its cause consists in acts
and defilements (karmakleśāḥ); what is to be obtained (upādeya) is liberation

(mokṣa), and its cause consists in the knowledge of true reality (tattvajñāna; PVṬ Ñe
D51a6–7/P59b1–3 ≈ PVSVṬ 592,8–9).
42 PVSV 164,24–176,16
[listening to] just the talk [of their predecessors];40 and so the conse‐
quence is not similar. But we shall explain this41 in due course.
42Objection: Isn’t it the case that such a thing as the arrangement of the
world, even though it is not an object [accessible] to reason,43 is known

[by you Buddhists] from the statement of a person which must be as‐
sumed [to be true, on the basis of his reliability in regard to other
40 At least according to Śākyabuddhi and Karṇakagomin, Dharmakīrti alludes here to
the champions of practical rationality (the so‐called “practically rational persons”

[prekṣāpūrvakārin]), the Buddhists (bauddha; PVṬ Ñe D51a6/P59a8–b1 = PVSVṬ
592,7), for “even if the words [of their scriptures] can be [as] polysemic [as those of
the Veda], the Buddhists (saugata) determine, for [their] scriptures, only [such] a
meaning [as is] rational (yuktiyukta) [and] serves a human goal, [but they do] not
[ascertain it] merely by [resorting to] the teaching of other [persons].” (PVṬ Ñe
D51a7–b1/P59b3–4 ≈ PVSVṬ 592,9–13; see also Vibh 406 n. 2: etad uktaṃ bhavati /
anekārthatvasambhave ’pi śabdānāṃ yuktiyuktaṃ puruṣārthopayoginam evāgamār‐
thaṃ niścinvanti saugatā na paropadeśamātreṇa /.) For a parallel passage, see TSPK
735,22–23/TSPŚ 891,23–892,4 (tatra nyāyam evānupālayantaḥ saugatāḥ sudhiyaḥ
pravartante na pravādamātreṇa /); on the Buddhist epistemologists’ “rhetoric of
reason,” see McClintock 2010; on practical rationality, see McClintock 2010: 58–62
and Eltschinger 2007b.
41 I.e.,
that rational persons proceed towards scripture (āgame pravartanam) by
following reasoning (nyāyānupālana; PVṬ Ñe D51b1–2/P59b5–6 = PVSVṬ 592,14).
We are not aware of any subsequent statement of Dharmakīrti to this effect. But one
may think of PV 2 as a whole, and especially PV 2.29–33 (see Eltschinger 2001:
110–114).
42 The Mīmāṃsaka opponent now objects that if the ascertainment of a scripture’s
meaning (āgamārthaniścaya) follows reasoning alone, there cannot be any ascer‐

tainment regarding the transcendent things dealt with in scripture (PVSVṬ 592,16–
17, to be compared with PVṬ Ñe D51b2–3/P59b6–7).
43 Śākyabuddhi and Karṇakagomin refer here to Buddhist cosmology as it appears,
e.g., in AK 3.45 (which they quote): tatra bhājanalokasya sanniveśam uśanty adhaḥ /
lakṣaṣoḍaśakodvedham asaṅkhyaṃ vāyumaṇḍalam //. “Here is how it is thought
that the receptacle world is arranged: at the bottom there is a circle of wind, immea‐
surable, with a height of sixteen hundred thousand leagues.” (Pruden 1988–1990:
II.451, translating La Vallée Poussin 1980: II.138). Other instances of things that are
not accessible to reason (because they are beyond reach in terms of space, time, and
nature, deśakālasvabhāvaviprakṛṣṭa) include the desirable and undesirable results
of intentions such as giving and injury (dānahiṃsādicetanānām iṣṭāniṣṭaphaladānā‐
di), i.e., karmic retribution (PVṬ Ñe D51b3–4/P59b7–60a2 ≈ PVSVṬ 592,17–22).
TRANSLATION 43
things]?44, 45 [Answer:] No, because [we have] no confidence [in such a

person].46 It is indeed not the case that, since [a person has been ob‐
served] not to err with respect to a certain [matter], everything [that
person says] is like that [i.e. true, and this for two reasons: first], be‐
cause one observes that [people who are known to be reliable in re‐
gard to a certain thing do in fact] err [in regard to other things]; and
[second,] because a concomitance between the [verbal] activity of this
[allegedly superior person] and reliability is not established.47 48Be‐
yond that, the [aforementioned] definition of scripture has been ac‐
cepted for lack of [any other] recourse.49 There is [indeed] no ascer‐
44 I.e., if someone’s statement concerning things that are not accessible to perception
and inference proves to be reliable (avisaṃvādin), then one may assume (sambhāv‐
yate) that this person’s statement regarding the third order (tṛtīyasthāna) of things,
viz., transcendent things, is veracious (satyārtha, PVṬ Ñe D51b5/P59a2–4 ≈ PVSVṬ
592,23–24).
45 Consequently, the Buddhists, too, learn about transcendent things from the state‐
ments of a person whom they merely trust, and thus do not deal with the meaning
of their scriptures by following reasoning alone. Therefore, it is the case that the
same consequence (see above, PVSV 167,14–16) applies to them (PVṬ Ñe D51b5–
6/P60a4–5 = PVSVṬ 592,24–26).
46 Dharmakīrti now argues that one cannot ascertain (niś√ci) someone’s reliability
(saṃvāda) in regard to something transcendent on account of his reliability in re‐

gard to things accessible to perception and inference (PVṬ Ñe D51b7/P60a6–7 and
PVSVṬ 592,28–29).
47 Consider the following argument (PVṬ Ñe D52a2–3/P60b1–2): “With regard to
something supersensible also, the verbal activity of this superior person, because it
is his verbal activity, is reliable, like [his] verbal activity regarding things that are
accessible to the [ordinary] pramāṇas.” Here, the concomitance between the pro‐
perties “being reliable” (probandum) and “being this person’s speech” (probans)
cannot be established for want of a valid cognition that excludes (bādhaka) the pre‐
sence of the probans (here: “verbal activity”) in cases dissimilar to the probandum
(here: “not being reliable”; PVṬ Ñe D52a3/P60b3–4 = PVSVṬ 593,12).
48 The Mīmāṃsaka opponent now objects that in PV 1.215, Dharmakīrti himself ac‐
cepts a definition of scripture (āgamalakṣaṇa) that makes the scripture’s partial

reliability (ekadeśāvisaṃvādana, i.e., in regard to empirically accessible things) a
reason for its overall reliability (i.e., in regard to transcendent things as well; PVṬ
Ñe D52a4/P60b4–5 ≈ PVSVṬ 593,12–15). On Dharmakīrti’s doctrine of scriptural
authority, see below, pp. 83–118; see also PVSV 173,26–174,2 below, pp. 63–65.
49 According to Dharmakīrti, people eager to engage in religious practice (pravṛttikā‐
ma) cannot live without resorting to scripture (PVSV 108,2–5) because of their co‐
44 PVSV 164,24–176,16
tainment [of supersensible things] from [scripture thus defined, and]

this is the reason why [we have] also stated [above] that scripture is
not a means of valid cognition.50 Since the meaning of authorless words
[can] be known neither from tradition, nor from reason, nor from the
[ordinary] world,51 it is [only] proper [to say] that there is no cognition
[of the meaning] in this case. Objection: In their case too,
ordinary parlance, which is [commonly] established, (PV
1.319a = PVin 2.37a)
is the cause of [our] cognition [of the meaning].52
[Answer:] What [human being] is there in this [world able] to
perceive supersensible things, who [could] discriminate the
[proper] meaning of [Vedic] words which have multiple mean‐
ings?53, 54 (PV 1.319bcd = PVin 2.37bcd)
gnitive limitations. Now, according to PVSV 108,5–6, if someone is to act according

to scripture (āgamāt pravṛttau), he would do better (varam) to follow a scripture
whose statements about empirical matters, at least, have proved to be veracious
(PVṬ Ñe D52a4–5/P60b5–6 ≈ PVSVṬ 593,15–16). See the references provided
above, n. 48, and PVSV 174,1 below, p. 64.
50 See below, pp.102–105.
51 Not from tradition, because there is no preceptor (daiśika) at the beginning of a
tradition (PVṬ Ñe D52a6/P60b8 = PVSVṬ 593,19); not from reason, because the
pramāṇas do not function with regard to something transcendent (PVṬ Ñe D52a6–
7/P60b8–61a1 = PVSVṬ 593,20); and not from the world, because there is no con‐
cern with following ordinary linguistic conventions (PVṬ Ñe D52a7/P61a1–2
≈ PVSVṬ 593,20–21).
52 I.e., the cause of our ascertainment (niścaya) of the meaning of the Veda (PVinṬms
71b6). And indeed, according to the Mīmāṃsaka opponent, neither is dog meat (śva‐
māṃsa) the thing signified (vācya) by the word agnihotra, nor is eating (bhakṣaṇa)
the meaning of juhuyāt (PVinṬms 71b6). Or, according to Manorathanandin (PVV
405,20), the word agni ordinarily refers to something capable of burning (dāhā‐
disamartha), whereas hotra (havana) points to pouring ghee, etc. (ghṛtādiprakṣepa)
into fire.
53 I.e., whose specific meaning is doubtful (āśaṅkyamānārthaviśeṣa) because it is not
restricted to a single meaning (ekārthapratiniyama; PVṬ Ñe D52b2/P61a4–5

≈ PVSVṬ 593,26–27). Here, Dharmottara refers to the well‐known example of the
nine meanings of the word “cow” (gośabdasya navārthāḥ pratītāḥ; PVinṬms 72a3).
On this example, see AKBh 80,28–81,3: tad yathā gaur ity eṣa śabdo navasv artheṣu
kṛtāvadhiḥ / vāgdigbhūraśmivajreṣu paśvakṣisvargavāriṣu / navasv artheṣu medhāvī
TRANSLATION 45
Indeed, such ordinary [linguistic] practice is not due to an authorless

relation between word and meaning, but to convention55 – just like the
[linguistic] practice of the Pāṇinīyas, etc., [regarding words like vṛddhi
or guṇa] is due to the convention [fixed long ago] by [Pāṇini,] the
founder of their science – because it depends on instruction.56 57In‐
struction, indeed, is not justified in the case of such an authorless [rela‐
tion], since no one knows the [meaning of a Vedic word], because it is
supersensible, [and] because if it were sensible, it would follow that it
would be cognized of itself [i.e., without any instruction], like visible
form, etc. And there can be no confidence in58 the instruction of arbi‐
trary59 human beings [about supersensible matters], because one can‐
not establish the reliability [of their instruction] by [verifying that they
are] teaching truthfully. Objection: Like the Veda, an authorless expla‐
gośabdam upadhārayed iti //. “For example, the word ʻcowʼ has been assigned to
nine meanings [, as it is said]: ʻThe wise should consider the word ʻcowʼ [as being
used] in regard to nine meanings: speech, region, land, light and diamond, as well as
cattle, eye, heaven, and water.ʼ” Note also Yaś omitraʼs (AKVy 183,16–17) comment
on tad yathā gaur ity eṣa śabdo navasv artheṣu kṛtāvadhiḥ: etena saṅketāpekṣaḥ śab‐
do ’rthaṃ pratyāyayati /. “Thus [it is only] in dependence on a convention [that] a
word communicates a meaning.”
54 I.e., there is no person (such as Jaimini or Śabarasvāmin) able to discriminate (*vive‐
cana) the proper meaning of the Vedic words, which are not preceded by a speak‐
er’s intention (PVṬ Ñe D52b3–4/P61a6–7, to be compared with PVSVṬ 593,27–28).
55 Defined as the communication of an intention (abhiprāyakathana; PVṬ Ñe D52b4/
P61b1 = PVSVṬ 594,8). See below, PV 1.327 and PVSV 172,19–24, and Eltschinger
2007a: 134–143.
56 On Pāṇini (and Piṅgala) in Mīmāṃsā, see ŚBh on MīSū 1.1.5/I.63,5–66,2, ŚBhF
42,16–44,12. See also Biardeau 1964: 157–159 and D’Sa 1980: 95–96.
57 In PVSV 168,9–11, Dharmakīrti has argued that if ordinary linguistic practice also
(laukiko ’pi vyavahāraḥ) were established naturally (nisargasiddha), it would not re‐
quire another person’s instruction (paropadeśa). But it actually requires it. There‐
fore, it is due not to an authorless relation between words and meanings, but to
convention. In PVSV 168,11–15, Dharmakīrti now explains why no one can know
the meaning of the Veda through conventional linguistic practice (*sāmayikaśābda‐
vyavahāra; PVṬ Ñe D52b6–7/P61b3–5, to be compared with PVSVṬ 594,12–14).
58 I.e.,
no ascertainment (niścaya) of the meaning of the Veda through… (PVṬ Ñe
D53a4/P62a3 = PVSVṬ 594,25).
59 Because, due to their being overcome by defilements such as desire (rāgādyabhi‐
bhava), those teaching the meaning of the Veda (vedārthopadeṣṭṛ) lack any correct
knowledge (samyagjñāna; PVṬ Ñe D53a3–4/P62a2–3 ≈ PVSVṬ 594,24–25).
46 PVSV 164,24–176,16
nation of it has also come down [to us] through an unbroken tradition,
[and it is] by means of this [explanation that] the meaning [of the Veda]
is established. [Answer:] This [explanation] also, being verbal in na‐
ture, [is confronted with] the same question [as the Veda]: How is its
meaning known? A human being [founding a tradition] can indeed im‐
part ostensively (śṛṅgagrāhikayā) the meanings of words, for which he
himself has fixed conventions, to someone who is not yet (tāvat) fa‐
miliar [with them]. Therefore, there is a means of gaining access to the
meaning of words having an author. An authorless word, however,
does not do this, and nobody can know the restriction of the relation
[of the word] to a certain [supersensible meaning]. Therefore, [there
can be] no cognition of its meaning at all. Moreover, the Veda or its ex‐
planation [, even though authorless, still] follows an uninterrupted tra‐
dition taught by one human to [another] human. Thus in this case too,
an oath (samaya) [swearing that it is true] is all you have to fall back
on.60 For one observes that the destroyers of scripture falsify [it] either
out of pride, or due to hostility towards its doctrines (darśana),61 or out
of sheer malevolence in order to humiliate those who adhere to them
(pratipanna), or for some other reason. Moreover, out of attachment to
your own theory you have forgotten the color of your [own] face
(mukhavarṇa)62 now that it comes to (atra) [determining the meaning
60 Because there is no
pramāṇa enabling one to ascertain (*niścāyakapramāṇa) its
meaning (compare PVṬ Ñe D53b4/P62b6 and PVSVṬ 595,17–18). samaya is ex‐
plained as śapathādi in PVṬ D53b4/P62b6–7 ≈ PVSVṬ 595,18. On the Indian oath,
see Hara 1991; on the juridico‐philosophical use of the oath, see Eltschinger 2007a:
302 n. 339.
61 Out of pride, as the Sāṅkhyasiddhānta was altered (anyathāracana) by Mādhava, the
destroyer of Sāṅkhya (sāṅkhyanāśaka; PVṬ Ñe D53b5–6/P62b8–63a1 = PVSVṬ

595,21–22; on Mādhava as a sāṅkhyanāśaka, see Hattori 1968: 134 n. 4.16 and 155
n. 5.40); due to hostility, as the enemies of the Mahāyāna (mahāyānavidviṣṭa) com‐
posed (pseudo‐)Mahāyānasūtras expounding a counterfeit Mahāyāna (mahāyāna‐
pratirūpaka; PVṬ Ñe D53b6/P63a1–2 ≈ PVSVṬ 595,22–23).
62 So that you are now looking for (pra√arth) the ascertainment of the meaning of the
Veda from humans who, according to your most characteristic doctrine, are tarni‐
shed by defilements such as desire (rāgādimalina; PVṬ Ñe D54a1–2/P63a5–6
= PVSVṬ 595,29). Skt. svam eva mukhavarṇam is rendered bdag ñid kyis khas blaṅs
pa in PVSVTib D361a2/P528b3–4 and PVṬ Ñe D54a1/P63a5, and explained by the
synonymous svābhyupagama in PVSVṬ 595,28–29: “the color of your own face,” viz.,
your most characteristic tenet, which Dharmakīrti gives in PVSV 169,1–2.
TRANSLATION 47
of the Veda, namely, the idea that] since a human being afflicted with
[moral defects] such as desire, may also speak what is false, his speech
is not a means of valid cognition. Why don’t [you] consider63 whether
this [falsehood] is present or not in this [tradition of exegesis] too?
[For,] due to being afflicted [by moral defects] this very [person]
teaching the Veda or the meaning of the Veda may also teach falsely.
64Indeed, we hear from certain people65 that there are certain Vedic
schools which have reappeared after nearly dying out – even today,
some have very few reciters.66 [Therefore, those who teach and recite
the Veda now may very well do it falsely] for, like the [schools which
now have only a few reciters, those schools] which have numerous re‐
citers could have somehow declined at a certain time [in the past],
since one can suppose that [the reciters] grew again in number due to
[their] confidence67 in a person esteemed [to be an authority]. And
[thus, there is doubt about their reliability for three reasons: First,] be‐
cause those who restored [the Vedic schools which had declined],
having sometimes forgotten the recitation [which they had] learned,
could have taught it falsely on account of such [motives] as fear that
they would lose the esteem (sambhāvanā) of others;68 and [second,]
63 I.e., without partisanship for Jaimini, etc. (*jaiminyādipakṣapāta; PVṬ Ñe D54a3–
4/P63a7–b1).
64 In talking, in PVSV 169,4–15, about the demise of Vedic schools (śākhā), Dharma‐
kīrti provides additional reasons for distrusting (āśaṅkākāraṇ[āntar]a) traditional

Vedic exegesis (PVṬ Ñe D54a5/P63b3 ≈ PVSVṬ 596,11).
65 I.e., from Jaimini, etc., according to Śākyabuddhi (PVṬ Ñe D54a5/P63b3), or from
Yājñavalkya, etc., according to Karṇakagomin (PVSVṬ 596,11–12). However, there is

no mention of such a thing in either the MīSū or the Yājñavalkyasmṛti.
66 PVSVṬ 596,14–15 adduces the Āhūrakas, etc., as an example. Āhūraka seems not to
be attested as a designation for a Vedic school/recension, but according to AVP

49.2,1, the Āhvarakāḥ form one of the twenty‐four divisions (bheda) of the Yajur‐
veda.
67 I.e., due to the fact that they acknowledge his (personal) authority (prāmāṇya; PVṬ
Ñe D54b2/P64a1 ≈ PVSVṬ 596,20–21). See also below, p. 48 n. 69.

68 I.e., of other reciters who have placed their faith in them (tadabhiprasanna; PVṬ Ñe
D54b5/P64a6 = PVSVṬ 596,28). Their fear can be expressed as follows: “If I didn’t
at least (api) teach [something] otherwise [i.e., falsely, rather than saying nothing at
all], they [would] immediately lose [their] confidence in me as someone who is
trustworthy” (PVṬ Ñe D54b5–6/P64a6–7 ≈ PVSVṬ 596,29–30: yady aham anyathāpi
nopadiśeyaṃ nūnam ete mayy āptasambhāvanāṃ jahātīti /). (I.e., if I don’t fake it, I’ll
48 PVSV 164,24–176,16
because, due to [their] confidence in the [person believed to be an

authority], his devotees [could have] practiced (pratipatteḥ) [his
recitation] carelessly, for one observes [nowadays] that among the
many reciters [of the Veda, recitation] practice [proceeds] mostly be‐
cause of a person who is esteemed [as trustworthy];69 [and third,] be‐
cause deception can arise somehow70 from that [sort of person] also.
Furthermore, we hear that your own [tradition] regarding the [inter‐
pretation of the Veda has] also [consisted in] a succession of human
expositors [at times] limited in number. Since among them, one could
also have been either hostile, ignorant, or malevolent, there can be no
confidence [in your explanation of the Veda].
Therefore, the meaning of the Veda [can be] established neither
through an authorless explanation nor through ordinary [linguistic]
practice, which is conventional. [Ordinary linguistic practice cannot es‐
tablish it for two reasons: first,] because even if [Vedic words] were not
conventional, words are seen to have various meanings in [ordinary
linguistic] practice; therefore doubt [about the meaning] would not
cease for any [word], 71for [even in the case of Vedic words] one ob‐
serves that a certain [word] of uncommon meaning or an uncommon
[word] has [to be] explained anew.72 73[Second,] because, whatever
lose their trust!) Additional motives include pride, hostility, and malevolence (see
above, PVSV 168,24–26; PVṬ Ñe D54b6/P64a7–8 ≈ PVSVṬ 596,30–31).
69 I.e., because of a person whose personal authority (puruṣaprāmāṇya; PVSVṬ 597,
11–12) is acknowledged by the reciters, or due to their confidence in a human being

who is or has become an authority (*pramāṇabhūtapuruṣa; PVṬ Ñe D55a1/P64b2–
3). On pramāṇabhūta(puruṣa), see Krasser 2001.
70 I.e., because of motives such as a teacher’s fear of losing the esteem of others (PVṬ
Ñe D55a2–3/P64b4–5 ≈ PVSVṬ 597,15).

71 The Mīmāṃsaka opponent now claims that only ordinary (laukika) words are poly‐
semic (nānārtha), not Vedic ones (PVṬ Ñe D55b1/P65a5–6 = PVSVṬ 597,26).

72 As a consequence, one may again wonder whether the interpreter teaches the cor‐
rect (yathāsthita) meaning of these Vedic words or an erroneous (viparīta) one

(PVṬ Ñe D55b2–3/P65a7–8 ≈ PVSVṬ 597,28–30).
73 The Mīmāṃsaka opponent now claims that, since Vedic words do not generate the
cognition (*pratīti) of their meaning if they are used in a way contrary (*anyathā) to
the desired meaning (*abhimatārtha), they have only one meaning (*ekārtha). In
other words, these words behave differently according to whether they are used to
designate a desired (*iṣṭa) or undesired (*aniṣṭa) meaning (PVṬ Ñe D55b5–
6/P65b3–4, to be compared with PVSVṬ 598,12).
TRANSLATION 49
meaning [the word] may be employed for, all [words, both ordinary
and Vedic,] generate the understanding [of a meaning] indifferently
(avaiguṇyena) according to a convention, since there is no difference74
between [meanings which are] desired and undesired [for followers of
the Veda].75 Who among average human beings, who do not perceive
the supersensible, could discriminate for [Vedic words which are in
themselves] indifferent with regard to all meanings, the one [correct]
meaning, whose connection [with the word] is supersensible, so that
(yataḥ) the understanding could arise from the [ordinary] world?
Moreover, even the [Mīmāṃsaka] himself does not comply with com‐
mon usage76 in every case, because
words such as77 svarga and urvaśī are seen to express an
unconventional meaning (PV 1.320ab = PVin 2.38ab)
when [they] are interpreted by him. According to ordinary parlance (iti
lokavādaḥ) heaven is the abode of special beings (puruṣaviśeṣa) supe‐
rior to [ordinary] human beings, a place of divine pleasure, offering
abundant enjoyments, [while] the one called Urvaśī is a nymph resid‐
ing there. How can the [Mīmāṃsaka], disregarding this [ordinary par‐
lance and] proposing a totally new (anya) meaning,78 set up common
74 I.e., no difference in generating the understanding (pratītijanana; PVṬ Ñe D55b6/
P65b5 = PVSVṬ 598,13).

75 Because
no words are either inherently close to or remote from any particular
meaning (pratyāsattiviprakarṣābhāvena; PVṬ Ñe D55b6/P65b5 = PVSVṬ 598,13).
See above, p. 41 n. 36.
76 So that one could gain access to the meaning of the Veda through commonly estab‐
lished linguistic practice (compare PVṬ Ñe D56a1/P65b8–66a1 and PVSVṬ 598,17–

18).
77 Additional words with unconventional meaning include nandanavana (Tib. dga’ ba’i
tshal, i.e., nandana‐vana, “[Indra’s] pleasure grove;” PVṬ Ñe D56a2–3/P66a2

= PVSVṬ 598,19).
78 According to ŚBh V.72,6–7 on MīSū 4.3.15 as well as ŚBh V.179,11–180,5 on MīSū
6.1.2, svarga is to be defined as happiness or delight (prīti; see below, pp. 139–140
n. 78; cf. Eltschinger 2007a: 244 n. 130). As for urvaśī, it is explained by Dharma‐
kīrti’s commentators as a piece of wood used for kindling fire (araṇ) or as a vessel
(pātrī; PVṬ Ñe D56a4–5/P66a5–6 ≈ PVSVṬ 598,25–26; PVV 406,9; note that
PVinṬms 72a has only araṇi). The apsaras Urvaśī is prominent in Vedic and Hindu
mythology. Dharmakīrti must have been familiar with the story of Urvaśī and her
human consort, king Purūravas, which is told in the HV (App. I.6 [found in some
50 PVSV 164,24–176,16
usage as the criterion in the case of other words [such as agnihotra]?

Objection: Since in the case [of words such as agnihotra] there is no
contradiction [of the postulate of the common meaning], we accept
mss. after HV 21.10 of the critical edition]) and various Purāṇas, and which is the
theme of Kālidāsa’s drama the Vikramorvaśīya; it goes all the way back to ṚV 10.95
and ŚB 11.5.1. Other stories of Urvaśī – about her seducing Ṛṣyaśṛṅga and
attempting to seduce Arjuna – are found in the Mahābhārata (Hopkins 1974: 162).
A reference to urvaśī as araṇi remains to be traced in the Mīmāṃsā literature.
Nevertheless, one may guess that Dharmakīrti was aware of Mīmāṃsā explanations
of the mantras to be applied during the kindling of the fire by friction
(agnimanthana) in the Soma sacrifice, namely, agnér janítram asi vŕ̥ṣaṇau stha
urváśy asy āyúr asi purūrávā ghṛ́tenākté v ṛ́ṣaṇaṃ dadhāthām (TaitS 1.3.7.1h–l; for
the corresponding brāhmaṇa see TaitS 6.3.5.2–3) (also: agnér janítram asi v ṛ́ṣaṇau
stha urváśy asy āyúr asi purūrávā asi; MaitS 1.2.7; brāhmaṇa MaitS 3.9.5). The
adhvaryu priest mutters, “Thou art Urvaśī,” while picking up the lower araṇi and,
“Thou art Purūravas (or: Āyus),” while picking up the upper. In other words, Urvaśī
is taken by the ritualists to refer, not to a heavenly nymph, which is the common
meaning of the word, but to one of the kindling sticks! See also VādhŚSū 5.2.1.4
(Voegeli 2010: II.24), BaudhŚSū 20.27; BhārŚSū 7.9.13–15; ĀpŚSū 7.12.13–14;
VaikhŚSū 8.5, 10.10. For a discussion, see Voegeli 2010: I.165–167; see also Gonda
1988: 229 (VājS 5.2c and ŚB 3.4.1.22). We are not aware of any identification of
urvaśī with a pātrī (a kind of dish by which the adhvaryu brings the iḍā portion of
the paśupuroḍāśa to the brahman priest, or to the āgnīdhra priest, according to
Voegeli 2010: I.173). nandanavana, the other example mentioned by Karṇakagomin
and Śākyabuddhi of a word used by Mīmāṃsakas in violation of its common
meaning (see above, p. 49 n. 77), poses a special problem, because not only is there
no known Mīmāṃsā treatment of it but it does not occur in any Vedic mantra, ei‐
ther. In epic and later literature it is the name of the pleasure grove in Indra’s city
where gods and saints dwell (MBhār 1.84.17, 3.78.3, 12.329.19; see Hopkins 1974:
141). It seems relevant to mention here that it was a general principle of Mīmāṃsā
that proper names in the Veda do not refer to “non‐eternal” entities; there is
“merely a similarity of sound” of some words in the Veda (Śabara mentions babara
prāvāhaṇi and kusumavinda auddālaki) to proper names (MīSū 1.1.31, paran tu śru‐
tisāmānyamātram). They, in effect, refer to concepts. It is possible that urvaśī was
explained in this way as well by some Mīmāṃsakas. Yāska gives etymologies of ur‐
vaśī at Nir 5.13: uru abhyaśnute, ūrubhyām aśnute, urur vā yaśo ’syāḥ, “She ranges
widely, or she pervades with her thighs, or her desire is extensive.” Whether Dhar‐
makīrti had in mind this kind of explanation of urvaśī or, as his commentators im‐
ply, the explanation about Urvaśī rubbing together with Purūravas (as one of the
kindling sticks), he must have thought it was a good joke! See also below, pp. 126–
127.
TRANSLATION 51
[it].79 [Answer:] No, because in the case [of words such as svarga and
urvaśī] a contradiction of the [common meaning, which is] something
supersensible, is not established, [and] because in the other case [i.e.,
in the case of words such as agnihotra, the fact that it] is not contra‐
dicted is hard to grasp.80 Even if the attainment of heaven by [perform‐
ing] the Agnihotra were contradicted [by other pramāṇas], this
[expositor of the Veda] would not even notice it due to [his] dullness.
Moreover, contradiction and non‐contradiction are [nothing but, on the
one hand,] the occurrence of a means of valid cognition which negates
[something, e.g., non‐perception,] and [, on the other hand,] the [occur‐
rence] of a means of valid cognition which establishes [something, i.e.,
perception or inference]; but neither of these is admitted [as possible]
in the case of something supersensible. How, then, could one ever un‐
derstand [that a Vedic word has the commonly accepted meaning or
some other meaning] due to them?81 82And it is not the case that there
is no contradiction [of the commonly accepted meaning] just due to the
79 On the other hand, since there is a contradiction of the common meaning in the case
of words such as svarga and urvaśī, we don’t accept it (PVṬ Ñe D56a7/P66a8–b1

≈ PVSVṬ 598,30–31).
80 In the case of words such as svarga and urvaśī one would require a pramāṇa that
opposes (bādhaka) the common meaning (PVṬ Ñe D56b1/P66b3 ≈ PVSVṬ 599,8) in

resorting to another meaning, viz. non‐perception (anupalambha; PVṬ Ñe D56b4
[with no equivalent in P66b6] = PVSVṬ 599,14). In the case of words such as agni‐
hotra one would require a pramāṇa that establishes (sādhaka) the common mean‐
ing (compare PVṬ Ñe D56b2/P66b4 and PVSVṬ 599,9–10), viz., perception or infer‐
ence (PVṬ Ñe D56b4/P66b6 = PVSVṬ 599,14–15). But since these words refer to
supersensible things and no pramāṇa, negative or positive, functions with regard to
such things (PVṬ Ñe D56b5/P66b7–8 ≈ PVSVṬ 599,16–17), a contradiction can be
neither established nor ruled out.
81 I.e., that svarga and urvaśī do not have the commonly established meaning (apra‐
siddhārtha) because of a contradiction (virodhavaśāt), and that agnihotra has the

commonly established meaning (prasiddhārtha) because of non‐contradiction (avi‐
rodhavaśāt; compare PVṬ Ñe D56b6–7/P67a1–2 and PVSVṬ 599,18–19; PVṬ Ñe
D56b6/P66b8 = PVSVṬ 599,18).
82 The Mīmāṃsaka opponent now claims that non‐contradiction does not consist in
the operation (vṛtti) of a positive pramāṇa (which, he seems to agree, is impossible

in regard to the supersensible realm), but just in the occurrence (pravṛtti) of a state‐
ment of the agnihotra, termed “scripture” (āgamasañjñita), as expressive (vācaka)
of the commonly established meaning (PVṬ Ñe D56b7–57a1/P67a2–3 ≈ PVSVṬ
599,20–22).
52 PVSV 164,24–176,16
occurrence of the [Vedic] statement [in the case of agnihotra], for

[that83] would follow in the other case as well [i.e., in the case of words
such as svarga and urvaśī].84 [Furthermore, your] scripture [called the
Veda] is [supposedly] authorless [but it does not reveal its own mean‐
ing by itself; on the contrary, you say] its meaning is established
through the [ordinary way of] talking. But if in [following the latter
you] are concerned about contradictions, then there can be no confi‐
dence in [any part of that] scripture.85 For even if this [scripture] is
[authorless], the meaning [of svarga and urvaśī understood by you] is
unlike [the commonly accepted one]; therefore, another [word such as
agnihotra] too can be suspected [of having a different meaning], for
there is no pramāṇa.86 [Objection:] When [you] state that one can con‐
ceive the [Vedic statement] “One who desires heaven should perform
the Agnihotra” as enjoining87 (deśanā) that one should eat dog meat,
[we reply that] this is not [the case], because this [statement] is re‐
peated in [exactly] this way in other passages [of the Veda itself]. [An‐
swer:] No, because, the meaning of this [statement occurring in other
passages] not being comprehended [either], the conjecture that [these]
other [Vedic] passages also have such a meaning cannot be excluded. If
this authorless multitude of words had a known meaning in one par‐
ticular [passage], then one could indeed understand the meaning [of
these other passages] from that. But since, even though there are so
many of them, all these [words remain] completely obscure, they are
fixed [in their meaning] arbitrarily. Therefore,
83 I.e., non‐contradiction with the commonly established meaning (PVṬ Ñe D57a1–
2/P67a4–5).
84 Because one observes that in common usage (loka) words such as svarga and urvaśī
are employed to refer to a specific place (sthānaviśeṣa) and a specific nymph (apsa‐
roviśeṣa). Since that is the same for agnihotra – it is commonly used to refer to a sac‐
rifice – one should adopt (parigraha) the commonly established meaning either for
both – i.e., svarga/urvaśī and agnihotra – or for neither (PVṬ Ñe D57a2–3/P67a5–6
= PVSVṬ 599,24–26).
85 Because with regard to something supersensible it is impossible to ascertain either
a contradiction or the absence of one (PVṬ Ñe D57a4–5/P67b1 ≈ PVSVṬ 599,29–

30).
86 I.e., there is no pramāṇa allowing one to adopt (grahaṇa) the commonly established
meaning in this case (PVṬ Ñe D57a7–b1/P67b5–6 = PVSVṬ 600,14–15).

87 On deśanā/codanā, see Gnoli 1960: 170 n. ad line 6.
TRANSLATION 53
in the case of other [explanatory] statements of this type,88 the

conjecture ought to be exactly the same (PV 1.320cd = PVin
2.38cd)
as for the sentence “One who desires heaven should perform the Ag‐
nihotra.”89 Moreover,
common usage is human parlance (nṛṇāṃ vādaḥ); and this
[parlance] is not accepted [by you] as a pramāṇa. But on the
other hand it is from that that one gains access to the meaning
[of the Veda]. Why [do you both] approve and disapprove of
this? (PV 1.321 = PVin 2.40)
What is called common usage is nothing other than the way humans
talk; and since [they are] filled with [moral defects] such as desire, and
by nescience, all people make statements whose truth cannot be as‐
sumed.90 Therefore, their talk is not a pramāṇa. 91Indeed, [the fact that
there are] many [people who conduct themselves in a certain way] is
worthless [as an argument], if there is not even one [person among
them] who acts correctly, just like [incest does not become acceptable
just because all] Persians misbehave with [their] mothers.92 [But]
88 I.e.,
in similar statements that serve as explanations (vyākhyābhūta) of phrases
(vākya) such as agnihotraṃ juhuyāt (PVṬ Ñe D57b5/P68a5 = PVSVṬ 600,27–28).
89 Between PV 1.320 (= PVin 2.38) and PV 1.321 (= PVin 2.40), PVin 2 has one stanza
with no equivalent in the PV. PVin 2.39 (see Steinkellner 1979: 74): aprasiddhārtha‐
yogasya tatprasiddhiprasādhane / nāsiddhārthaḥ svayaṃ śaktas tulyaparyanuyoga‐
taḥ //. “[An additional Vedic passage, being] of unknown meaning [also], is by itself
[i.e., independently of any human intention,] incapable of establishing that common
usage [as the meaning] of a [statement] with an unknown connection with its mean‐
ing, because the same question [would arise regarding this additional passage too,
namely, How does one ascertain its connection to its meaning?].” For Dharmottaraʼs
commentary on this stanza, see PVinṬms 72b1–2/PVinṬTib Dze 228b6–229a1.
90 For
analyses of the compound asambhāvanīyayāthātathyavacanāḥ (called a ’bru
maṅ po’i sñiṅ po can gyi ’bru maṅ po pa’o, i.e., a *bahuvrīhigarbho bahuvrīhiḥ, PVṬ Ñe
D58a3/P68b5), see PVSVṬ 601,16–17, and especially PVṬ Ñe D58a2–3/P68b3–5.
91 The Mīmāṃsaka opponent now claims that since these people are many (bāhulyāj
janasya), their talk (tatpravāda) is a pramāṇa after all (PVṬ Ñe D58a4/P68b6–7

= PVSVṬ 601,20).
92 On this and similar statements regarding Persians, see Eltschinger 2007a: 312 n.
377, Silk 2008a and 2008b.

54 PVSV 164,24–176,16
since, on the other hand, it is from the speech of those very human be‐
ings that the imperceptible (parokṣa) meaning is arrived at, how can
this [ordinary speech] be simultaneously approved and disapproved
of?
But if [you see] no reason to ignore common usage and postu‐
late [another meaning], what reason [do you have] to adopt it,
given that common usage is not a pramāṇa? (PV 1.322 ≈ PVin
2.41)
Objection: Using [words] in another [meaning] contrary to the [mean‐
ing] obtained [from common usage] is justified [only if] one observes
an advantage [in the uncommon meaning] and a defect [in the common
one]. [If this is not the case,93] then [we] follow common usage. [An‐
swer:] No, because the obtaining [of a meaning] is defined as the occur‐
rence of a pramāṇa [which establishes it. But] for the [Mīmāṃsaka],
who does not set up common usage as a pramāṇa, the understanding
[of the meaning obtained] by means of this [common usage] is a haph‐
azard grasping (yatkiñcanagrahaṇa) indeed, because [by rejecting
common usage] he [in effect] prohibits [the meaning] from being ob‐
tained according to a rule (nyāyāt).94 Since the way [your] own and
[your] adversary’s conceptions [of the meaning95 are arrived at] is the
same [as regards its arbitrariness] in both ways [i.e., whether one pre‐
serves common usage or not],96 what is [this] priviledge [given] to
common usage? Moreover,
it is precisely because of common usage that PVSVthisPVSV doubt
regarding the ascertainment of the meaning of words has
93 I.e., if, as in the case of words such as agnihotra, one observes neither a defect in the
commonly established meaning nor an advantage in the unconventional one (apra‐

siddha; PVṬ Ñe D58b3–4/P69a7–8 ≈ PVSVṬ 602,11–12).
94 Consequently, since the meaning arrived at is devoid of any pramāṇa (*niṣpramāṇa‐
ka), the meaning of words such as agnihotra can only be postulated arbitrarily (ic‐
chayā parikalpanīyaḥ; compare PVṬ Ñe D58b6–59a1/P69b5–7 and PVSVṬ 602,19–
21).
95 For the Mīmāṃsaka the meaning of words such as agnihotra consists in a burning
substance, etc. (dahanadravyādi), whereas for his adversary, it consists in the eating
of dog meat (śvamāṃsabhakṣaṇa; PVṬ Ñe D59a1–2/P69b7–8 ≈ PVSVṬ 602,22–23).
96 I.e., in both cases, the adoption (parigraha) of the meaning is made arbitrarily for
want of any pramāṇa (PVṬ Ñe D59a2/P69b8–70a1 = PVSVṬ 602,23–25).

TRANSLATION 55
arisen, for in that [common usage] words [like agni] are seen to
have various meanings.97 (PV 1.323 = PVin 2.42)
It is not the case that one can ascertain the one [correct] meaning [of
Vedic] words from common usage, for it is precisely due to it that
doubt arises, because words [like agni] are seen to have various mean‐
ings in the world. But common usage (pratīti) is ordinary parlance,
[and] it is due to this [ordinary parlance that words] have various
meanings. Therefore, one is not justified in restricting [Vedic words] to
a single meaning on the basis of it.
Because it is not impossible that an utterance of itself pos‐
sessed of various capacities, [could refer] otherwise [than to
97 Between PV 1.323 (= PVin 2.42) and PV 1.324 (= PVin 2.45), PVin 2 has two stanzas
that have no equivalent in the PV. PVin 2.43–44 (see Steinkellner 1979: 75–76): na
yuktibādhā yatrāsti tad grāhyaṃ laukikaṃ yadi / gṛhyate vātaputrīyaṃ kiṃ na yuk‐
tyā na bādhitam // āgamārthāśrayā yuktir atyakṣeṣu na cetarā / tadarthasyāprati‐
ṣṭhānād yukter atra na sambhavaḥ //. “[Objection: Only] this [ordinary meaning] is
to be adopted for which there is no invalidation by reasoning [i.e., by the pramāṇas].
[Answer:] If [only] the ordinary [meaning that is not invalidated by reasoning is to
be adopted], why is the swindlerʼs [speech] not adopted [as long as it is] not invali‐
dated by reasoning? [Moreover, the kind of] reasoning [that operates] regarding im‐
perceptible [things] is [that] based on [a trairūpya derived from] the meaning of
scripture, and not the other [kind, i.e., the kind whose trairūpya proceeds by the
force of something real. But] since the meaning of this [scripture remains] unestab‐
lished, reasoning is impossible here.” For Dharmottaraʼs commentary on these two
stanzas, see PVinṬms 73a1–6/PVinṬTib Dze 229a6–b4 and PVinṬms 73a6–b3/
PVinṬTib Dze 229b4–230a1. The kind of reasoning alluded to here, i.e., the āgamā‐
rthāśrayā yuktiḥ, is of course the scripturally based inference (āgamāpekṣānumāna)
of PV 1.215/PVSV 109,1–3 (see Eltschinger 2007a: 105–109), PV 4.48–51 and 106–
108 (see Tillemans 2000: 78–82 and 147–153). See below, pp. 111–115. All the ele‐
ments that constitute such an inference, including the three characteristics of the
logical reason, are based on scripture (āgamasiddha, NBṬ 81,19; see also NBṬV
130,1–3 and PVV 410,18–411,1; see Moriyama, forthcoming). Note that PVin 2.44cd
resorts to the two successive uses or stages of reasoning that apply in the exegesis
of “ordinary” treatises (śāstra): “Indeed, the entire meaning of a treatise must be
determined by reasoning. And once the meaning of the treatise has been deter‐
mined [by reasoning], scripturally based reasoning proceeds by resorting to the
meaning [as it is] known from other passages [in the same treatise]” (PVinṬms
73b1–2/PVinṬTib Dze 229b6–7: yuktyā hi sarvaḥ śāstrārthoa vyavasthāpanīyaḥ /
vyavasthite ca śāstrārthe yuktirb āgamāśrayāc pradeśāntaraprasiddhārthāpekṣayā
pravartate /. a PVinṬTib bstan bcos thams cad kyi don [*sarvaśāstrārthaḥ]; b PVinṬTib
rjes su dpag pa [anumānam]; c PVinṬTib luṅ gi don la brten pa [*āgamārthāśrayaṃ]).
56 PVSV 164,24–176,16
the desired meaning], there must necessarily arise doubt for

those who do not see any restricting [factor].98 (PV 1.324
= PVin 2.45)
This is an intermediate stanza.
Therefore, Jaimini, when explaining [the Veda], attributes a single [de‐
sired] meaning to words whose [proper] interpretation (arthavibhāga)
is unknown, [a meaning] whose connection [with the word] is super‐
sensible [and] whose ascription is without foundation;99 [in this way,]
it is just his own conception which he formulates in this guise [i.e., in
the guise of the Veda itself].100 Thus, we do not see how he differs from
other founders of religions.101 For to say [that the Veda,] which [in fact]
lacks the capacity to express that [desired] meaning, [expresses it] by
[falsely]ascribing that [capacity] to it, amounts to nothing more than
one’s own assertion. Somebody who does that only brings to light his
own bad breeding,102 for
[Asked about the way to Pāṭaliputra,] one [person says,] “This
post says that this is the way,103” [whereas] another [answers,]
“[The post can’t say anything,] I myself say [that this is the
way].” One should inquire [whether there is any] difference be‐
tween the two. (PV 1.325 = PVin 2.46)
98 I.e., any pramāṇa establishing (sādhaka) the object to which this utterance is res‐
tricted (pratiniyataviṣaya; PVṬ Ñe D59b1/P70b2 = PVSVṬ 603,16–17). PVinṬms

74a1 explains: niyāmakaṃ hetum apaśyatām, and Manorathanandin (PVV 407,8–9):
anekārthasyaikavṛttiniyamakāraṇam apaśyatām.
99 I.e.,
devoid of any pramāṇa (niṣpramāṇaka; PVṬ Ñe D59b3/P70b4–5 ≈ PVSVṬ
603,20–21).
100 I.e., by referring to the Veda (vedopakṣepa), saying that it is the Veda that says so
(PVṬ Ñe D59b4/P70b6 = PVSVṬ 603,22–23).

101 Because the founders of religions (tīrthakara) say honestly (nirvyājam) that they
are speaking on their own (svayam; PVṬ Ñe D59b4/P70b6–7 ≈ PVSVṬ 603,23–24).

102 Because he is like a person who would indicate a wrong way (lam log pa) in order to
mislead a group of persons having lost their way (lam draṅ po; PVṬ Ñe D59b7–
60a1/P71a3–4).
103 I.e., indicates the way in the guise (vyājena) of a post that actually lacks the capacity
of speaking (PVṬ Ñe D60a2/P71a5–6 = PVSVṬ 604,12–13).

TRANSLATION 57
One indicates [the way] by ascribing [expressiveness] to the post

which is devoid of intention, exertion,104 and speech, and (vā) [another
indicates the way] independently.105 When it comes to following their
statements, there is no difference [between them] except for the slow‐
ness (pratipattimāndya) of the stupid [person who fails to understand
that the former is speaking “in the guise of the post”].
Moreover, if [a given Vedic word really] were restricted to a single
meaning, Jaimini would know it.
[But] where does PVSVthis veryPVSV restriction to illumining a
single meaning come from for a PVSVwordPVSV which is capable of
[referring to] any [meaning]? (PV 1.326ab ≈ PVin 2.47ab)
Indeed, there is no meaning of a word that is restricted [to it] by na‐
ture, because it is suitable for any [meaning],106 and because if it were
not suitable [for any meaning], humans could not [arbitrarily] apply
[it] or no longer apply [it] to something. For it would not be under
[their] control, since it would never lose this [nature of not being suit‐
able for any meaning].
107Or, by whom [are] supersensible [things such as semantic re‐
strictions] known in the absence of a statement of the speaker’s

intention? (PV 1.326cd = PVin 2.47cd)
104 Intention is defined as the resolve (cetanā) to do this or that; exertion, as the effort
(prayatna) that follows (pūrvaka) such an intention (PVṬ Ñe D60a3/P71a8 = PVSVṬ

604,16–17).
105 I.e., without resorting to the post (sthāṇunirapekṣa; PVṬ Ñe D60a4/P71b1 = PVSVṬ
604,19).
106 According to Dharmakīrti’s doctrine of yogyatā, any word is suitable (yogya) for any
meaning whatsoever. On yogyatā, see Tillemans 1997 and Eltschinger 2007a: 134–
138.
107 Dharmakīrti now provisionally accepts (bhavatu vā) that Vedic words are restricted
to a single meaning (ekārthaniyama; compare PVṬ Ñe D60b3/P72a3 with PVSVṬ

605,11–12). In PV 1.326cd, Dharmakīrti conflates two Mīmāṃsā doctrines into one
argument. Even if natural semantic restrictions exist, they are supersensible; now,
the Mīmāṃsaka rejects the possibility that humans perceive supersensible things
(PVSVṬ 605,12–13). Moreover, these semantic restrictions are authorless according
to Mīmāṃsā; therefore, they do not originate from an original speaker’s (vaktṛ)
intention (abhiprāya), the announcing (kathana) of which provides the only clue to
58 PVSV 164,24–176,16
In the case of a word pronounced by a person with a certain intention,

he might indeed at a certain time communicate108 that [intention] to a
certain [hearer]; therefore, one could very well understand the seman‐
tic restriction (arthaniyama) of [those] words [which are] preceded by
a speaker’s intention. But in the case of an authorless [word], how
could [this] semantic restriction be cognized, even if it exists? For [Ve‐
dic words] do not have a specific nature;109 or, if they did, it would fol‐
low, if it were perceptible, that it would be cognized by itself. [And] if it
were imperceptible, it could not be known by anybody, either. And
there is no difference [i.e., no specific nature at all], for words are never
[inherently] close to or remote from any meaning.110 Therefore,
the speaker’s intention is the cause of PVSVthesePVSV [words’111]
being restricted [to a single meaning, and] the convention [is
that which] reveals this [intention].112 [Since] an authorless
understanding the meaning (PVSVṬ 605,14–15). Unless otherwise stated, subse‐

quent occurrences of “(speaker’s) intention” translate the Sanskrit term vivakṣā.
108 I.e., teach that this or that is intended (vivakṣita) as the meaning (vācya) of the word
he uses (PVṬ Ñe D60b4–5/P72a5 ≈ PVSVṬ 605,17–18).

109 I.e., Vedic words do not have the nature of being restricted to a single desired (abhi‐
mata) meaning, the cognition (darśana) of which nature would bring about the
understanding (pratīti) of the desired meaning (iṣṭārtha; PVṬ Ñe D60b6/P72a7–8
≈ PVSVṬ 605,20–21; or, according to PVinṬms 74b1: the cognition of which would
enable one to know the word’s restriction to its meaning [yaddarśanād arthaniya‐
mapratītiḥ syāt]). Dharmakīrti uses a reductio ad absurdum (PVṬ Ñe D61a1/P72b2
≈ PVSVṬ 605,26) to disprove such a nature, first by considering the hypothesis that
this nature is perceptible (pratyakṣa), i.e., that its cognition does not depend on in‐
struction (upadeśanirapekṣa; PVṬ Ñe D60b7/P72b1 = PVSVṬ 605,23), and second,
by considering the hypothesis that it is imperceptible (apratyakṣa) to persons of
limited cognitive ability (arvāgdarśana; PVṬ Ñe D61a1/P72b2 = PVSVṬ 605,24–25;
PVṬ Ñe D60b6–7/P72a8–b1 = PVSVṬ 605,21–23).
110 See above, PVSV 167,11, and p. 41 n. 36.
111 I.e., of these words that are inherently suitable for any meaning, or naturally (sva‐
bhāvatas) the same (tulya) for all meanings (sarvārtha; PVṬ Ñe D61a2/P72b4
= PVSVṬ 605,28–29).
112 I.e., it is due to a convention that one understands that a certain person intended a
certain meaning (ayam arthas tena vivakṣita iti; PVṬ Ñe D61a2–3/P72b5 ≈ PVSVṬ
605,30–31). According to Dharmottara (PVinṬms 74b3–5, to be compared with
PVinṬTib Dze 230b8–231a1), a speaker “conventionalizes” (saṅketayati) that very
meaning (abhidheya) which he intends; those instructed (vyutpanna) at that time
TRANSLATION 59
[word] lacks this [intention], how does it have a single mean‐

ing? (PV 1.327 = PVin 2.48)
[It is] indeed due to the speaker’s intention [that] a word is restricted
to a [certain] meaning, not its nature, for the [word] is the same for
every [meaning] since it lacks a relation to anything. [Words] are not
even restricted to designating that to which [they do have] a relation,
otherwise all words would designate the vocal organs.113 Therefore, in
order to reveal the speaker’s intention, a convention is made, which is
defined as the communicating of an intention. But in the case of an au‐
thorless [word, there is] neither a speaker’s intention nor, [precisely]
because of the lack of anyone’s intention, a convention [for revealing
it].114 Thus [there is] neither restriction [to a certain meaning,] nor
[could there be] knowledge of it [if there were one].
If [one were to accept] a natural restriction, [then a Vedic word]
could not be connected again by the [speaker’s intention] with
another [meaning]. (PV 1.328ab = PVin 2.49ab)
If [one were to accept that] a word is joined with meanings due to its
very nature, independent[ly] of [any] convention, [then the reply] to
will instruct their own pupils (svaśiṣya) in turn, and those pupils their pupils, and so
on. In other words, whether it is a matter of a scripture that has been created (kṛtri‐
ma) or ordinary linguistic transactions (lokavyavahāra), one can know that there is
a semantic restriction due to an uninterrupted scholarly tradition (āmnāyāvicche‐
da). On convention, see above, pp. 13–15, and Eltschinger 2007a: 134–143.
113 Dharmakīrti accepts only two types of relation (pratibandha): identity (tādātmya)
and causality (tadutpatti). That words are not the nature (rūpa) of objects and that
objects are not the nature of words has been explained in PV 1.229cd and PVSV
114,25–27 (see Eltschinger 2007a: 253–254 and n. 159). But words are not the
cause of their meanings, either, no more than meanings are the cause of words. Ra‐
ther, words are related to the causal complex that gives rise to them, which includes
the speaker’s intention (which is why, according to passages such as PVSV 113,25–
114,3, 118,14–17 and 120,2–6 [Eltschinger 2007a: 140–142], PV 2.1c2–2 [Katsura
1984: 219], and PVin 2.1–7 [Steinkellner 1979: 73–73], one can infer the meaning or
cause, i.e., the speaker’s intention, from the word or effect), as well as the places of
articulation (sthāna) and organs (karaṇa) involved in the utterance of sounds (PVṬ
Ñe D61a5/P72b8–73a1 ≈ PVSVṬ 606,14–15). See also PV 1.336–337/PVSV 175,10–
24 below, pp. 73–75, and p. 73 n. 155.
114 See above, pp. 13–15.
60 PVSV 164,24–176,16
this has [already] been stated:115 Since [there is] no [real] relation,
[there is] no restriction. Moreover, if the relation between signified and
signifier were natural, [a word] could not be employed again arbitrar‐
ily according to the speaker’s intention.116
And a convention would be useless. (PV 1.328c = PVin 2.49c)
Indeed, a specific nature accessible to the senses does not need any‐
thing like an explanatory rule117 to be cognized (svapratīti), just like a
specific [instance of] blue, etc. On the contrary, [those things] the cog‐
nition of which depends on this [sort of thing] do not have the nature
of something real (vastusvabhāva), but are conventional, like the insig‐
nia of a king, etc.118 And that which is conventional cannot be restricted
by nature, for it exists [merely] according to [one’s] wish. 119Therefore,
115 See PVSV 172,19–20, above, p. 59.
116 I.e., the eye (cakṣus), which is naturally restricted (svabhāvato niyataḥ) to revealing
visible things/colours (rūpaprakāśana), cannot be used (ni√yuj) to reveal sounds

(śabda). Now, a word is used arbitrarily (yatheṣṭam), and therefore is not restricted
by nature (svabhāvaniyata) to revealing a certain meaning, like a hand‐gesture (has‐
tasañjñā; PVṬ Ñe D61b4–5/P73b3–4 ≈ PVSVṬ 606,29–31).
117 Explanation, PVṬ Ñe D61b6–7/P73b6–7 = PVSVṬ 607,11: paribhāṣā saṅketaḥ / ādi‐
śabdāt saṅketasmṛtyādiparigrahaḥ /. This is the only occurrence of paribhāṣā listed

in Ono et al. In Pāṇini a paribhāṣā is a rule that explains how rules of the grammar
are to be interpreted, hence a “metarule.” Thus, A 1.1.68: svaṃ rūpaṃ śabdasyāsaṃ‐
jñā. The word itself is generally to be understood when mentioned in a rule (not its
meaning), except in the case of a technical expression (such as gha, which stands for
the affixes tara and tama, A 1.1.22). There are some fifty such sūtras identified in A.
Insofar as a paribhāṣā clarifies the meaning of a rule or an element of a rule that
would not be evident otherwise it can be considered a statement of a convention.
For an overview of the use of paribhāṣās in Vyākaraṇa see Devasthali 1969 and
Cardona 1980: 167–170.
118 Additional instances of purely conventional things include hand‐gestures (hasta‐
sañjñā; PVṬ Ñe D62a1–2/P74a1–2 = PVSVṬ 607,15–16). See also above, p. 60 n.

116.
119 The Mīmāṃsaka opponent now claims that, although the specific nature (svabhāva‐
viśeṣa) of a Vedic word is naturally (nisargata eva) restricted to its meaning, this na‐
ture is revealed by a convention (saṅketa; PVṬ Ñe D62a3/P74a3–4 = PVSVṬ
607,19–21). This claim comes very close to the Mīmāṃsaka and Vaiyākaraṇa (but at
least partly also Vaibhāṣika) doctrine according to which the relation between word
and meaning, though real and permanent (be it just sāṃvyavahārikanitya), must be
learned by a convention. Note that for Kumārila, it is not always learned from the di‐
TRANSLATION 61
[when it comes to] the specific nature [which is supposedly restricted

to a single meaning] –
how could it be that [it] is necessarily revealed (PV 1.328d
= PVin 2.49d)
by a convention? There is no constraint (noparodho ’sti) that this con‐
vention, which exists [merely] according to one’s wish, be made with
regard to the [desired meaning] only [and] not with regard to another
one. And there is no necessity that this [convention], which is made by
human beings according to their wish, [should] reveal only that nature
[which is restricted to a certain meaning and] not another one.
In this [convention], where there is arbitrariness, how can
there be necessity? Thus, the convention cannot bring to light
just a desired capacity [of the Vedic word and no other]. (PV
1.329 = PVin 2.50)
This is an intermediate stanza.120
121For which reason, probably (kila)122 [ – that is, in light of all
of the above – ] another [philosopher] has said that just as a

[Vedic] sentence,123 such as [the one to the effect that] fire re‐
rect statement of a convention, but also from the observation of linguistic behavior.
See PV 1.227cd/PVSV 113,14–23 (Eltschinger 2007a: 246–248) and Eltschinger
2007a: 122–134.
120 This closes Dharmakīrti’s examination of the doctrine of authorlessness as the
definition of scripture (apauruṣeyatvam āgamalakṣaṇam, PVṬ Ñe D62a6–7/P74b1
= PVSVṬ 607,29), which began at PVSV 112,6. See above, pp. 18–21.
121 In PV 1.330–335/PVSV 173,16–175,10, Dharmakīrti criticizes another definition of
scripture (āgamalakṣaṇa), viz., the reliability of one part (ekadeśāvisaṃvādana) of

the scripture as a criterion of its overall truth (PVṬ Ñe D62a7/P74b1–2 ≈ PVSVṬ
608,12). On the identity of his mysterious opponent, see above, pp. 18–21. For a
(short) parallel passage, see PVin 2 72,10–11 and Steinkellner 1979: 77–78.
122 According to PVṬ Ñe D62a7/P74b2 = PVSVṬ 608,13, the word kila expresses (dyo‐
taka) lack of intention (anabhiprāya), i.e., the apauruṣeyatvavādin’s being forced to

change his definition of scripture. This is, at least, the way PVṬ Ñe D62b2/P74b5–6
≈ PVSVṬ 608,17 account for PVSV 173,16 (parityajya): yathoktadoṣopahatatvāt,
“reeling from the aforementioned errors.”
123 For an earlier occurrence of the Vedic statement agnir himasya bheṣajam (“Fire is a
remedy for cold”), see PVSV 152,1–4, where the context is similar; see also PVSV
173,18–19 and PVṬ Ñe D62b1/P74b3–4 = PVSVṬ 608,13–14. The statement can be
62 PVSV 164,24–176,16
moves cold, is true, any other one [, such as “One should per‐
form the Agnihotra,”] is like this [i.e., true], because it is a part
of the [same] Veda.124 (PV 1.330)
But another [theorist], renouncing the definition of scripture as [that
which is] authorless [because of a similar criticism, and] wishing to
prove the authority of the Veda in another way, says, “[Those] Vedic
sentences for which there is no cognition [on the part of Buddhists as
being true] are [in fact] true, because they are part of the [same] Veda,
just like the sentence ‘Fire is the remedy for cold,’ etc.” The [argument]
of this [adversary]
is [an inference of the type known as] śeṣavat, like [the infer‐
ence that something has a certain] taste [as other fruits] from
having the same color and like [the inference that something is]
cooked [from being] in one [and the same] pot.125 This type of
[inference] has been rejected by the Logician126 because it devi‐
ates [from the property‐to‐be‐proved]. (PV 1.331)
traced back, at least in part, to TaitS 7.4.18.2 (where agniḥ is not mentioned but
might be the answer to the riddle [kíṃ svid dhimásya bheṣajám?] posed by the text –
which is incidentally the interpretation of BaudhŚSū 15.28 [agnir himasya
bheṣajam]) and ŚB 13.2.6.12, and to Śrautasūtra literature (see e.g. ĀśvŚSū 10.9/2b
and ŚāṅkhŚSū 16.5.4). The statement is quoted in NV 255,1 ad NSū 2.1.60. As an
additional example of such trivially true Vedic statements, PVṬ Ñe D62b5/P75a1–2
= PVSVṬ 608,22–23 quote: dvādaśa māsāḥ saṃvatsaraḥ, “Twelve months are a
year,” which already appears in NV 254,1. The source of this statement must be
KāṭhS 19.8, 19.9, 19.9, and 29.8. Note, however, that the KāṭhS reads dvādaśamāsaḥ
saṃvatsaraḥ, “A year has twelve months.” In ŚV codanā 121–122ab, Kumārila also
criticises the ekadeśāvisaṃvāda argument; see Kataoka 2011a, n. 377.
124 In this argument, īdṛśam – “is like this,” i.e., true – is the property to be proved (sā‐
dhya, or *sādhanaphala; PVṬ Ñe D62b1/P74b3 ≈ PVSVṬ 608,14–15), satyaṃ yathā‐

gniḥ śītanodano vākyam is the example (*dṛṣṭānta; PVṬ Ñe D62b1–2/P74b4), vedai‐
kadeśatvāt is the reason (hetu; PVṬ Ñe D62a7/P74b2–3 = PVSVṬ 608,15). For other
formulations of the argument, see PVSV 175,2–4 and PVṬ Ñe D67a6–7/P81a3–4
≈ PVSVṬ 614,11–13, as well as PVinṬTib Dze 231a1–b3 (PVinṬms 75a is unfortunate‐
ly missing!).
125 On the second example, see PV 1.13d = PVin 2.65d, PVSV 10,15–17 ≈ PVin 2 92,4–6
and Steinkellner 1979: 114 and n. 434.

126 I.e.,
by Ācārya Dignāga in his Pramāṇasamuccaya (PVṬ Ñe D63a1–2/P75a7–8
≈ PVSVṬ 608,25–26) when criticizing the deviating character (vyabhicāra) inherent
in the Naiyāyikas’ śeṣavadanumāna (PVṬ Ñe D63a3/P75b1–2 = PVSVṬ 609,7–8).
TRANSLATION 63
This kind of inference was declared to be not [really] a proof by the

master [Dignāga] himself in pointing out the deviating character of the
Naiyāyikas’ śeṣavat‐inference, like the [so‐called] proof that fruit [one
has not tasted] has the same taste [as fruit one has tasted] because it
has the same color, and the [so‐called] proof that rice grains one has
not observed are cooked, like those which one has observed, because
they are in one [and the same] vessel. And the manner in which this
[śeṣavat‐inference] is not a proof has [already] been stated earlier.127
128And [true,] we have stated this definition of scripture [too].129 How‐
See PS 2.28b and PSVK 115b2–4, PSVV 33a1–2 ad loc. in Horst Lasic’s provisional
Sanskrit reconstruction: śeṣavaty api ced vatiḥ / śeṣavaty api ced vatiḥ kriyate –
śeṣeṇa pratyakṣeṇa tulyo viṣayo yasya, taj jñānaṃ śeṣavad iti, tad api vyabhicārān na
pramāṇam. na hy avaśyaṃ rūpasāmy(amātr)ād rasādi tulyaṃ bhavati. evaṃ śeṣavaty
api vatir na yujyate. On Dharmakīrti’s definition of the śeṣavadanumāna, see below,
n. 127, pp. 102–105, and Steinkellner 1979: 114–116 n. 436.
127 I.e., in PV 1.14 = PVin 2.66 and PVSV 10,19–23 ≈ PVin 2 92,7–12 (PVṬ Ñe D63a5–
6/P75b5 = PVSVṬ 609,11–12; see Steinkellner 1979: 114–116): kiṃ punar etac che‐
ṣavat / yasyādarśanamātreṇa vyatirekaḥ pradarśyate / tasya saṃśayahetutvāc che‐
ṣavat tad udāhṛtam // sa tasya vyatireko na niścita iti vipakṣe vṛttir āśaṅkyeta / vy‐
atirekāsādhanasyādarśanamātrasya saṃśayahetutvāt / na sarvānupalabdhir gami‐
kā /. “But what does this [inference] ‘with a remainder’ consist in? The [reason]
whose absence [in the dissimilar instances] is established by mere non‐cognition is
named ‘with a remainder,’ because it is a cause of doubt. Since the absence of this
[reason in the dissimilar instances] is not ascertained, one may suspect that it oc‐
curs in the dissimilar instances, for mere non‐cognition, which does not [success‐
fully] prove absence, is a cause of doubt. [Indeed,] not every [kind of] non‐percep‐
tion is conclusive.” In light of this definition, the parallel discussion in PVin 2 72,10–
11 makes it very clear why the argument of Dharmakīrti’s opponent must be
termed śeṣavat: ekadeśāvisaṃvādanam apy āgamalakṣaṇam āhuḥ / tad vipakṣe ’dar‐
śanamātrād avyabhicārāsiddhyāniścitārtham /. “[Certain theorists] claim that the
reliability of one part [of a scripture] is the definition of scripture. [But] since non‐
deviation [can]not [be] established through the mere non‐cognition [of the reason]
in the dissimilar instances, this [argument] is uncertain.” See Steinkellner 1979: 77–
78 and nn. 251–253.
128 The opponent now objects that Dignāga, in PS 2.5ab, has also made use of partial re‐
liability (ekadeśāvisaṃvādana) as a definition of scripture (PVṬ Ñe D63a6/P75b5–7

≈ PVSVṬ 609,12–16). On PS 2.5ab and Dharmakīrti’s interpretation of it, see below,
pp. 85–100.
129 I.e., Dharmakīrti has indeed at least implied that the definition of scripture consists
in partial reliability (ekadeśāvisaṃvādirūpam āgamalakṣaṇam) in his explanation of

PS 2.5ab (see above, n. 128), stating that “this human being cannot live without re‐
64 PVSV 164,24–176,16
ever, this [is justified only] if, for every object capable of being exam‐
ined, there is correctness (viśuddhi) of positive and negative assertions
by appropriate means of valid cognition.130 [And] even if there is no
necessary relation between words and [their] meanings [, which would
ensure the validity of scripture], it is better that a [person] act in [a
state of] doubt [when it comes to matters relating to worldly prospe‐
rity and salvation]; for [scripture] may occasionally be reliable in this
case.131 But in the opposite case,132 a human being who has observed a
conflict with a means of valid cognition would not [be justified] in act‐
sorting to the authority of scripture” (PVSV 108,2–3: nāyaṃ puruṣo ’nāśrityāgama‐

prāmāṇyam āsitum samarthaḥ; PVṬ Ñe D63a6–7/P75b7–8 ≈ PVSVṬ 609,16–19). On
PVSV 108,2–3 and Dharmakīrti’s interpretation of PS 2.5ab, see below, pp. 85–100,
and above, pp. 18–21. In PVSV 173,26–174,2, Dharmakīrti is going to show that his
own and his opponent’s accounts of ekadeśāvisaṃvādana are methodologically very
different; for Dharmakīrti’s version is not based on the truth of scripture about a
single trivial matter (atyantaprasiddhaikaviṣayasatyatāśraya; PVṬ Ñe D63a6–7/
P75b7–8 ≈ PVSVṬ 609,16–19).
130 I.e., according to PV 1.215 and PVSV 108,16–109,4, by perception and the two types
of inference. The establishment of a positive assertion (vidhisiddhi) functions as fol‐

lows: That which is recognized (abhimata) as perceptible (pratyakṣaviṣaya) in the
treatise (śāstra) under consideration must indeed be perceptible, e.g., cognition
(buddhi) as understood in Buddhist doctrine (bauddhasiddhānte); that which is re‐
cognized as an object of an inference based on the reality of entities (vastubalāyā‐
tānumānaviṣaya; āgamānapekṣānumānaviṣaya) must indeed be inferable by an infe‐
rence of that type, e.g., the Truth of Suffering (duḥkhasatya); that which is recog‐
nized as an object of a scripturally based inference (āgamāpekṣānumāna; see
Eltschinger 2007a: 105–109, and Moriyama, forthcoming; see also above, p. 55 n.
97) must indeed be inferable by an inference of that type. The establishment of a
negation (pratiṣedhasiddhi) functions as follows: That which is recognized as not
being the object of perception, an “objective” inference, or a “scriptural” inference
must be imperceptible, non‐“objectively” inferable, and non‐“scripturally” inferable.
(PVṬ Ñe D63b1–4/P76a2–6 ≈ PVSVṬ 609,22–27.) See below, pp. 86–87. For a trans‐
lation of PV 1.215/PVSV 108,16–109,4, see Yaita 1987: 7–8, Dunne 2004: 362–363,
and Eltschinger 2007a: 221–224.
131 I.e., in the case of scripture that has not been observed to err (adṛṣṭavyabhicāra) in
regard to empirically verifiable matters (PVṬ Ñe D63b5/P76a7–8 ≈ PVSVṬ 609,29–

30). See PV 1.213 and PVSV 107,19–108,6 (Yaita 1987: 6–7, Dunne 2004: 361,
Eltschinger 2007a: 217–220) and above, pp. 43–44 n. 49.
132 I.e., in the case of a scripture that can be proved not to possess the aforementioned
reliability in regard to empirically verifiable matters, i.e., the necessary correctness

(*viśuddhi; PVṬ Ñe D63b5–6/P76a8–b2).
TRANSLATION 65
ing. But he who, pointing out that the statement (abhidhāna) that fire –
an object [accessible] to ordinary persons – has the capacity to elimi‐
nate cold, is veracious, declares the entire treatise [known as the Veda]
to be veracious, though it is mostly incorrect due to [its] being contra‐
dicted by the means of valid cognition even in regard to something that
can be determined – [ such a person’s audacity knows no bounds].
[The Veda] says that a permanent soul is the agent [of action],
[indeed] that there are permanent entities, [and] that super‐
sensible [things] are sensible. [It declares] a wrong cause, a
wrong duration as well as a [wrong] cessation of entities, or
[puts forward yet] other [things] whose possibility133 (gocara)
is excluded by the two means of valid cognition or contradicted
by inference based on scripture. He who would pretend that
[such a treatise] is veracious without having set aside [its] con‐
tradictions and without exhibiting the purpose of the treatise,
would surpass an unchaste woman in audacity.134 (PV 1.332–
334)
The Veda declares that a soul (puṃs), which neither loses its former
nature nor assumes a new one, [i.e., which is permanent] is succes‐
sively the agent of [good and bad] deeds and the experiencer of the
fruits of [those] deeds. [It is supposedly the experiencer] due to being
the inherence cause (samavāyikāraṇa) [of pleasant and unpleasant
sensations,] and [the agent] due to assuming the supervision (adhi‐
ṣṭhāna) [of bodily actions], etc.135 And this has repeatedly been shown
133 gocara is explained as avakāśa in PVṬ Ñe D64a5/P77a2 ≈ PVSVṬ 610,22–23.
134 For a parallel passage, see TSK 2775/TSŚ 2774 and TSPK 736,4–18/TSPŚ 892,13–
893,10.
135 Although Dharmakīrti’s intention in PV 1.332a and PVSV 174,14–16 remains un‐
clear, these two passages seem to target primarily Vaiśeṣika and Naiyāyika ideas re‐
garding the self. (1) The concept of inherence cause (samavāyikāraṇa) is at home in
Vaiśeṣika, as is the description of the self as a samavāyikāraṇa. On the samavāyi‐
kāraṇa in general, see VSū 1.1.14 and DPŚ 91 (Miyamoto 2007: 26); on samavāya
and samavāyikāraṇa in connection with the self, see VSū 10.1 (ātmasamavāyaḥ su‐
khaduḥkhayoḥ) and especially DPŚ 10 (Miyamoto 2007: 10: *ka ātmā? yo buddhisu‐
khaduḥkhecchādveṣaprayatnasaṃskāradharmādharmāṇāṃ samavāyikāraṇaṃ jñā‐
notpādanalakṣaṇaṃ sa evātmā /. “What is the self? The self is that which is the inhe‐
rence cause of cognition, pleasure, pain, desire, aversion, effort, residual traces, me‐
rit, and demerit, [and which is] characterized as producing cognition”). Here is the
66 PVSV 164,24–176,16
Vaiśeṣika’s classic statement regarding the inference of the self (VSū 3.2.4): prāṇā‐
pāṇanimeṣonmeṣajīvanamanogatīndriyāntaravikārāḥ sukhaduḥkhe icchādveṣau pra‐
yatnaś cety ātmaliṅgāni /. “The marks of the self are breathing in and out, closing
and opening [the eyes], life, the movements of the mental organ, and the transfor‐
mations of the other sense organs; pleasure and pain; desire and aversion; and ef‐
fort.” (On this sūtra see, e.g., Preisendanz 1994: 263–274 and Oetke 1988: 319–322
and 334–340.) Now, Praśastapāda seems to divide these inferential marks into two
categories: those that allow one to infer the self as a supervisor (adhiṣṭhātṛ) and
those that are the marks of a property‐possessor (guṇin), i.e., an inherence cause.
Here are Praśastapāda’s remarks as regards the first inference (PDhS 15,8–10):
śarīrasamavāyinībhyāṃ ca hitāhitaprāptiparihārayogyābhyāṃ pravṛttinivṛttibhyāṃ
rathakarmaṇā sārathivat prayatnavān vigrahasyādhiṣṭhātānumīyate, prāṇādibhiś
ceti. “[The self] is inferred [as] the body’s supervisor endowed with effort from the
undertaking [of action] and abstaining [from it], which inhere in the body [and] are
suitable for obtaining what is useful and rejecting what is harmful, just as a chariot‐
eer [is inferred] from the action of the chariot. And [the self is also inferred thus]
through breath, etc.” (See Oetke 1988: 279, arguments no. 3 and 4.) And here are his
views regarding the second inference (PDhS 16,3–4): sukhaduḥkhecchādveṣa‐
prayatnaiś ca guṇair guṇy anumīyate. “And [the self] is inferred [as] a property‐
possessor from the properties that are pleasure, pain, desire, aversion, and effort.”
(See Oetke 1988: 280, argument no. 9; to be compared with NSū 1.1.10: icchādveṣa‐
prayatnasukhaduḥkhajñānāny ātmano liṅgam /. “The marks of the self are desire,
aversion, effort, pleasure, pain, [and] cognitions.”) Thus, whereas the first inference
(→adhiṣṭhātṛ) seems to be concerned with the self as an agent (kartṛ) inasmuch as it
presides over bodily actions, the second one (→guṇin, i.e. samavāyikāraṇa; note VSū
1.1.5: buddhayaḥ sukhaduḥkhe icchādveṣau prayatnaś ca guṇāḥ /, and VSū 1.1.14:
kriyāvad guṇavat samavāyikāraṇam iti dravyalakṣaṇam /) seems to deal with the
self as an experiencer (bhoktṛ) inasmuch as it experiences pleasure and pain, but
also – and this has eschatological consequences – merit and demerit (dharmādhar‐
ma, see PDhS 16,7–8: tasya guṇā buddhisukhaduḥkhecchādveṣaprayatnadharmā‐
dharma°, and DPŚ 10, quoted above). In other words, the self is an agent in that it
supervises bodily action and an experiencer in that it is the inherence cause of
pleasant and unpleasant sensations as well as merit and demerit. (2) While explain‐
ing PVSV 174,14–16, Karṇakagomin quotes three half‐verses that recur at the
beginning of Śāntarakṣita’s examination of the Vaiśeṣikas’ and Naiyāyikas’ concep‐
tion of the self as well as in Prajñākaramati’s BCAP. Here are TSK/Ś 171–173: anye
punar ihātmānam icchādīnāṃ samāśrayam / svato ’cidrūpam icchanti nityaṃ sarva‐
gataṃ tathā // śubhāśubhānāṃ kartāraṃ karmaṇāṃ tatphalasya ca / bhoktāraṃ
cetanāyogāc cetanaṃ na svarūpataḥ // jñānayatnādisambandhaḥ kartṛtvaṃ tasya
bhaṇyate / sukhaduḥkhādisaṃvittisamavāyas tu bhoktṛtā //. “However others admit
that the self is the substrate (samāśraya) of desire, etc., while not consisting of con‐
sciousness by itself, that it is permanent and omnipresent, that it is the agent of acts
[both] good and bad, and the experiencer of the fruit of these [acts and, finally,] that
it is conscious due to its association with consciousness, but not by nature. Its con‐
TRANSLATION 67
nection with cognition, effort, etc., is called its agency, whereas the inherence of the
awareness of pleasure, pain, and so forth [in the self] is [its] being an experiencer.”
Prajñākaramati introduces the two stanzas of the TS he quotes (= TSK/Ś 171–172) as
follows (BCAPSh 327,7–13/BCAPLVP: 295,17–23): tatra naiyāyikās tāvan nityaṃa
sarvagataṃ pratiprāṇibhinnam acetanaṃb cetanāyogāt tu cetanaṃ sukhādiguṇā‐
dhāraṃ śubhāśubhkarmakartāraṃ tatphalopabhoktāraṃ paralokinaṃ cātmānam
icchanti / naiyāyikavad vaiśeṣikā api / tad uktam /. a nityaṃ BCAPLVP: nitya‐ BCAPSh;
b pratiprāṇibhinnam acetanaṃ BCAP
LVP: pratiprāṇibhinnacetanaṃ BCAPSh. “In this
regard, first of all the Naiyāyikas admit that the self is permanent, omnipresent, dis‐
tinct according to each living being, not conscious [by nature] and yet conscious be‐
cause of [its] association with consciousness; [it is] the substrate (ādhāra) of the
properties pleasure, etc., the agent of actions [whether] good or bad, the experi‐
encer of the fruit of these [actions], and it transmigrates to another world. Like the
Naiyāyikas, the Vaiśeṣikas too [admit this]; this has been said in [the following two
stanzas].” As we can see, both Śāntarakṣita and Prajñākaramati hold these ideas to
be common to Naiyāyikas and Vaiśeṣikas. These conceptions form the background
of Karṇakagomin’s explanation (PVSVṬ 611,7–12) of PVSV 174,14–16: kena
prakāreṇa bhoktā kartā cety āha / pūrvakarmajanitasukhaduḥkhādisaṃvittiṃ prati
samavāyikāraṇabhāvenātmā karmaphalānāma bhoktāb / tad uktam – sukhaduḥkhā‐
disaṃvittisamavāyas tu bhoktṛteti / śubhāśubhakarmakaraṇe jñānaprayatnādikaṃ
praty adhiṣṭhānabhāvenātmā karmaṇāṃc kartā / tad uktam – jñānayatnādisamban‐
dhaḥd kartṛtvaṃ tasya bhaṇyata iti / ādigrahaṇād / jaḍarūpasyāpy ātmanaś cetanā‐
yogena bhoktṛtvaṃ gṛhyate / tad uktam – bhoktā ca cetanāyogāc cetanaṃ na sva‐
rūpata iti /. a karmaphalānām em.: karma phalānām ed.; b bhoktā em.: bhoktāḥ ed.; c
karmaṇāṃ em.: karmaṇā ed.; d jñānayatnādisambandhaḥ em.: jñānādisambandhaḥ
ed. “[In order to explain] in what way [the self] is an experiencer and an agent
[Dharmakīrti] says [that] the self is the experiencer of the fruits of actions insofar as
it is the inherence cause with respect to the awareness of pleasure and pain, etc.,
which result from previous actions. This has been said in [the following stanza]:
‘Whereas the inherence of the awareness of pleasure, pain, and so on [in the self] is
[its] being an experiencer’ (= TSK/Ś 173cd, above). [And] the self is the agent (kartṛ)
of actions insofar as it assumes the supervision (adhiṣṭhāna) of cognition, effort, and
so on in the performance of acts that are either good or bad. This has been said in
[the following half‐verse]: ‘Its connection with cognition, effort, etc., is called its
agency’ (= TSK/Ś 173ab, above). And because of the word ‘etc.’, [we should]
understand that the self is an experiencer although its nature is insentient, due to its
association with consciousness. This has been said in [the following half‐verse]:
‘And [that] it is the experiencer [and, finally,] that it is conscious due to its
association with consciousness, but not by nature (= TSK/Ś 172cd, above).’” Note,
however, that Śākyabuddhi (PVṬ Ñe D64b3–6/P77b1–5) favors another explana‐
tion: de la sems las byuṅ ba bsod nams daṅ bsod nams ma yin pa de’i ’phrod pa ’du ba
can gyi rgyu ñid kyis byed pa po yin te / blo bdag la ’du ba daṅ ldan pa ñid kyi phyir
ro // lus daṅ ṅag gi las byin gyis brlabs (D: P rlobs) pa’i ṅo bo byed pa po yin te / bdag
ñid kyis brlabs pa’i lus daṅ ṅag dag ni raṅ gi bya ba rnams la ’jug pa’i phyir ro // … de
68 PVSV 164,24–176,16
to be incorrect.136 And [the Veda also states] the permanence of certain

entities, [which] is incorrect, because a non‐momentary [entity] vio‐
lates the criterion (dharma) of something real.137 [Moreover, the Veda
bźin du las kyi ’bras bu bde ba daṅ sdug bsṅal dag gi ’phrod pa ’du ba’i rgyu ñid kyi
phyir za ba po yin te / bde ba daṅ sdug bsṅal ba dag ni bdag daṅ ’du ba daṅ ldan pa
ñid kyi phyir ro // bde ba (P: D lus) daṅ sdug bsṅal sgrub par byed pa mṅon par ’dod
pa daṅ / mṅon par ’dod pa ma yin pa’i yul rnams byin gyis brlabs pa’i ṅo bo’i phyir za
ba po yin no. “Here, [the self] is an agent (*kartṛ) inasmuch as it is the inherence
cause (*samavāyikāraṇatvena) of mental [factors] (*caitta, *caitasika) such as merit
and demerit, [and this] because cognition (*buddhi) inheres (*samavāyin?) in the
self. [And the self] is an agent in assuming the supervision (*adhiṣṭhānabhāvena) of
corporeal and verbal actions (*kāyavākkarman), because [, insofar as they are]
supervised by the self, body (*kāya) and speech (*vāc) proceed (*[pra]vṛtti?) to
their own [characteristic] actions (*svakriyā?) … Similarly, [the self] is an experien‐
cer (*bhoktṛ) because it is the inherence cause of the pleasant and unpleasant re‐
sults of actions (*sukhaduḥkhakarmaphala), [i.e.,] because pleasure and pain (*su‐
khaduḥkha) inhere (*samavāyin?) in the self. [The self] is an experiencer because it
assumes the supervision of [psycho‐physical activities with regard to those] desira‐
ble and undesirable things (*viṣaya) that are the means of realizing (*sādhana)
pleasure and pain.” Thus, according to Karṇakagomin the self’s being an agent is to
be explained in terms of adhiṣṭhāna, whereas its being an experiencer is to be ac‐
counted for in terms of samavāyikāraṇa. But according to Śākyabuddhi, the self’s
being an agent and the self’s being an experiencer are both to be explained in terms
of both samavāyikāraṇa and adhiṣṭhāna. (Although he does not allude to adhiṣṭhāna
in this context, Uddyotakara too accounts for both kartṛtva and bhoktṛtva in terms
of samavāya in NV 337,14–15 ad NSū 3.1.6: kiṃ punar idaṃ kartṛtvaṃ kiṃ vā
bhoktṛtvam iti jñānacikīrṣāprayatnānāṃ samavāyaḥ kartṛtvaṃ sukhaduḥkhasaṃvit‐
samavāyo bhoktṛtvam. “[But] how (kim) [to account for the self’s] being an agent,
and (vā) how [to account for its] being an experiencer? The inherence of cognition,
desire to act, and effort is [the self’s] being an agent; the inherence of the awareness
of pleasure and pain is [the self’s] being an experiencer.” For other statements re‐
garding the self as an agent due to its adhiṣṭhāna, see SK 17 and ŚV ātmavāda 76.
136 I.e., by Dharmakīrti (śāstrakāra) himself while demonstrating the impossibility of
causality (kāryakāraṇabhāvāsambhava) for permanent things, and by his coreli‐

gionists (*svayūthya, most probably Vasubandhu) in their own treatises (*svaśāstra;
compare PVṬ Ñe D64b6/P77b6–7 and PVSVṬ 611,17). For references to Dharma‐
kīrti’s PVSV and PVin 2 as well as Vasubandhu’s KSP and more ancient Yogācāra and
Sautrāntika sources, see Yoshimizu 1999. See also below, n. 137.
137 I.e., causal efficiency (arthakriyāvirodha). According to Dharmakīrti (see, e.g., PV
1.166ab and PVSV 84,5–6) to be a real entity (vastu) is to be causally efficient (ar‐
thakriyāsamartha). Something non‐momentary (akṣaṇika) could not have any cau‐
sal efficiency (because it could not produce an effect either successively or at once
TRANSLATION 69
says that things which are] indeed strictly imperceptible, such as uni‐
versals,138 are perceptible, and [declares] a wrong origination, dura‐
tion, and cessation of entities: [Indeed, it proclaims that something]
which initially is not an agent [and which, being permanent,] cannot
receive [any] new property (viśeṣa), can generate [an effect] through
dependence on [something] else;139 [that something] whose nature is
no [longer] to be brought about since it has [already] been completed
[by its own causes can] last by virtue of a substratum [upon which it
depends];140 and [that entities] perish due to a cause,141 etc. [The Veda
states] yet other [things] which are contrary to what is established by
perception and inference and are negated by inference based on scrip‐
ture, such as the capacity of the Agnihotra and [ablutions] to purify one
of sin, etc.142 [He who,] failing to set aside the contradictions of the
[kramayaugapadya]), hence it could not be something real (PVṬ Ñe D64b7–65a1/

P77b7–78a1 ≈ PVSVṬ 611,19–21). See Yoshimizu 1999.
138 For additional examples of (pseudo‐)things erroneously held to be perceptible by
Sāṅkhya and Vaiśeṣika authors, see PVSV 108,22–24 and Yaita 1987: 8, Dunne
2004: 362–363, Eltschinger 2007a: 222–223 and n. 25; see also PVṬ Ñe D65a1/
P78a1–2 = PVSVṬ 611,21–22, and above, p. 20.
139 I.e., a cooperating factor (sahakārin; PVṬ Ñe D65a3/P78a4–5 = PVSVṬ 611,26).
140 That which depends on nothing (sarvanirāśaṃsasya; thams cad la ltos pa med par
grub pa) cannot last (sthāna) by force of something else (anyabalena; compare PVṬ
Ñe D65a5/P78a7–8 with PVSVṬ 611,29–30). See SPV 270,5–8 on SP 3cd (saṃś ca
sarvanirāśaṃso bhāvaḥ katham apekṣate; see Vibh 410 n. 8): yod na yaṅ kun la rag
ma las te / raṅ gi ṅo bo thams cad skyes pa bltos pa med pa’i dṅos po bltos pa gaṅ gis
na ’brel par ’gyur ba ji ltar bltos pa yin /. “Wenn [das betreffende Ding] dagegen et‐
was Seiendes ist, durch welche Bedingtheit kann dann ein vollkommen unabhängi‐
ges Ding, d.h. (ein Ding), das seinem ganzen Wesen nach entstanden und daher
nicht bedingt ist, verbunden sein; wieso kann es bedingt sein?” Translation Frau‐
wallner 1934: 284.
141 I.e., due to a cause of destruction (vināśahetu; PVṬ Ñe D65a6/P78a8–b1 = PVSVṬ
611,31). According to Dharmakīrti destruction is without a cause (nirhetukatvaṃ

vināśasya; PVSVṬ 612,6) or incompatible with the fact of having a cause (rgyu daṅ
ldan pa ñid du ’gal ba; PVṬ Ñe D65a6/P78b1). On Dharmakīrti’s vināśitvānumāna,
see Sakai 2011.
142 This refers back to PVSV 109,1–3 (see Yaita 1987: 8, Dunne 2004: 363, Eltschinger
2007a: 105–109 and 223, and below, pp. 77–78 nn. 172 and 174). If a given treatise
teaches that sin or demerit (adharma) consists in defilements such as desire (rāgā‐
dirūpa) and the actions born of them (tatprabhavaṃ karma), it cannot prescribe
practices such as the agnihotra and ablutions (tīrthasnāna) as means of eliminating
70 PVSV 164,24–176,16
pramāṇas in the entire body of the treatise143 [and] failing to exhibit

[that it has] the properties of a [sound] treatise,144 viz., [its] expressing
[internal] consistency, appropriate means, and a human purpose,145
demerit or increasing merit (dharmavṛddhi; PVṬ Ñe D65b2/P78b6, to be compared

with PVSVṬ 612,10). For such practices, since they do not counteract the causes
(nidāna) of demerit (especially nescience [avidyā]), cannot remove it; and since they
do not consist in the the absence of greed (alobha) or the actions born of it, they
cannot increase merit (PVṬ Ñe D65b3–5/P78b6–79a2 ≈ PVSVṬ 612,11–15). Look‐
ing for this kind of internal contradiction (pūrvāparavirodhacintā) is the only thing
a cognitively limited person (arvāgdarśin) can do to assess the statements regard‐
ing supersensible matters of the treatise under scrutiny. This is what Dharmakīrti
calls an “inference based on scripture” (āgamāpekṣānumāna, see above, p. 55 n. 97).
On Dharmakīrti’s threefold analysis, see above, p. 64 n. 130, and below, pp. 86–87
and n. 15.
143 I.e., in regard to matters (vastu) taught by the treatise, which are threefold (trivi‐
dha): those that bear on perceptible things, those that pertain to things “objectively”
inferable, and those concerning things “scripturally” inferable (PVṬ Ñe D65b5/
P79a2–3 = PVSVṬ 612,16–17). Dharmakīrti’s notion of pratisamādhāna (PVSV
174,25; see also PVṬ Ñe D64a7/P77a5–6 = PVSVṬ 610,27–28 and PVṬ Ñe D65b5–6/
P79a3 = PVSVṬ 612,17) is likely to echo the VY’s codyaparihāra. According to Vasu‐
bandhu an apt commentator of the Buddhist sūtras (sūtrārthavādin), after having
indicated the purpose (prayojana) of the sūtra, provided a summary (piṇḍārtha) of
it, explained the meaning of the words (padārtha), and explicated the connection
(anusandhi) of its various parts, should formulate hypothetical objections and refute
them (codyaparihāra; see Cabezón 1992: 237–238 n. 16, Skilling 2000: 318–319,
and Verhagen 2008). What Dharmakīrti’s opponent fails to do is to refute the objec‐
tions raised against the Veda by the application of pramāṇas.
144 I.e., the properties which are causes of one’s engagement with a treatise (śāstre pra‐
vṛttyaṅgabhūtā dharmāḥ; PVṬ Ñe D65b6/P79a3–4 = PVSVṬ 612,17–18). These

three properties (see below, n. 145) constitute the purpose of the treatise (śāstrār‐
tha) alluded to in PV 1.334b (PVṬ Ñe D64a7–b1/P77a5 = PVSVṬ 610,28–29).
145 This refers back to PV 1.214 and PVSV 108,6–16; see Yaita 1987: 7, Dunne 2004:
361–362, and Eltschinger 2007a: 102–104 and 220–221. Here, Dharmakīrti formu‐
lates the three properties that a treatise must possess in order to be qualified for
the evaluation of its reliability (< parīkṣādhikṛta), namely, (1) internal consistency
(sambandha, sambaddhatā, saṅgatārthatā), i.e., the mutual subordination (aṅgāṅgī‐
bhāva) of its statements, or the fact that its statements (vākya) converge towards a
single meaning (ekārthopasaṃhāra, mīlana; PVṬ Ñe D65b6–7/P79a4–5 ≈ PVSVṬ
612,19–20); (2) the existence of appropriate means (anuguṇopāya) for attaining its
purpose, such as the cultivation of selflessness (nairātmyabhāvanā, TSPK 877,24–
25/TSPŚ 1062,21–22; PVṬ Ñe D65b7/P79a5–6 = PVSVṬ 612,20–21); (3) the indica‐
tion of a human purpose (puruṣārthābhidhāyakatva), such as heaven (svarga), libe‐
ration (apavarga), worldly prosperity (abhyudaya), or the highest good (niḥśreyasa;
TRANSLATION 71
[and even] wishing to prove, just by [resorting to] the [occasional]

truthful indication of something trivial, that the Veda which says these
things is equally faultless when it comes to [those] profundities which
can scarcely be penetrated by great insight – he surpasses the unchaste
woman in audacity. There was (kila) a certain unchaste woman who
was reproached by [her] husband after catching [her] in the act of
adultery (vipratipattisthāna). She answered him: “Behold, mothers,146
the faithlessness (vaiparītya) of [this] man! Rather than putting his
trust in me, [his] lawful wife,147 he puts it in those two water‐bubbles
he calls his eyes! Though I was propositioned by this old, one‐eyed vil‐
lage wood gatherer,148 I never (na) had sex [with him]. Indeed, because
I am attached to [your] beauty and virtues, I [only] love [you, this]
young man who is chief among ministers.149” Of this sort150 is [our
adversary’s] inference, by means of the statement that fire counteracts
cold, that [the Veda] is also reliable with regard to the imperceptible,
even though it is seen to be contradicted by the means of valid cogni‐
tion!151
PVṬ Ñe D65b7–66a1/P79a6 = PVSVṬ 612,21–22). See also PVṬ Ñe D66a1–

3/P79a7–b1 = PVSVṬ 612,24–27.
146 For an explanation of the form mātaḥ (= mātaraḥ), see PVSVṬ 613,12–17.
147 According to PVṬ Ñe D66a7/P79b7–8 ≈ PVSVṬ 613,17–18, dharmapatnī is to be
analyzed as follows: dharmasya sādhanabhūtā patnī dharmapatnīti madhya[ma]‐

padalopī samāsaḥa /. a PVṬ reads mar gyi bum pa źes bya ba la sogs pa lta bu’o after
samāsaḥ, i.e., *tailaghaṭa ityādivat. “A wife who is the means of accomplishing duty
is a lawful wife. [This is] a compound that omits the middle word, PVṬlike
tailaghaṭaPVṬ.”
148 “Old” (jarat) points to the lack of vigour (vayovaikalya); “one‐eyed” (kāṇa), to ugli‐
ness (vairūpya, because the eye is the supreme locus of beauty, paraṃ rūpasthānaṃ
cakṣuḥ); “of the village” (grāmya), to the lack of qualities such as skillfulness
(vaidagdhyādiguṇavaikalya); “wood gatherer” (kāṣṭhahāraka), to poverty (dāridrya;
PVṬ Ñe D66b2/P80a2–3 = PVSVṬ 613,20–22).
149 “Beauty” (rūpa) points to elegance (prāsādikatva); “virtues” (guṇa), to skillfulness,
etc. (vaidagdhyādi); “young man” (dāraka), to vigour (vayas); “chief among minis‐
ters” (mantrimukhya), to power (aiśvarya; PVṬ Ñe D66b5–7/P80a7–b1 ≈ PVSVṬ
613,28–614,7).
150 I.e., like the unchaste woman’s reply (PVṬ Ñe D67a5/P81a1 = PVSVṬ 614,8–9).
151 PVṬ Ñe D67a7–b4/P81a4–b2 ≈ PVSVṬ 614,14–19: tatra dharmapatnīsthānīyo ve‐
daḥ / vipratipattitulyaṃ nityasya puṃsaḥ kartṛtvādyabhidhānam / netratulye praty‐

akṣānumāne / na ca dṛṣṭavyabhicārāyāḥ patnyā vacanaṃ garīyas tasya puruṣasya
72 PVSV 164,24–176,16
152If [the entire Veda] were to be established as a means of valid
cognition in this way, then what in this [world] would not be a

means of valid cognition? Indeed, for [any] person who says
lots of things there is [at least] one veracious [statement]. (PV
1.335)
Just as always (atyantam) saying what is true is extremely difficult,153
so also is always saying what is not true. In that case, if due to the fact
yena [sva]yaṃ vipratipattiṃ dṛṣṭvāpi svadarśanam apramāṇīkṛtya tasyā vacanaṃ

yathārthaṃ kuryāt / evaṃ vedoktārthabādhakayoḥ pramāṇayor aprāmāṇyaṃ kṛtvā
na vedasya patnīsthānīyasya dṛṣṭavyabhicārasya vacanād atyantaparokṣaṃ pratipa‐
dyemahīti /. “Here, the Veda is represented by the lawful wife. To claim, for exam‐
ple, that a permanent soul is the agent [of deeds], is similar to the crime [of adul‐
tery]. As for [the two pramāṇas] perception and inference, they are similar to the
[husband’s] eyes. And it is not the case that for this man the words of [his] wife
whose transgression [he has] witnessed (dṛṣṭa) are of greater weight [than his own
eyes], so that, even though he has himself witnessed [her] crime, he should disre‐
gard the evidence of his own vision (svadarśanam apramāṇīkṛtya) and consider her
words veracious. Thus, we cannot, considering the two pramāṇas that contradict
the things said in the Veda to be invalid (aprāmāṇyaṃ kṛtvā), know that which is
transcendent from a statement of the Veda – [here] representing the wife – whose
unreliability has been [duly] witnessed.” See also Śākyabuddhi’s additional explana‐
tions in PVṬ Ñe D67b1–3/P81a5–8. Earlier in this passage Śākyabuddhi already
provided a lengthy explanation of Dharmakīrti’s intention; see PVṬ Ñe D66b7–
67a5/P80b2–81a1. One can understand Dharmakīrti’s image as follows: (1) The
husband’s eyes catch the wife in flagrante delicto. ≈ The two pramāṇas find the Veda
to be mistaken. (2) The wife replies, in order to deny her guilt: ≈ (The opponent
has) the Veda answer, in order to deny its erroneousness: (3) I wasn’t unfaithful to
you, don’t trust your (“water‐bubble”) eyes. ≈ The Veda isn’t wrong, don’t trust your
pramāṇas! (4) Since everybody admits that one cannot have sex with a stupid, poor,
ugly old man, what I say is true, I wasn’t unfaithful to you. ≈ Since everybody admits
that fire is a remedy for cold, what the Veda says is true, not erroneous. (5) In the
same way as I am truthful on this point, I am to be trusted when I say that I love you,
who are young, handsome, talented, and rich. ≈ In the same way as the Veda is
truthful about empirical matters, it is to be trusted when it concerns transcendent
things.
152 In PV 1.335 and PVSV 175,7–10, Dharmakīrti points out another fault in his oppo‐
nent’s argument (PVṬ Ñe D67b4–5/P81b2 = PVSVṬ 614,20).

153 According to Śākyabuddhi (PVṬ Ñe D67b6–7/P81b5–6) Dharmakīrti says “extreme‐
ly difficult” (atiduṣkara) and not “totally impossible” (*atyantāsambhava), because

always saying what is true can indeed be achieved by those who have eliminated all
the obstructions (*prahīṇasarvāvaraṇa) and by those who have completely incorpo‐
TRANSLATION 73
that [just] one statement [made by a person] is somehow154 reliable,

the rest of his statements, which are vast in number, were like that [i.e.,
reliable, then] there would be no person who would not be trustwor‐
thy.
155Moreover,
This audible sound (dhvani) [which rests and originates] in the

speaker is neither the nature nor the effect of the entities [ex‐
pressed by it].156 Now, apart from these two [logical reasons]
there is nothing [else that might come into consideration] that
does not deviate. (PV 1.336)
First, speech is not the nature of the [things] signified [by it], nor is it
their effect, either; for it exists merely due to the speaker’s intention
even when these [things] are absent. Now, there is no other [kind of
logical reason] for something [than these two] that does not deviate.
And since, there being a deviation, a [word] could also occur otherwise
(tato ’nyathāpi) [i.e., in the absence of that which is signified by it], the
rated compassion (sñiṅ rje goms par gyur pa can, *karuṇāsātmībhūta), i.e., Buddhas
and higher Bodhisattvas.
154 I.e., according to PVṬ Ñe D68a1/P81b7–8 = PVSVṬ 614,26–27, “after the manner of
the crow and the palm‐fruit,” (kākatālīya; see Monier‐Williams 1963: 266c s.v. ), i.e.,
completely accidentally.
155 In PVSV 175,10–176,12, Dharmakīrti leaves the question of ekadeśāvisaṃvādana
and returns, first (PV 1.336–338/PVSV 175,10–176,4), to the issue of the relation
between words and meanings introduced at PV 1.213/PVSV 107,19 (see below, p.
75 n. 162), and second (PV 1.339/PVSV 176,5–12), to the issue of non‐perception
which provided the original impetus for his excursus on scriptural authority (PVSV
107,14–176,4; see below, p. 75 n. 162). According to Dharmakīrti (see above, PVSV
172,20–22 and p. 59 n. 113) a relation between word and meaning could only con‐
sist in their identity (tādātmya, tādrūpya, tatsvabhāvatā) or in one’s causing the
other (tadutpatti), for something makes something known (gamaka) only if it has
the same nature as that thing or is produced by it (tatsvabhāvas tajjanyo vā san), but
not otherwise (nānyathā; PVṬ Ñe D68a3/P82a2–3 = PVSVṬ 615,10–11).
156 I.e.,
the audible sound, which has the speaker (vaktṛ) for its cause (nimitta; PV
1.336b vaktari is analyzed as a locative indicating the nimitta, PVṬ; note also PVṬ Ñe
D68a4/P82a4 = PVSVṬ 615,13: yasmād vaktari [s]ati dhvanir bhavati), is related to
the speaker’s intention only (icchāmātrapratibaddha); it does not depend on the
entities signified (vācyavastu) by it (compare PVṬ Ñe D68a5/P82a4–5 with PVSVṬ
615,14–15).
74 PVSV 164,24–176,16
understanding of the [thing signified] cannot be due to the presence of

that [word].
Objection: 157[Words are indirectly the effects of the things
signified, for] the occurrence158 of [these] signifiers is caused by
the perceptual cognition of the [things] signified. (PV 1.337ab)
Suppose the following [be urged]: Speech is indeed an effect, because
the signifier occurs due to the perceptual cognition of the [thing] signi‐
fied [, which in turn gives rise to the intention to express it. To this, we
answer:] If it were so,
how could this [occurrence of an utterance] present [, accord‐
ing to the scriptures considered,] mutually contradictory
meanings concerning one [and the same thing]?159 (PV
1.337cd)
If [there were] such a restriction that a word does not occur without
the thing signified, [then] an utterance occurring in different doctrines
would not attribute160 contradictory natures [such as “permanent” and
“impermanent”] to a single entity. It is indeed impossible that one [and
the same entity such as a] sound could be [both] permanent and im‐
permanent without [alternative] modes of being.161
157 The opponent now admits that there can be neither identity nor direct causality be‐
tween words and meanings, but contends that an indirect causality (< pāramparye‐
ṇa vacanam arthakāryam) between them obtains. Thus, signifiers being the indirect
effects of things signified, the former still make the latter known (gamaka; PVṬ Ñe
D68b2–3/P82b3–4 ≈ PVSVṬ 615,24–25). This indirect causality is accounted for as
follows: thing signified (vācyārtha) → perceptual cognition (darśana) of it → spea‐
ker’s intention (vivakṣā) → signifier (vācaka; PVṬ Ñe D68b3–4/P82b5 ≈ PVSVṬ
615,26–28).
158 I.e., either in the form of physical production (utpatti), as a Vaiśeṣika or a Buddhist
would conceive, or in the form of manifestation (abhivyakti), as a Mīmāṃsaka or a

Grammarian would contend (PVṬ Ñe D68b3/P82b4 = PVSVṬ 615,25–26).
159 I.e., it couldn’t, for it would have to have the same meaning (ekārtha) in all the
doctrinal systems (sarvapravāda) considered (PVṬ Ñe D68b5–6/P82b8 = PVSVṬ

616,11).
160 On
upasaṃhāra employed in a sense approximating samāropa (PVṬ Ñe D68b7/
P83a2 = PVSVṬ 616,14–15), see BHSD 142ab s.v. upasaṃharati and upasaṃhāra.
161 I.e., “without another aspect” (prakārāntareṇa vinā; PVṬ Ñe D69a1/P83a4 = PVSVṬ
616,16). According to Śākyabuddhi (PVṬ Ñe D69a1–2/P83a4–6) if one said “perma‐

TRANSLATION 75
162Therefore, scriptures necessarily related to entities are not in
any way established for the [person] engaged in [religious]

practice. How [then could this person ever] ascertain some‐
thing through them? (PV 1.338)
It is the ignorant person who looks at scripture as a means of valid
cognition in order to put into practice its teachings [, and this for two
reasons: first,] because [those] who have thoroughly understood the
truth163 do not [need to] resort to the instruction [of others any
longer]; and [second,] because the ignorant person cannot discriminate
human beings whose [mental] qualities are inaccessible to the senses
[and therefore he cannot recognize a trustworthy person].164 Someone
nent” with reference (rjes su ’jug pas) to the existence of a permanent aspect (*nitya‐
prakāra) and “impermanent” on the basis of (*apekṣya) a momentary phase (*kṣa‐
ṇa), then there would be no fault; but to assert opposed properties (*viruddha‐
dharma) without alternative modes of being (*niṣparyāyam) is contradictory in re‐
gard to one and the same thing.
162 After his long excursus on scriptural authority (PV 1.213–339/PVSV 107,14–176,4)
Dharmakīrti now returns to the issue at stake in PVSV 107,14–25 (Yaita 1987: 5–6,
Eltschinger 2007a: 217–218). In PV 1.199 and PVSV 101,23–102,8 (≈ PVin 2 65,1–9,
see Steinkellner 1979: 62, Yaita 1985: 215–214) he had shown that the silence or
non‐operation (nivṛtti) of the three means of valid cognition (pratyakṣa, anumāna,
āgama), i.e., mere non‐perception (anupalabdhimātreṇa; PVSV 101,23; cf. pramāṇa‐
trayanivṛttilakṣaṇā ’nupalabdhiḥ; PVSV 176,7), does not allow one to establish the
non‐existence of an entity. In PVSV 107,14–17, an opponent attacks this conclusion
and contends that the silence of scripture does allow one to establish the non‐exist‐
ence of an entity because scripture covers (vy√āp) all that exists. In PVSV 107,17–
19, Dharmakīrti summarizes his previous arguments and then says (PV 1.213; for
PVSV 107,22–25 [≈ PVin 2 66,4–7] thereon, see Yaita 1987: 6, Eltschinger 2007a:
218, and Steinkellner 1979: 65): nāntarīyakatā’bhāvāc chabdānāṃ vastubhiḥ saha /
nārthasiddhis tatas te hi vaktrabhiprāyasūcakāḥ //. “Due to the absence of a neces‐
sary relation of words with things, one [can]not establish [i.e., ascertain] any object
through them [i.e., words], because they [merely] indicate the speaker’s intention.”
163 I.e., those who know the true nature of entities (padārthānām aviparītaṃ rūpam)
and thus have thoroughly understood ultimate reality (adhigataparamārtha; PVṬ

Ñe D69a5–6/P83b2–3 ≈ PVSVṬ 616,25–26).
164 I.e., a cognitively limited person (arvāgdarśin) cannot distinguish (< vibhāgakriyā)
whether someone is omniscient (sarvajña) or not, or whether someone says what is

true (avitathābhidhāyin; PVṬ Ñe D69a6–7/P83b3–4 ≈ PVSVṬ 616,28–617,12). As a
consequence, since he cannot ascertain this omniscient or veracious person’s
speech as being scripture (āgamatvena), this arvāgdarśin cannot engage in religious
practice (PVṬ Ñe D69a7/P83b4–5). This refers back to PV 1.218–219 and PVSV
76 PVSV 164,24–176,16
who sees that statements occur even in the absence of the desired
meaning must indeed be beset with doubt – [asking himself,] “Is it true
or not?” – even in the case of the discourse of people whose utterances
have not been observed to deviate [from the truth]. Therefore, the [ig‐
norant person] cannot ascertain anything by means of someone’s
statement.
Therefore,165 the non‐existence of an entity [to be negated] is
not established by the silence of [scripture] either. (PV
1.339ab)
[You] have stated [above] that non‐perception defined as the silence of
[scripture] proves non‐existence. For everything falls within the scope
of scripture, so that if something exists it will refer (vṛtti) [to it] in a
reliable way.166 167[To this we reply:] Even if everything falls within its
scope, this [ascertaining the non‐existence of something] would be the
case [only] if [scripture] were [indeed] silent when something does not
exist. But that is not the case.168 [Thus we have] stated that the [per‐
son] who wishes to engage in [religious] practice cannot establish
[anything] through it [i.e., scripture].
Therefore, it is not established that non‐perception results in
the ascertainment [of something] as non‐existent. (PV 1.339cd)
109,24–110,15, where Dharmakīrti demonstrates that the contents of other per‐

sons’ mental streams (santāna) are neither within the scope of (ordinary) percep‐
tion (since they are supersensible) nor inferrable from physical and verbal behav‐
iours (kāyavāgvyavahāra, since human beings are often seen to deliberately be‐have
in a deceptive way). See Yaita 1988: 434–435, Akimoto 1993, Dunne 2004: 366–
368, and Eltschinger 2007a: 92–96 and 227–230.
165 I.e., because of the impossibility of any ascertainment (aniścaya; PVṬ Ñe D69b3/
P84a1 = PVSVṬ 617,20–21).

166 See above, p. 75 n. 162.
167 Rather than repeating the arguments put forward in PVSV 102,2–8 (e.g., the argu‐
ment that scripture does not deal [aviṣayīkaraṇa] with matters not related to hu‐
man goals [puruṣārthāsambaddha] and thus does not embrace all possible objects),
Dharmakīrti here provisionally accepts that everything falls within the scope of
scripture (PVṬ Ñe D69b5–6/P84a4–5 ≈ PVSVṬ 617,26–28).
168 Because scripture also exists (vṛtti) independently of real entities (vastv antareṇa;
PVṬ Ñe D69b7/P84a6–7 = PVSVṬ 618,6).

TRANSLATION 77
Therefore, the non‐existence of [things] beyond the reach [of ordinary

cognition] cannot be ascertained even if all three means of valid cogni‐
tion [should] fail to operate.
169[Believing in the] authority of the Veda,170 claiming some‐
thing [permanent] to be an agent,171 seeking merit in ablu‐
tions,172 taking pride in one’s caste,173 and undertaking penance
169 Having shown throughout PV 1 that non‐Buddhists (tīrthika) only talk nonsense
(ayuktābhidhāyitva), Dharmakīrti concludes the work by pointing out, in a merely

indicative way (diṅmātram), the five principal signs of their stupidity (jāḍya; com‐
pare PVṬ Ñe D70a2–3/P84b1–2 with PVSVṬ 618,12–13). That Dharmakīrti alludes
to only five signs is a synecdoche (*upalakṣaṇa), for the signs of the stupidity of the
outsiders are innumerable (*aprameya; PVṬ Ñe D70b6/P85a8–b1, to be compared
with PVSVṬ 619,7–8).
170 As the Brahmins do, thus revealing their own ignorance (*ajñāna; PVṬ Ñe D70a3–
4/P84b3–4). For grammatical explanations of the compound vedaprāmāṇya, see

PVṬ Ñe D70a4–5/P84b4–6 = PVSVṬ 618,15–17.
171 The words kasyacit kartṛvādaḥ can be interpreted in at least two ways. Śākyabuddhi
interprets kasyacit as *ātmāder nityasya (PVṬ Ñe D70a5/P84b6), thus understand‐

ing kartṛ as “agent” (see above, PV 1.332a and PVSV 174,14–16; see above, pp. 65–
66): “claiming something [permanent] to be the agent [of deeds].” As for Karṇakago‐
min, he interprets kasyacit as naiyāyikādeḥ (PVSVṬ 618,18) and kartṛ as “creator”
(īśvaras tattvādīnāṃ karteti; PVSVṬ 618,18): “someone’s claiming that there is a
creator [god].”
172 Merit (dharma) consists in the absence of attachment (*sneha), hostility (*dveṣa),
and error (*moha), as well as the physical and verbal actions born of these mental
factors (tajjanitaṃ kāyavākkarma). Now, ablutions consist in mere contact with wa‐
ter (jalasaṃśleṣamātralakṣaṇa) and thus cannot bring about merit (PVṬ Ñe D70a6–
b1/P84b8–85a1 ≈ PVSVṬ 618,20–23). Although Dharmakīrti implies a critique of
ablutions along these lines in PVSV 109,1–3 (see Yaita 1987: 8, Dunne 2004: 363,
Eltschinger 2007a: 105–109, 223, and above, pp. 69–70 n. 142), Śākyabuddhi and
Karṇakagomin (ibid.) refer back to the “detailed refutations” (cf. vistareṇa nirākṛ‐
taḥ) of tīrthasnānavāda by Ācārya Vasubandhu (in AKBh 282,8–9?) and other
Buddhist masters. See La Vallée Poussin 1980: III.135 n. 2 and Eltschinger 2007a:
108 n. 126 for references.
173 I.e., though devoid of qualities such as morality (śīlādiguṇavaikalye ’pi), someone
takes pride in his caste by saying, “I am a Brahmin” (PVṬ Ñe D70b1/P85a1–2

≈ PVSVṬ 618,24). Note PVṬ Ñe D70b2–3/P85a2–4 ≈ PVSVṬ 618,25–28: [tathā hi
brāhmaṇatvasāmānyaṃ gṛ]hītvāa jātivādāvalepaḥ syāt / brāhmaṇena pitrā brāhma‐
ṇyā garbhe ya utpādas taṃ vā samāśritya / tatra vastubhūtasāmānyaniṣedhān na
pūrvaḥ pakṣaḥ / nāpy uttaraḥ / brāhmaṇabrāhmaṇīśarīrayor aśucisvabhāvatvena
śarīrāntarād aviśeṣāt /. a According to PVṬ de ltar na bram ze ñid kyi spyi gzuṅ nas.
78 PVSV 164,24–176,16
to remove sin,174 these are the five signs of complete stupidity

devoid of any discrimination.175 (PV 1.340)
“Indeed, one could take pride in one’s caste either by accepting the universal ‘Brah‐
minhood’ or by appealing to one’s birth from a Brahmin father in the womb of a
Brahmin woman. Among these [two possibilities] the first hypothesis is excluded
(na), because real universals have been refuted [in the section devoted to apoha].
Nor is the second [hypothesis] sound, because the bodies of the Brahmin male and
Brahmin woman do not differ from [any] other body in being of an impure nature.”
On Dharmakīrti’s critique of caste, see PVSV 157,9–18 and Eltschinger 2000: 103–
115; on the two arguments of Śākyabuddhi and Karṇakagomin, see Eltschinger
2000: 139–140.
174 I.e., tormenting one’s own body (śarīrapīḍana) by practices such as fasting (anaśa‐
nādi; PVṬ Ñe D70b4/P85a5–6 = PVSVṬ 619,1). The stupidity of these undertakings

is like that of ablutions (see above, p. 77 n. 172). The root cause (nidāna) of all evil
(pāpa) consists of greed, hostility, and error (lobhadveṣamoha), which are mental
factors. Undertaking physical penance (santāpārambha) is not contradictory (avi‐
ruddha) to these factors (i.e., to the causes of evil), hence it cannot purify one from
evil (pāpaśuddhi; PVṬ Ñe D70b4–5/P85a6–7 ≈ PVSVṬ 619,2–4).
175 I.e., devoid of rational knowledge (nyāyānusārijñānarahita), viz., of any knowledge
that derives from the force of the pramāṇas (pramāṇabalaṃ vijñānam; PVṬ Ñe
D70b5–6/P85a7–8 = PVSVṬ 619,5–7).

Appendix: Philological Remarks
PVSV 166,20: read jñānayāthātathyaṃ (PVSVTib D359b2/P526b5, PVṬ

Ñe D49a4–5/P56b2, PVSVṬ 589,17).
PVSV 169,9: delete vā (PVSVṬ 596,28, Sāṅkṛtyāyana’s edition of the

Svavṛtti [596,5], PVṬ Ñe D54b5/P64a6). Note that PVSVTib D361a5/
P528b7 reads (de dag) daṅ (gźan).
PVSV 169,23 – PVSVTib D361b2/P529a7 and PVṬ Ñe D56a1/P65b7–8

read ’jig rten gyi tha sñad las, suggesting *lokavyavahārāt instead of
lokāt (PVSVṬ 598,16, Sāṅkṛtyāyana’s edition of the Svavṛtti [598,2]).
PVSV 170,20–21 – °mithyācāra° (ms A, Sāṅkṛtyāyana’s edition of the

Svavṛtti [601,5]), against (1) °maithunācāra° (ms B, and probably
PVSVṬ 601,23 [°maithunācaraṇa°], PVṬ Ñe D58a5/P68b8 ['khrig pa
spyod par byed]), and (2) *°vivāhamithyācāra (PVSVTib D362a5/P530a6
[bag mar len pa'i log pa'i spyod pa]; cf. PVSV 125,23 and PVSVTib
D333b2/P493a3: mātṛvivāha = ma bag mar len pa).
PVSV 171,26: read parīkṣyatām.
PVSV 174,22: read prasiddhaviparyayam (PVSVṬ 612,7–8, PVSVTib

D364a6/P533a7, PVṬ Ñe D65a7/P78b2–3).
PVSV 175,8–9: read saṃvādenāvaśiṣṭasya (PVSVTib D364b4/P533b8,

PVṬ Ñe D68a2/P81b8).

Part 2 – Studies

HELMUT KRASSER
Logic in a Religious Context:

Dharmakīrti in Defence of āgama1

1. INTRODUCTION
Contrary to what many have believed, Dharmakīrtiʼs (~550)2 view of
scripture (āgama)3 is unambiguous and clear.4 In sharp contrast to
Vasubandhu, who, in accordance with the Bodhisattvabhūmi and other
earlier Yogācāra texts, accepted three means of valid cognition (pramā‐
ṇa), namely, perception, inference, and – as the third – either the Bud‐
dha or the Sons of Buddha or their teaching,5 and also in contrast to
Dignāga, who accepted scripture as a means of valid cognition but sub‐
sumed it under inference, Dharmakīrti neither counts scripture as a
separate means of valid cognition nor subsumes it under inference. For
Dharmakīrti no scripture, including even the Buddhist teaching, is a
1 A short version of this paper was presented at the 14th World Sanskrit Conference,
September 1–5, 2009, Kyoto University.

2 For the proposed date of ca. the middle of the sixth century for Dharmakīrti’s time
of activity, see Krasser 2012.

3 The translation “scripture” for āgama is problematic (see Eltschinger 2007a: 17–
20) and only adopted for want of a better one. āgama should also comprise, e.g., the
eternal words of the Veda or the oral teachings of the Buddha or Mahāvīra that have
been handed down without interruption.
4 For
a general overview of Dharmakīrti’s philosophical thought, see Eltschinger
2010.
5 See AKBh 76,22–23: kiṃ kāraṇam / pramāṇābhāvāt / na hy eṣāṃ dravyato ʼstitve
kiñcid api pramāṇam asti pratyakṣam anumānam āptāgamo vā, yathā rūpādīnāṃ
dharmānām iti /; AKBh 460,2–3: saddharmanītau tu punar buddhā eva pramāṇaṃ
buddhaputrāś ca /. See also BoBhū 25,19: pratyakṣam anumānam āptāgamaṃ
pramāṇaṃ niśritya. For the acceptance of these three kinds of pramāṇas in the VY,
see Verhagen 2008; for their acceptance in other Yogācāra texts, see Eltschinger,
forthcoming, n. 79.
84 HELMUT KRASSER
means of valid cognition, and there can be no certainty based on scrip‐

ture – quite a remarkable position for a Buddhist monk! The reason
why this unambiguous understanding of scripture has not always been
recognized as such is Dharmakīrti himself. He does not proclaim it very
loudly – for, again, it entails that even the buddhavacana cannot be re‐
garded as a pramāṇa – probably not only because it contradicts the
definition of scripture given by Dignāga, but also because it was at odds
with the attitude of the Buddhist community. Such a view certainly
would not have been favoured by the spiritual authorities.
2. PVSV 108,1–109,22 ON PS 2.5ab

Dharmakīrti deals with Dignāgaʼs de inition of scripture in the context
of his theory of concept formation and language, apoha. One of the
cornerstones of his apoha theory, as formulated in PV 1.213, is that
words have no connection with the external world; they can only make
known the intention of the speaker, not external objects.6 As this
contradicts Dignāgaʼs de inition of scripture, immediately after
explaining this verse in his Svavṛtti Dharmakīrti raises the following
question in order to deal with this problem:7
How then [should it be understood] when [Dignāga by his definition of
āptavāda,] “The statement of a credible person is inference, because it is
equal in not belying,” said that scripture is inference?8
6 PV 1.213: nāntarīyakatābhāvāc chabdānāṃ vastubhiḥ saha / nārthasiddhis tatas te
hi vaktrabhiprāyasūcakāḥ //. “Because words are not invariably concomitant with

real entities (vastu), therefore [they] do not establish real entities (artha). For they
make known [only] the speakerʼs intention.”
7 Dharmakīrtiʼs treatment of Dignāgaʼs āptavāda definition has been translated sev‐
eral times. For PV 1.213–217 with PVSV, see Yaita 1987 (included in Yaita 2005:
442–449); for PV 1.214–223 with PVSV, see Dunne 2004: 361–373; for PV 1.213–
268 with PVSV, see Eltschinger 2007a: 217–385.
8 PVSV 108,1–2: yat tarhīdam – āptavādāvisaṃvādasāmānyād anumānatā (PS
2.5ab) ity āgamasyānumānatvam uktam, tat katham /. A more accurate translation
will be offered below. As I intend to deal with Dignāgaʼs understanding of this de i‐
nition elsewhere, in the following I will discuss only the problems related to Dhar‐
makīrtiʼs treatment of it. For (different) interpretations of PS 2.5ab, see Eltschinger
2007a: 70 (with n. 9), 218ff., and Lasic 2010.
LOGIC IN A RELIGIOUS CONTEXT 85
Dharmakīrti devotes 4 verses with commentary to the solution of this

problem, in which he provides two different interpretations of this
definition.9 Of these interpretations I will refer only to those state‐
ments that are necessary for my argumentation.
2.1. Why did Dignāga provide a definition of scripture?

Immediately following the posing of the above question Dharmakīrti
tells us why we are in need of āgama at all.
A person cannot live without relying on the validity of scripture because
he has heard that in the case of certain [activities] the results of which
cannot be perceived, engaging or not engaging [lead to] extremely
praiseworthy or disastrous [results], and because he does not see any‐
thing that contradicts the presence of those [results]. Therefore [Dignāga]
taught that the validity [of scripture] is due to a critical examination [of it,
having in mind] “given that one has to proceed, it is better to proceed in
such a way [i.e. critically examining the scripture]10.”11
... If that [treatise,] upon being examined, is not liable to making false as‐
sertions (na … visaṃvādabhāk), [then] the one who proceeds might shine
(śobheta)!12
In other words: We do need religion because it is the only means to es‐

cape saṃsāra. Thus, we have to base our religious activities on āgama.
9 For a helpful synopsis and concise summary of this excursus on PS 2.5ab, see Yaita
1987: 2–3 and Dunne 2004: 240–241. For proposed changes, see below p. 100 n. 38.
10 Cf.
PVSVṬ 390,30–391,9: tat sati pravartitavye varam evam āgamaṃ parīkṣya
pravṛtta ity āgamasya parīkṣayā prāmāṇyam āhācāryaḥ.
11 PVSV 108,2–6: nāyaṃ puruṣo ʼnāśrityāgamaprāmāṇyam āsituṃ samarthaḥ, atyakṣa‐
phalānāṃ keṣāñcit pravṛttinivṛttyor mahānuśaṃsāpāyaśravaṇāta tadbhāve virodhā‐

darśanāc ca. tat sati pravartitavye varam evaṃ pravṛtta iti parīkṣayā prāmāṇyam
āha.
a °āpāya° PVSV 25a2 (TSP 4,22–23; PVSVṬ 140b6; ṅan soṅ PVSV D322a4/
ms ms Tib
P478b3; see Yaita 2005: 443 n. 159) : °āpāpa° PVSV, PVSVṬ 390,25.
12 PVSV 108,15–16: tad yadi na parīkṣāyāṃ visaṃvādabhāk pravartamānaḥ śobheta.
86 HELMUT KRASSER
2.2. PV 1.214–215: Which āgama should be followed?

In order to decide which āgama we should follow – this is explained in
PV 1.214 with commentary – we have to look for a scripture that is
coherent (sambaddha) and teaches a proper human aim (puruṣārtha)
as well as a suitable means for obtaining it (anuguṇopāya). If we find
such a scripture, for example that of the Buddhists, then we have to ex‐
amine it in its entirety, as explained in PV 1.215 with PVSV. Whatever it
teaches that can be checked by perception and inference has to be true.
Anything taught in such a scripture must not be contradicted by per‐
ception or inference. Moreover, there cannot be a single internal con‐
tradiction in it. Such internal contradictions are checked with the help
of the so‐called āgamāpekṣānumāna (PVSV 109,1),13 i.e., an inference
that, based on passages of the scripture, tests its consistency.14 If an
āgama passes this threefold test15 – whereby it is tested in all testable
13 Later (PVSV 174,22–23; see above, p. 69) Dharmakīrti uses the formulation āgamā‐
śrayeṇa cānumānena. In contrast to this kind of inference an inference that operates

with visible matters is called anāgamāpekṣānumāna (PVSV 108,24). In the fourth
chapter Dharmakīrti terms this latter kind of anumāna together with perception as
vastubalapravṛttapramāṇa (PV 4.108 ≈ PVin 3 31,11: yac chāstraṃ vastubalapravṛt‐
tena pramāṇena svavacanena cābādhitaṃ dṛśyādṛśyayor viṣayayoḥ, tad grāhyam iti).
These two kinds of inferences are already distinguished by Pakṣilasvāmin (NBh
3,12 on NSū 1.1.1: pratyakṣāgamāśritaṃ cānumanam).
14 On the application of this method as described in PV 1.214–215 by Śāntarakṣita and
Kamalaśīla, see McClintock 2010: 318ff.

15 A possible source of this threefold analysis, which came to be known as dpyad pa
gsum in the later Tibetan tradition, might be Vasubandhuʼs VY; see Verhagen 2008:
244–247 (Section Four, “Three pramāṇas”) with the appendix (pp. 253–258). A si‐
milar idea is encountered in the definition of āptāgama in ASBh 153,5ff.: yatropa‐
deśe tat pratyakṣam anumānaṃ ca sarvathā na virudhyete na vyabhicarataḥ sa āptā‐
gamaḥ sampratyayitvāt. “Die Unterweisung, zu der diese Wahrnehmung und [diese]
Schlußfolgerung in keiner Weise im Widerspruch steht [und von der] diese nicht
abweichen, ist āptāgama, weil er Vertrauen erweckt.” Translation Oberhammer et
al. 1991: 122b–123a s.v. āptāgama. On the background of the Tibetan “threefold ana‐
lysis (dpyad pa gsum)” see Tillemans 1993: 10ff., and Keira 2006: 182 with n. 15,
who informs us that the notion of “purified through the threefold analysis (tshul
gsum gyi brtag pas yoṅs su dag pa)” is already found in Kamalaśīlaʼs MĀ (D148b4–
5): [deʼi phyir] bkaʼ gaṅ źig tshul gsum gyi brtag pas yoṅs su dag pa thog ma daṅ tha
ma daṅ bar du dge bar ṅes pa de ni mi slu baʼi phyir de mkhas pa rnams kyis brten par
bya ba ñid do //. “[Therefore,] since statements ascertained as faultless by means of
three kinds of investigation (tshul gsum gyi brtag pas) and as excellent for the first
cases (śakyaparicchedāśeṣaviṣayaviśuddhi, PVSV 109,3–4) – and is thus

established to be without contradictions, its being pure or cleansed
(viśuddhi)16 constitutes its non‐belying nature or reliability (avisaṃ‐
vāda, PVSV 109,4). “Non‐belying” is of course the definitional charac‐
teristic of a valid cognition.17 If a scripture is “purified” in this way,
then we can infer that this scripture or testimony of a credible person
(āptavāda) might be true also with regard to the non‐empirical or
transcendent things18 it teaches, because the scripture or credible
person is the same. And if then, based on that scripture, one acts to‐
wards non‐empirical objects one might be lucky and obtain what one
was looking for (see n. 12).
2.3. Does PV 1.216 comment upon PS 2.5ab?

Up to this point the existing interpretations and translations of PV(SV)
(see n. 7) more or less agree, and I have followed them thus far. The
situation changes, however, with the next verse, i.e., PV 1.216. Scholars
disagree about the meanings of the words agatyā and sāmānyāt and
[stage of practice, i.e., hearing (śruti)], the last [stage, i.e., meditation (bhāvanā)],
and the middle [stage, i.e., consideration (cintā)], are non‐belying (avisaṃvāda),
scholars should rely upon such [statements].” On the resemblance of Dharmakīrtiʼs
threefold test to Āryadevaʼs CŚ 12.5 (= 12.280 in Tillemans counting), see Tillemans
1990: I.29–32.
16 The choice of the term viśuddhi might have been inspired by a similar use in the Ślo‐
kavārttika, where śuddhatva of the causes guarantees the validity of the resulting
cognition; see ŚV codanā 44ab: tasmāt kāraṇaśuddhatvaṃ jñānaprāmāṇyakāraṇam.
“Therefore (tasmāt), the cause of validity in a cognition (jñānaprāmāṇyakāraṇam) is
the excellence of causes (kāraṇaśuddhatvam).” Translation Kataoka 2011a: 248. See
also Kataokaʼs n. 189 on the translation of k. 46, where the term śuddhyasambhava is
used.
17 See PV 2.1ab1: pramāṇam avisaṃvādi jñānam. “Valid cognition is non‐belying cogni‐
tion.”
18 In this context Dharmakīrti refers to these
kinds of things as “not knowable by
perception or inference” (pratyakṣānumānāgamye ʼpy arthe PVSV 109,8). Later he
also uses the term atyantaparokṣa (PVSV 153,7, 155,14, 175,3–4, PV 1.314c, 1.316a,
3.94, 4.210). For his usage of parokṣa in the sense of atyantaparokṣa, see Tillemans
1986: 44 n. 14 (= 1999b: 34 n. 13). The term atyantaparokṣa is also used by Uddyo‐
takara (NV 204,12 on NSū 2.1.31) and occurs also in YD 70,15 on SK 4ab1, 99,12 on
SK 6cd, and 101,7 still on SK 6cd.
88 HELMUT KRASSER
about the overall purport of the verse. Thus, it seems appropriate to

have a closer look at the different positions.
2.3.1. Previous interpretations of PV 1.216

Let me begin with Tom Tillemans, to whom we owe much of our
understanding of the Buddhist epistemologistsʼ approach to scripture.
In his article “Dharmakīrti, Āryadeva, and Dharmapāla on scriptural
authority” he provides us with the following translation of the state‐
ment of Dignāgaʼs we have been considering:
PS II, k. 5a: Because authoritative words (āptavāda) are similar [to an
inference] in not belying, they are [classified as] inference. (Tillemans
1986: 32 = 1999b: 27ff.)
In his summary of Dharmakīrtiʼs PV 1.216, which if I understand him

correctly19 he considers together with PV 1.215 to be the core of Dhar‐
makīrtiʼs explanation of PS 2.5ab, Tillemans gives a slightly different
account of Dignāga’s statement. But first, let us have a look at Tille‐
mans’ translation of PV 1.216.
āptavādāvisaṃvādasāmānyād anumānatā /
buddher agatyābhihitā parokṣe ʼpy asya gocare // PV 1.216
As authoritative words are similar in being avisaṃvāda, the understand‐

ing of their imperceptible (parokṣa) object is also termed an inference, for
[otherwise] there would be no way [to know such objects]. (Tillemans
1993: 11)20
On the basis of this translation of PV 1.216 Dignāga’s position in PS

2.5ab is summarized by Tillemans as follows:
(d) v. 216: Dignāgaʼs point in saying that authoritative words were an
inference was that when an authorityʼs words (= scripture) have been
found to be non‐belying on rationally decidable matters, then we are jus‐
19 I derive this understanding from the fact that Tillemans presents Dignāgaʼs PS 2.5ab
together with PV 1.215–216 twice in order to outline “the epistemological schoolʼs

position” on scripture (Tillemans 1990: I.24ff.) and the background of the Tibetan
“threefold analysis (dpyad pa gsum)” (Tillemans 1993: 10ff.).
20 Also translated in Tillemans 1990: I.25.
tified to understand radically inaccessible matters based on that scrip‐

ture.... (Tillemans 1999a: 399–400 = 1999b: 41–42)
I would subscribe to Tillemans’ translation of PV 1.216, though we

shall have to come back to the expression agatyā later (below, pp. 108–
111). Nevertheless, Dharmakīrti’s understanding of PS 2.5ab in the
way Tillemans is suggesting poses a problem. I cannot see any basis in
the wording of PS 2.5ab itself for saying that our understanding of the
transcendent object of scripture is termed an inference. It is one thing
to say that āptavacana is anumāna and another that our cognition of its
object is anumāna. In other words, did Dharmakīrti over‐interpret or
even misinterpret Dignāga, or must we understand Dignāga in a differ‐
ent way?
As the āgama‐section of Dharmakīrtiʼs PVSV has been translated by
Hideomi Yaita in 1987, by John Dunne in 2004, and by Vincent
Eltschinger in 2007a, we may gain some help from considering their in‐
terpretations. Before looking at their analyses it should be mentioned
that PS(V) 2.5ab is incorporated by Dharmakīrti in his PV(SV) twice,
and that he interprets it somewhat differently each time. First, he re‐
fers to it when initially raising the question of its meaning, and then
about a page later in the Gnoli edition, in concluding his first interpre‐
tation of PS(V) 5ab. On its first occurrence it is marked as a quotation,21
on the second Dignāgaʼs wording is used without explicitly characteris‐
ing it as a quotation. The latter passage, namely PV 1.216, is com‐
mented upon by Dharmakīrti as follows:
āptavādāvisaṃvādasāmānyād anumānatā /
buddher agatyābhihitā parokṣe ʼpy asya gocare // PV 1.216
tasya cāsyaivambhūtasyāptavādasyāvisaṃvādasāmānyād adṛṣṭavyabhi‐

cārasya pratyakṣānumānāgamye ʼpy arthe pratipattes tadāśrayatvāt tad‐
anyapratipattivad avisaṃvādo ʼnumīyate / tataḥ śabdaprabhavāpi satī na
śābdavad abhiprāyaṃ nivedayaty evety arthāvisaṃvādād anumānam api /
(PVSV 109,8–11)
21 PVSV 108,1–2: yat tarhīdam āptavādāvisaṃvādasāmānyād anumānatety āgamasyā‐
numānatvam uktam, tat katham.

90 HELMUT KRASSER
Dunneʼs translation of the verse with commentary reads,22

[Dignāga] said that, since the statements of a credible person are gener‐
ally trustworthy, a cognition from such statements of those statementʼs
object is a well formed inference of that object, even though the object is
epistemically remote. The cognition is said to be an inference because
there is no other way to know that object. [PV 1.216]
These kinds of statements of a credible person – those [described by

Dignāga] and those [accordingly delineated above] – are generally (sāmā‐
nya) trustworthy. Hence, not observing those statements to be mislead‐
ing, one infers the trustworthiness of a cognition of those statementsʼ
object, even though it is not knowable by perception and empirical infer‐
ence. One infers that such a cognition is trustworthy because it is based
on those statements, just like the other cognitions based on those state‐
ments that can be verified by perception and empirical inference.
Hence, even though that cognition comes from language, it does not make
known just the speakerʼs intention like a cognition coming from [ordi‐
nary] language because in this case the cognition is also an inference of
the statementʼs objects, since it is trustworthy with regard to those ob‐
jects (artha). (Dunne 2004: 363ff.)
Here it is clear, as we have already seen in the translation of Tillemans,

that in the verse our cognition of a transcendent object derived from
the statements of a credible person is said to be an inference. However,
Dunne’s translation of the earlier occurrence of PS 2.5ab, where Dhar‐
makīrti quotes it (PVSV 108,1–2) to introduce his discussion of scrip‐
ture, construes the testimony of a credible person as a source for an in‐
ference.
[Dignāga] said, “The testimony of a credible person is the source for an in‐
ference because it is generally trustworthy,”... (Dunne 2004: 361)
In a section earlier in his book which gives a general account of scrip‐

tural inference, titled “Scriptural Inference and Dharmakīrtiʼs Rejection
of Credibility,” Dunne translates the same passage in another way.
22 In the translation offered by John Dunne his explanatory remarks are not always
written in brackets in order to present a smooth text readable also for non‐special‐
ists. However, they easily can be recognized as such.
Since the statements of a credible person are generally trustworthy, a

cognition arising from them is an instrumental inference. (Dunne 2004:
239)
In the footnote to the translation Dunne justifies this interpretation

with reference to PV 1.216.
Note, first, that although the phrase “a cognition arising from them” must
be supplied by context, Dharmakīrti clearly understands this statement in
that fashion (see PV I.216, translated along with the rest of the section on
scripture in the appendix).
Interestingly enough, this double understanding of one and the same

passage – as referring to either scripture or the cognition that arises
from it as an inference – is also to be found in Hideomi Yaitaʼs treat‐
ment of Dharmakīrtiʼs view of scripture. In his synopsis he summarises
the passage we have just referred to as follows:
Opponent (108,1f.): Dharmakīrtiʼs statement is not consistent with Dignā‐
gaʼs statement in PS II 5ab, which recognized āgama (āptavāda) as a pra‐
māṇa, i.e. anumāna. (Yaita 1987: 2)
His translation of PVSV 108,1–2 reads,

(Opponent:) Then, how [do you explain] the fact that [Dignāga] said the
sacred tradition (āgama) is [a pramāṇa, that is] anumāna [in his follow‐
ing statement:] “Because the words of a credible person (āptavāda)
generally do not disagree, [the cognition based on it] is an inference (anu‐
māna)”? (Yaita 1987: 6)
The above interpretation of PS 2.5ab in this pūrvapakṣa is obviously

based on PV 1.216.
Because the word of a credible person (āptavāda) generally does not dis‐
agree [with our experience], the cognition [based on āgama] even with
regard to its (= āgamaʼs) object beyond the range [of any means of knowl‐
edge] is said [by Dignāga] to be an inference, for there is no [other] possi‐
bility (agatyā) [to explain it]. (Yaita 1987: 8)23
23 Van Bijlert (1989: 123–124) is also of the opinion that PV 1.216 is an explanation of
PS 2.5ab: “In the next verse Dharmakīrti uses this general conception of trustwor‐
thiness to give his first explanation of PS II.5ab … (PV I.216).”
92 HELMUT KRASSER
Eltschinger (2007a: 219, 224) does not read PVSV 108,1–2 in light of
PV 1.216 but treats both passages independently. Thus, the question
remains: Did Dignāga say in PS 2.5ab that scripture is a kind of infer‐
ence or did he term our understanding of their transcendent (parokṣa)
objects an inference, or did he say both?
2.3.2. On the context of PV 1.216 with PVSV

As I cannot see how one can harmonize these two interpretations, the
one clearly saying that āgama is inference (PVSV 108,1–2), the other
that our cognition is inference (k. 216), it seems worthwhile to have
another look at the context of PS 2.5ab.24 This has been nicely summa‐
rized by Tillemans when arguing against Hayesʼ interpretation of PS
2.5ab in his monograph Dignāga on the Interpretation of Signs.
2) In the immediately preceding passage to k. 5, Dignāga had made a
distinction between two types of inference, depending upon whether the
object is empirical or non‐empirical, arguing that in the former case we
can apply names to what is empirical, but in the latter case we only have a
concept (rnam par rtog pa = vikalpa) and do not cognize the svalakṣaṇa
object.
3) An opponent then tries to find an absurdity, saying that in that case au‐
thoritative statements about imperceptible objects would just express the
conceptually invented object and not the real particular at all: hence there
would be no difference between authoritative and unauthoritative state‐
ments.
4) Dignāga then replies that authoritative statements about heaven and

the like do not express just the conceptually invented object: they are
similar to normal inference because they too are non‐belying with regard
to the real particular. For, although the heavens and so forth are beyond
our sense range, authoritative people have directly seen them and hence
were able to apply the words “heaven,” etc. (Tillemans 1990: I.22)
Tillemans had to base his summary on the two barely intelligible Ti‐
betan translations of the PS(V) by Vasudhararakṣita and Kanakavar‐
man. With the help of Jinendrabuddhiʼs commentary in Sanskrit we
24 On the wider structural position of PS 2.5ab, see Lasic 2010: 514–522.

now can get a more nuanced picture. The opponent referred to by Til‐
lemans in (3) in fact asks how this latter cognition operating with a
non‐empirical object which is a mere concept of the object can be in‐
ference (kathaṃ tarhi tasyānumānatvam).25 Dignāga introduces his an‐
swer to this question by mentioning words such as heaven, the refer‐
ent of which we never have seen: “For (hi) words like ‘heavenʼ do not
express the mere object (arthamātra).”26 Next follows PS 2.5ab with
the Vṛtti.
āptavādāvisaṃvādasāmānyād anumānatvam / PS 2.5ab
āptavacanaṃ gṛhītvārthāvisaṃvādatulyatvāt tasyānumānatvam uktam.27
25 A
possible Sanskrit can be reconstructed as follows: adṛṣṭārtha arthavikalpamā‐
tram, na viśiṣṭārthapratītiḥ. kathaṃ tarhi tasyānumānatvam. na hi svargādiśabdair
arthamātram ucyate. (= PSVK 111a2–3: ma mthoṅ baʼi don la don du rnam par rtog
pa tsam yin gyi don gyi khyad par rtogs pa ni ma yin no // ji ltar de rjes su dpag pa
ñid yin te mtho ris la sogs paʼi sgra rnams kyis don tsam brjod ba ni ma yin no //.) The
Sanskrit is available in PSṬms 62b7: adṛṣṭārthe svargādāv arthavikalpamātram iti
(PSṬTib D92a2/P103b1–2: ma mthoṅ baʼi don la ni mtho ris la sogs pa la ste / don
du rnam par rtog pa tsam źes pa), quoted in PVSV 37,27; PSṬms 63a5: tataś ced
arthavikalpamātraṃ bhavati, na viśiṣṭārthapratītiḥ (PSṬTib D92a7/P103b7–8: de
las gal te don rnam par rtog pa tsam du ʼgyur gyi don gyi khyad par rtogs par mi
ʼgyur na); PSṬms 63a4: kathaṃ tarhi tasyānumānatvam iti (PSṬTib D92a6/
P103b7: ʼo na ci ltar de rjes su dpag pa yin źe na źes pa /); PSṬms 63a6: āha – na
hītyādi (PSṬTib D92b1–2/P104a2: bśad pa / ma yin źes pa la sogs pa ste /); PSṬms
63a7: tasmān nāptaprayuktaiḥ svargādiśabdair arthamātram ucyate. kiṃ tarhi.
laukikaśabdāsādhāraṇe bāhye ʼpy artha iti (PSṬTib D92b2–3/P104a3–4: deʼi phyir yid
ches pas rab tu sbyar baʼi mtho ris la sogs paʼi sgra rnams kyi (read: kyis) don
tsam brjod pa ma yin gyi / ʼo na ci źe na / ʼjig rten paʼi sgra daṅ thun moṅ ma yin
paʼi phyi rol gyi don yaṅ ṅo źes paʼo //).
26 na hi svargādiśabdair arthamātram ucyate, see above, n. 25.
27 For the reconstruction of the Sanskrit text (following mainly the Tibetan translation
of Vasudhararakṣita), see below, p. 94. The Tibetan translation reads:

yid ches tshig kyaṅ mi slu bar // mtshuṅs phyir rjes su dpag pa ñid // PS 2.5ab
yid ches paʼi tshig ñid bzuṅ nas kyaṅ mi bslu bar mtshuṅs paʼi phyir de yaṅ rjes su
dpag pa ñid du brjod do // PSVV 29a2–3.
yid ches tshig ni mi slu ba // spyi las rjes su dpag pa ñid // PS 2.5ab
yid ches paʼi tshig ñid bzuṅ nas don de la mi bslu baʼi phyir daṅ / mi ʼdra baʼi phyir
rjes su dpag pa ñid du bśad pa yin te // PSVK 111a3–4.
94 HELMUT KRASSER
... (PS 2.5ab). Having grasped the statement of a credible person, it (tasya)
has been said to be inference, because [it] is equal in not belying with re‐
gard to the object.
If I understand Dignāgaʼs wording in the commentary correctly, this is

the answer to the initial question kathaṃ tarhi tasyānumānatvam,
referring to our cognition of svarga etc., the object of which never has
been seen by us but only by the āptas. This interpretation goes
smoothly with Jinendrabuddhiʼs comments when he says that the ques‐
tion, now reformulated as kasmāt punas tasyānumānatvam to conform
to the wording he has previously used,28 has to be answered by arthā‐
visaṃvādatulyatvāt. Jinendrabuddhiʼs comments read,
āptavacanaṃ gṛhītvetyādi. āptavacanaṃ gṛhītvāvisaṃvādādhigamahe‐
tuṃ tasyānumānatvam uktama. kasmāt punas tasyānumānatvam arthā‐
visaṃvādatulyatvād iti vyākhyeyam. PSṬms 63b5–6 (yid ches paʼi tshig
bzuṅ nas źes pa la sogs pa ste yid ches paʼi tshig bzuṅ nas mi bslu bar rtogs
paʼi rgyur / de rjes su dpag pa ñid du gsuṅs so // ciʼi phyir de rjes su dpag
pa ñid yin źe na / don la mi bslu bar mtshuṅs pa ñid kyi phyir źes bśad
par byaʼo // PSṬTib D93a1–2/P104b2–4)
a uktam em. (gsuṅs so PSṬTib) : yuktaṃ PSṬms
“Having grasped the statement of a credible person” and so forth. It

(tasya) has been said to be an inference, having grasped the statement of
a credible person which is a logical reason (hetu) for an understanding
that [it] is not belying (avisaṃvāda). [The question] why now this (tasya)
is an inference is to be explained by “because [it] is equal in not belying
with regard to the object” (arthāvisaṃvādatulyatvāt).
tasya in Jinendrabuddhiʼs explanation thus would have the same mean‐

ing as it has in Dignāgaʼs initial question and refer to the inference
which is a mere concept, which would correspond to Dharmakīrtiʼs
buddheḥ in PV 1.216 or pratipatteḥ in the Svavṛtti. And Dharmakīrtiʼs
āptavādāvisaṃvādasāmānyāt of PV 1.216, finally, would be his
rephrasing of Dignāgaʼs arthāvisaṃvādatulyatvāt. Thus, I propose to
28 PSṬms 63a4: kathaṃ tarhi tasyānumānatvam iti (PSṬTib D92a6/P103b7: ʼo na ci
ltar de rjes su dpag pa yin źe na źes pa /); see above, p. 93 n. 25.

understand PV 1.216 together with the Svavṛtti as commenting upon

Dignāgaʼs Vṛtti on PS 2.5ab and not on PS 2.5ab itself.29
2.4. On sāmānyāt
The various interpretations of sāmānyāt have already been addressed
by Lasic (2010: 511–514), whose main concern is a possible inter‐
pretation of PS 2.5ab in its own right, not Dharmakīrtiʼs interpretation.
Thus, he does not offer a solution for how to understand it in Dharma‐
kīrti.
2.4.1 On the meaning of the ablative ending

The ablative ending of sāmānya has been translated in two ways: (1) as
“insofar,” etc., and (2) as “because,” etc. Hayesʼ (1980: 252, 1988: 238)
interpretation of the ablative as “insofar” has been refuted convinc‐
ingly by Tillemans (1990: I.20–21).30 The interpretation of the ablative
as “because,” “since,” and the like, to which I subscribe, is favoured by
the majority of scholars.
2.4.2. On the meaning of sāmānya

Also regarding the meaning of sāmānya there are two camps. One takes
it as “similar” or “equal,” etc., the other understands it as “in general” or
“generally.” The latter translation has been proposed by Yaita (1987: 6,
8) and van Bijlert (1989: 124), and it is also the one preferred by
Dunne. Dunne (2004: 363–364 n. 10) argues for this interpretation
against the understanding of Śākyabuddhi in order to avoid a contra‐
diction with PV 1.218. He does so even at the risk of attributing to
Dharmakīrti a form of fallacious reasoning that Dharmakīrti himself
refers to as śeṣavadanumāna – and which, as we will see later, also per‐
tains to Dignāga’s definition and Dharmakīrti’s interpretation of it, as
29 It goes without saying that, disregarding the context, tasya in PSV as well as in PSṬ
easily could be understood as referring to āptavacana. In that case, however, we

have to assume that the question kathaṃ tarhi tasyānumānatvam remains comple‐
tely ignored.
30 This refutation holds good also for van Bijlertʼs interpretation “in so far” (1989:
122).
96 HELMUT KRASSER
well as to Śākyabuddhiʼs comments – by apparently taking the term “in

general” in the stronger sense of “universally, without exception.” He
writes, “On this interpretation, the argument is that, since the state‐
ments of a particular author have been observed to be trustworthy in
terms of observable objects, this general trustworthiness may be ex‐
tended to unobservable objects.” This understanding leads Dunne to
the following translation of PVSV 109,7–9 (on k. 216, see above p. 90):
These kinds of statements by a credible person – those [described by Dig‐
nāga] and those [accordingly delineated above] – are generally trustwor‐
thy. Hence, not observing those statements to be misleading, one infers
the trustworthiness of a cognition of those statementsʼ object, even
though it is not knowable by perception and empirical inference.31
(Dunne 2004: 364)
This translation is fine, given that sāmānyāt means “generally.” Dhar‐

makīrti here speaks of a sāmānya in terms of avisaṃvāda of two kinds
of statements: tasya cāsyaivambhūtasyāptavādasya.32 The first one re‐
ferred to by tasya is, according to Karṇakagomin, the āgama said by
Dignāga to be anumāna. The second one addressed by asyaivambhūta‐
sya is the one that, in accordance with what has been said in k. 214, has
been found to be coherent (sambaddha) and to teach a proper human
aim (puruṣārtha) as well as a suitable means for obtaining it (anuguṇo‐
pāya), which then has been tested and ascertained to be trustworthy as
described in k. 215 with PVSV.33 The āgama at stake in Dignāga con‐
cerns heaven (svarga), which is not accessible to our knowledge. Thus,
31 PVSV 109,7–9: tasya cāsyaivambhūtasyāptavādasyāvisaṃvādasāmānyād adṛṣṭa‐
vyabhicārasya pratyakṣānumānāgamye ʼpy arthe pratipattes tadāśrayatvāt tadanya‐

pratipattivad avisaṃvādo ʼnumīyate.
32 Here one would expect another ca: evambhūtasya ca. But this is not supported by
the commentaries. PVSVTib D322b6–7/P479a7 (de daṅ ʼdi de lta bur gyur paʼi ñes pa
zad paʼi tshig) suggests the reading: tasyāsya caivambhūtasya or tasyāsyaivambhūta‐
sya ca. The text either needs to be corrected or we have to suppose that ca is mis‐
placed (bhinnakrama).
33 PVSVṬ 394,9–10: tasyāgamasyācāryadignāgena nirdiṣṭānumānabhāvasya. asyety
asmābhisa sambaddhānuguṇopāyamb ityādinā vicāritasya.

a asmābhiḥ refers to Dharmakīrti; cf., e.g., the comment on atrocyate (PVSV
39,30): atra sāmānyalakṣaṇe ʼsmābhir ucyate (PVSVṬ 174,21).

b sambaddhānuguṇopāyam em. (PV 1.214) : sambandhād anuguṇopāyam PVSVṬ,
PVSVṬms 142a1.
Dharmakīrti would be saying that “these kinds of statements by a

credible person – those [that speak about heaven] and those [accord‐
ingly delineated above] – are generally trustworthy.” This implies that
“statements speaking about heaven are generally trustworthy,”
whereby “generally trustworthy” means “checked by the threefold
test.” In order to avoid this problematic implication one can assume, of
course, that Dharmakīrti with tasya does not mean anything special
and only wants to inform us that Dignāga considers even that part of
āgama he isn’t specifically concerned with to be trustworthy.
Be that as it may be, it is in any case not clear why general trustwor‐
thiness justifies Dignāgaʼs holding the āgama that speaks of heaven to
be anumāna. This leads us to the question, On what basis can Dignāga,
according to Dharmakīrti, claim āgama to be anumāna, even when
sāmānya is understood as “similar” or “equal,” etc., in terms of its being
avisaṃvāda? First, however, one must answer the question, Similar to
what? It seems there are two possibilities: (a) either leave it open, as I
have done in the translation of PVSV 108,1–2 (see above, p. 84), or (b)
take it specifically as “similar to inference.” Those who have opted for
the first alternative are Hayes (1988: 238), “insofar as they have the
common character of not being false,” Tillemans (1990: 20), “because
authoritative words are similar in not belying” (similarly, Tillemans
1993: 10, 11), and Eltschinger (2007a: 219), “est semblable en fiabil‐
ité.” Yet the same scholars have also on occasion opted for the second
alternative: Hayes (1980: 252), “insofar as they have (in) common
(with inference the) character of not being false,” Tillemans (1986: 32),
“because authoritative speech (āptavāda) is similar [to an inference] in
being infallible,” and (1990: 22), “they are similar to normal inference
because they too are non‐belying,” and Eltschinger (2007a: 224), “est
semblable [à lʼinfé rence] en iabilité .”
Thus, the statement of a credible person with regard to transcend‐
ent objects such as heaven is either said to be inference on account of
the general trustworthiness of the words of this credible person re‐
garding perceptible and inferable matters, or on account of their being
similar to normal inference in not being belying. Both interpretations
are far from being self‐evident, nor is there any commentarial support
for them, as far as I can tell. There are, however, clear statements by
the commentators that Dignāga in PS 2.5ab intended what Dharmakīrti
would later call a svabhāvahetu and that anumāna should be under‐
98 HELMUT KRASSER
stood in the sense of logical reason (liṅga, hetu).34 This did not escape
Dunne, who provides the relevant material in the footnotes to his
translation. The commentators also tell us that the equality or similar‐
ity is between the statements of a credible person concerning matters
that can be checked by perception and inference and transcendent
matters such as heaven that we cannot check. Already the comments
by Śākyabuddhi (followed by Karṇakagomin) on the opening question
at PVSV 108,1–2 are quite telling.
Any statement of a credible person is non‐belying, like [a statement] such
as “All conditioned things are momentary.” And this is a statement with
regard to a transcendent object. Thus, this too is non‐belying. Because the
statement of a credible person is non‐belying (avisaṃvāditvāt, [PVSVṬ,
no equivalent in PVṬ]) as its character of being non‐belying is the
same in the way [explained] (evam), it is inference. Thus (iti), the
teacher Dignāga said that scripture is inference with regard to the exter‐
nal object.35 (Also translated in Dunne 2004: 361 n. 2)
Śākyabuddhi’s and Karṇakagomin’s comments on PV 1.216 are also

quite clear on the point that the character of being non‐belying in the
case of a credible personʼs statement concerning transcendent matters
is equal to or the same as this personʼs statement concerning matters
that can be checked. And it is for this reason that āptavāda serves as an
34 For āptavacana as a kāryahetu see Tillemans’ explanation on PV 4.92 (Tillemans
2000: 126ff.).
35 PVṬ Je D242b3–5/P285b6–286a1: gal te phyi rol gyi dṅos po la sgra tshad ma ñid
yod pa ma yin na / ʼon kyaṅ gaṅ ʼdi yid ches tshig ni mi slu ba spyi las (PVSV
108,1) te gaṅ daṅ gaṅ ñes pa zad paʼi tshig de daṅ de ni mi slu ba yin te / dper na ʼdus
byas thams cad ni skad cig ma źes bya ba la sogs pa lta buʼo // śin tu lkog tu gyur paʼi
don la ñes par zad paʼi tshig ʼdi yaṅ yin no // de bas na ʼdi yaṅ mi slu ba yin no źes bya
ba de lta na ñes pa zad paʼi tshig ni mi slu ba spyi las rjes su dpag pa ñid yin pa de ltar
na / slob dpon gyi (read: gyis) luṅ ni phyi rol gyi don la rjes su dpag pa ñid du gsuṅs
pa yin no // ≈ PVSVṬ 390,15–19: yadi bāhye vastuni śabdasya nāsti prāmāṇyam, yat
tarhīdam āptavādāvisaṃvādasāmānyāt. yo ya āptavādaḥ, so ʼvisaṃvādī, yathā
kṣaṇikāḥ sarve saṃskārā ityādikaḥ. āptavādaś cāyam atyantaparokṣe ʼpy arthe.
tasmād ayam apy avisaṃvādītyevam āptavādasyāvisaṃvādasāmānyād avisaṃvā‐
ditvād anumānatety āgamasya bāhye ʼrthe ʼnumānatvam uktam ācāryadignāgena.
inferential mark36 on account of which we can infer that our cognition

does not belie.
“Because that credible personʼs speech is the same in its trustworthiness
(avisaṃvādasāmānyāt). That is, just as the credible personʼs speech is
trustworthy with regard to an object that can be determined [through
perception and ordinary inference], likewise it is trustworthy with regard
to an extremely remote object also, precisely because it is the speech of a
credible person.” (Dunne 2004: 364 n. 10) It is therefore that the teacher
Dignāga termed an inference a cognition arising from an inferential mark
characterized as [this] personʼs credible speech that [such a cognition] is
not belying with regard to a real state of affairs (don la mi slu ba ... blo ni,
*arthāvisaṃvādabuddheḥ).37
This interpretation of Śākyabuddhi and Karṇakagomin is in accord

with what Dharmakīrti himself explained in PVSV 109,8–11 when com‐
menting on PV 1.216 (see above p. 89). Thus, I can see no harm in ac‐
cepting this interpretation and translating PS 2.5ab from the viewpoint
of Dharmakīrti as follows:
Because the character of being non‐belying of a statement of a credible
person is the same [in the case of transcendent objects as in the case of
objects that are accessible to us, it] is an inference [in the sense of an in‐
ferential mark].
sāmānyāt in PV 1.216 might be understood accordingly. It goes without

saying that for the time being such an understanding is justified only
36 Jinendrabuddhi (PSṬms 63b5–6, referred to above, p. 94) also explains the statement
of a credible person to be a logical reason: āptavacanaṃ gṛhītvāvisaṃvādādhigama‐

hetuṃ tasyānumānatvam uktam.
37 PVṬ Je D245a7–b1/P289b5–7: ñes pa zad paʼi tshig de ni mi slu bar mtshuṅs paʼi
phyir te / ci ltar mṅon sum daṅ rjes su dpag pas mi slu baʼi don yoṅs su gcod par nus
pa la ñes pa zad paʼi tshig mi slu ba de ltar śin tu lkog tu gyur pa (read: pa la) yaṅ yin
te / ñes pa zad paʼi tshig ñid yin paʼi phyir ro // de bas na don la mi slu ba ñes pa zad
paʼi tshig gi mtshan ñid can gyi rtags las byuṅ paʼi blo ni rjes su dpag pa ñid du slob
dpon gyis brjod do // ≈ PVSVṬ 393,25–28: tasyāptavādasyāvisaṃvādasāmānyāt.a
yathā śakyaparicchede ʼrthe āptavādasyāvisaṃvādaḥ, tathātyantaparokṣe ʼpi, āpta‐
vādatvād eva. tataś cāptavādalakṣaṇāl liṅgād utpannāyā avisaṃvādabuddher anu‐
mānatācāryadignāgenābhihitā.
a tasyāptāvādasyāvisaṃvādasāmānyāt PVSVṬ 141b6 (PVṬ) : tasyās tāvad asyā‐
ms
visaṃvādāt sāmānyāt PVSVṬ.
100 HELMUT KRASSER
for Dharmakīrti. Whether this interpretation is faithful to Dignāgaʼs un‐

derstanding still remains to be investigated.
2.5. PV 1.217 – a second interpretation of PS 2.5ab

The above interpretation probably is the best Dharmakīrti could come
up with without contradicting Dignāgaʼs wording, although I cannot
see anything in Dignāgaʼs phrasing that would suggest that a scripture
must be tested in its entirety, as proposed by Dharmakīrti. Perhaps be‐
cause this is hardly possible Dharmakīrti provides us with a second,
less demanding interpretation of Dignāgaʼs āgama definition, accord‐
ing to which a person’s reliability with regard to transcendent objects
is inferred from his knowledge of the main religious teachings that he
presents, such as the Four Noble[ʼs] Truths.38
heyopādeyatattvasya sopāyasya prasiddhitaḥ /
pradhānārthāvisaṃvādād anumānaṃ paratra vā // PV 1.217
heyopādeyatadupāyānāṃ tadupadiṣṭānām avaiparītyam avisaṃvādaḥ. ya‐

thā catūrṇām āryasatyānāṃ vakṣyamāṇanītyā. tasyāsya puruṣārthopayo‐
gino ’bhiyogārhasyāvisaṃvādād viṣayāntare ’pi tathātvopagamo na vipra‐
lambhāya, anuparodhān niṣprayojanavitathābhidhānavaiphalyāc ca vak‐
tuḥ. PVSV 109,15–19
Or [scripture] is inference with regard to the other [domain] due to its

being not belying with regard to the principal points [i.e. the Four No‐
ble(ʼs) Truths], because the nature (tattva) of what is to be abandoned
and what is to be realized together with their means is well
established [by it].
38 I consider the method proposed in PV 1.217 as an alternative to the threefold analy‐
sis, i.e., the second possibility proposed by Dunne (2004: 241): “This test either sup‐
plements or perhaps replaces … the threefold analysis.” Yaita and Tillemans consi‐
der it differently: “On the basis of credibility of āgama examined like that, Dharma‐
kīrti proves the non‐disagreement of the āgama even with regard to inaccessible
thing talked of by the author, in the following two ways (v. 216–7)” (Yaita 1987: 3);
“(e) v. 217 elaborates upon aspects of v. 216: when the scripture is non‐belying on
important rationally accessible things it should also be so on the inaccessible
things” (Tillemans 1999a: 400 = 1999b: 42).
Reliability consists in the fact that what is to be obtained and what is to

be avoided [together with] their means, which has been taught by that
[credible person], corresponds to reality (avaiparītya); like the Four No‐
ble(ʼs) Truths in the way it will be explained [in the Pramāṇasiddhi‐chap‐
ter]. Because that very (tasyāsya) [thing that has been taught], which
serves the human goal, [and hence] is suitable to being practiced, is relia‐
ble, the assumption that this is so also in the case of the other, transcend‐
ent realm may not lead to oneʼs deception. [And this is for two reasons:]
1) because there is no counter evidence (anuparodha), and 2) because it
is pointless for a speaker to make false statements without a purpose.
2.6. Conclusion of this excursus on PS 2.5ab

What we have seen so far certainly does not substantiate my claim that
for Dharmakīrti scripture is not a means of valid cognition. On the con‐
trary, we learn that the statements of a credible person and inferences
based on a scripture that has either been well checked or expounds the
main religious teachings, are reliable even with regard to real states of
affairs and not just the speaker’s intention. Thus – to invert a common
saying – what you get is what he has seen.
However, before refuting other ways of accounting for the reliability
of credible persons offered by opposing schools (PV 1.218ff.), Dhar‐
makīrti concludes his interpretation of Dignāgaʼs de inition with the
following words:
tad etad agatyobhayathāpy anumānatvam āgamasyopavarṇitam – varam
āgamāt pravṛttāv evaṃ pravṛttir iti. na khalv evam anumānam anapāyam,
anāntarīyakatvād artheṣu śabdānām iti niveditam etat. PVSV 109,20–22
Thus, this scripture has been explained in both ways to be inference for
want of [any other] possibility (agatyā), [having in mind:] “Given that one
has to proceed on account of scripture it is still better (varam) to proceed
in such a way.” However, inference in such a way indeed is not without
problems (na ... anapāyam), for words are not invariably concomitant
with [their] objects. This [we already] have explained [in PV 1.213].
Here we learn, if my interpretation of agatyā is correct, that scripture
can only be said to be anumāna (in the sense of an inferential mark
[liṅga] or logical reason [hetu]), because there is no better way to de‐
fine it and because it is still better to proceed on the basis of a checked
102 HELMUT KRASSER
or “purified” āgama than just blind faith, e.g., in the Veda. But, in which‐
ever way this definition is to be understood, to infer the reliability of a
credible personʼs statement with regard to transcendent matters is
problematic. Thus, although Dignāgaʼs de inition is in contradiction
with Dharmakīrtiʼs theory that words are not connected to their ob‐
jects, Dignāga nevertheless did the best he could under the circum‐
stances that we require scripture to live our lives and, thus, ideally
should rely on one that is optimally reliable. What kind of problem
Dharmakīrti had in mind, when he says na ... anapāyam, he does not
share with us here.
3. OTHER PV(SV) PASSAGES ON ĀGAMA

Alas, it is only much later that Dharmakīrti tells us specifically what
kind of problem he had in mind. This kind of inference where one in‐
fers correctness with regard to transcendent matters from the correct‐
ness of statements in one part is a fallacious inference called a śeṣavad‐
anumāna. Moreover, we know that Dignāga was aware of this type of
fallacy, because in his Pramāṇasamuccaya he refutes the śeṣavadanu‐
māna of the Naiyāyikas. Dharmakīrti refers, without actually saying so,
to PS 2.28b with Vṛtti where Dignāga refutes the Naiyāyikas using the
same examples as Dharmakīrti. The main difference between these two
kinds of śeṣavadanumāna applied by the opponent, in this case a Mī‐
māṃsaka, and by the Buddhist is that the Buddhist scripture is “puri‐
fied” by an examination and thus it might sometimes be reliable. Thus,
in this case the probability that it is true is greater, but there is no cer‐
tainty.
But another [theorist], renouncing the definition of scripture as [that
which is] authorless [because of a similar criticism, and] wishing to prove
the authority of the Veda in another way, says, “[Those] Vedic sentences
for which there is no cognition [on the part of Buddhists as being true]
are [in fact] true, because they are part of the [same] Veda, just like the
sentence ‘Fire is the remedy for cold,ʼ etc.” ...
... This kind of inference was declared to be not [really] a proof by the
master [Dignāga] himself in pointing out the deviating character of the
Naiyāyikasʼ śeṣavat‐inference [in PS(V) 2.28b],39 like the [so‐called] proof

that fruit [one has not tasted] has the same taste [as fruit one has tasted]
because it has the same colour, and the [so‐called] proof that rice grains
one has not observed are cooked, like those which one has observed, be‐
cause they are in one [and the same] vessel. And the manner in which this
[śeṣavat‐inference] is not a proof has [already] been stated earlier [in PV
1.14 with PVSV]. And [true,] we have stated this definition of scripture
[too]. However, this [is justified only] if, for every object capable of being
examined, there is correctness (viśuddhi) of positive and negative asser‐
tions by appropriate means of valid cognition. [And] even if there is no
necessary relation between words and [their] meanings [, which would
ensure the validity of scripture], it is better that a [person] act in [a state
of] doubt [when it comes to matters relating to worldly prosperity and
salvation];40 for [scripture] may occasionally be reliable in this case.41
Why there can be no certainty Dharmakīrti has already stated before,

where he also referred back to the definition of scripture I have dealt
with at the beginning of this paper and to his remark that such an in‐
ference is not without problems.
39 PSVK 115b2–4: lhag ldan la yaṅ dpe byed na // (= PS 2.28b: śeṣavaty api ced vatiḥ)
lhag ma daṅ ldan pa la yaṅ gal te dpe byed pa yin na / lhag mṅon sum daṅ mtshuṅs
paʼi yul gaṅ la yod pa deʼi śes pa ni lhag ma daṅ ʼdra baʼo źes bya ba ʼdi yaṅ ʼkhrul paʼi
phyir tshad ma ma yin te / gzugs mtshuṅs pas ro la sogs pa gdon mi za bar mtshuṅs
par ʼgyur ba ni ma yin no // de ltar na lhag ma daṅ ldan pa la yaṅ dper mi rigs pa yin
no //. See above, p. 62 and n. 126.
40 The idea that one acts in regard to transcendent objects from a state of doubt once
the scripture is purified is also propounded by Śāntarakṣita and Kamalaśīla; see

McClintock 2010: 322–324.
41 PVSV 173,16–174,1: anyas tv apauruṣeyam āgamalakṣaṇaṃ parityajyānyathā prā‐
māṇyaṃ vedasya sādhayitukāmaḥ prāha – avitathāni vedavākyāni yatrāpratipattiḥ,

vedaikadeśatvāt, yathāgnir himasya bheṣajam ityādi vākyam iti. ... svayam īdṛśam
ācāryeṇānumānaṃ naiyāyikaśeṣavadanumānavyabhicāram udbhāvayatā tulyarū‐
patayā phalānāṃ tulyarasasādhanavad ekasthālyantargamād dṛṣṭavad adṛṣṭataṇḍu‐
lapākasādhanavac cāsādhanam uktam. tadasādhanatvanyāyaś ca pūrvam evoktaḥ.
uktaṃ cedam āgamalakṣaṇam asmābhiḥ. tat tu sarvasya śakyavicārasya viṣayasya
yathāsvaṃ pramāṇena vidhipratiṣedhaviśuddhau nāntarīyakatvābhāve ʼpi śabdānām
artheṣu varaṃ saṃśayitasya vṛttiḥ, tatra kadācid avisaṃvādasambhavāt… On this
passage, see above, pp. 62–64.
104 HELMUT KRASSER
Objection: Isnʼt it the case that such a thing as the arrangement of the
world, even though it is not an object [accessible] to reason, is known [by
you Buddhists] from the statement of a person which must be assumed
[to be true, on the basis of his reliability in regard to other things]? [An‐
swer:] No, because [we have] no confidence [in such a person]. It is in‐
deed not the case that, since [a person has been observed] not to err with
respect to a certain [matter], everything [that person says] is like that [i.e.
true, and this for two reasons: first], because one observes that [people
who are known to be reliable in regard to a certain thing do in fact] err
[in regard to other things]; and [second,] because a concomitance be‐
tween the [verbal] activity of this [allegedly superior person] and relia‐
bility is not established. Beyond that, the [aforementioned] definition of
scripture has been accepted for lack of [any other] recourse. There is [in‐
deed] no ascertainment [of supersensible things] from [scripture thus de‐
fined, and] this is the reason why [we have] also stated [above] that scrip‐
ture is not a means of valid cognition.42
These inferences are mainly uncertain, because one cannot establish a

pervasion between scripture and its reliability, and only such a perva‐
sion could guarantee the certainty of the inference. Therefore, Dhar‐
makīrti repeats that this definition was assumed only in want of a bet‐
ter possibility and that there is no certainty from scripture. And this is
the reason why he also said that āgama is not a pramāṇa.
In the last chapter of his PV Dharmakīrti refers to these passages.
prāmāṇyam āgamānāṃ ca prāg eva vinivāritam / PV 4.101ab
Now, it had already been refuted earlier that scriptures were pramāṇas.
(Tillemans 2000: 141)
So far we have seen that by reading bits and pieces from different dis‐
cussions of āgama in the PVSV together the impression that we gained
from Dharmakīrtiʼs interpretation of Dignāgaʼs de inition of scripture,
namely that it is a real pramāṇa with regard to external objects such as
42 PVSV 167,23–168,2: nanu kaścil lokasanniveśādir ayuktiviṣayo ʼpi sambhāvanīyapu‐
ruṣavacanād arthaḥ pratipadyate. na, apratyayāt. na hi kvacid askhalita iti sarvaṃ

tathā, vyabhicāradarśanāt, tatpravṛtter avisaṃvādena vyāptyasiddheś ca. agatyā
cedam āgamalakṣaṇam iṣṭam. nāto niścayaḥ. tan na pramāṇam āgama ity apy
uktam. On this passage, see also above, pp. 42–44. The passage beginning with na hi
kvacit is also translated in Tillemans 1999a: 400; see also Kataoka 2011b: 256 n. 10.
heaven, has now changed and it has become clear that āgama is not a
pramāṇa at all. It is only under certain circumstances, namely when it
has been completely checked and purified, that there might be a chance
of proceeding successfully based on āgama. Or else, when an āgama or
āptavāda expounds the main teachings, as we have seen in PV 1.217
with PVSV. What Dharmakīrti formulates in PV 1.217 and his commen‐
tary thereon serves as a leitmotif for his second chapter entitled
Pramāṇasiddhi, which is entirely devoted to the proof that the Buddha
is pramāṇa.43 But in order to avoid the mistake of employing a śeṣavad‐
anumāna, the Buddha is proved to be pramāṇa only metaphorically.
Just as a pramāṇa qua cognition is not belying (pramāṇam avisaṃvādi
jñānam PV 2.1ab1) and makes known a hitherto unknown object (ajñā‐
tārthaprakāśo vā PV 2.5c), in the same way the Buddha is not belying
with regard to and makes known the Four Noble(ʼs) Truths, which hi‐
therto were unknown to us.44
43 PV 2.32 (cf. PV 1.217): heyopādeyatattvasya sābhyupāyasya (: hānyupāyasya Miyasa‐

ka ed.) vedakaḥ / yaḥ pramāṇam asāv iṣṭo na tu sarvasya vedakaḥ //. “He who makes
known the nature of what is to be obtained and what is to be avoided, together with
their means, is assumed to be a pramāṇa, but not he who makes everything known.”
PV 2.145–146ab: tāyaḥ svadṛṣṭamārgoktir vaiphalyād vakti nānṛtam / dayālutvāt
parārthañ ca sarvārambhābhiyogataḥ // tataḥ pramāṇam, tāyo vā catuḥsatyaprakā‐
śanam /. “‘Protection’ [i.e., causing people to cross,] is stating the path seen by him‐
self. He does not tell a lie, (1) because it is pointless, (2) because he is compassion‐
ate, and (3) because he perseveres in all his undertakings for the sake of others.
Therefore [he is] a pramāṇa. Or, protection is illuminating the four Noble(’s)
Truths.” On these one‐and‐a‐half verses, see, e.g., Kataoka 2011b: 258–259 with n.
20.
44 See Krasser 2001. It goes without saying that I do not agree with Franco who con‐
cludes his very learned chapter “The Framework and Proof‐Strategy of the Pramā‐
ṇasiddhi‐Chapter” by comparing Dharmakīrtiʼs strategy with that of Pakṣilasvāmin
Vātsyāyana. Dharmakīrti was well aware of the fact that Dignāgaʼs way of under‐
standing āgama was similar to that of Pakṣilasvāmin and that both operated with a
false śeṣavadanumāna, a mistake that Dharmakīrti avoided. Franco (1997: 39–40):
“To conclude, I have tried to show that there are strong resemblances and parallel‐
isms between Dharmakīrtiʼs proof of the validity of the Buddhist āgama, which is
based on the Buddhaʼs being pramāṇabhūta, which is based in its turn on the esta‐
blishment of the four noble truths, and Vātsyāyanaʼs proof of the validity of the
Veda, which is based on its composition by āptas, whose status is based in its turn
on the effectiveness of the statements of the Āyurveda, the mantras, etc. I would
even go so far as to say that although structurally the Pramāṇasiddhi‐chapter is con‐
strued as a commentary on the five epithets of the Buddha, the interpretation of the
106 HELMUT KRASSER
4. MORE ON PVSV 108,1–109,22

These two accounts of āgama, according to which, on the one hand, it
cannot be counted as a pramāṇa at all and according to which, on the
other, it yields correct knowledge even of external objects, seem com‐
pletely incompatible to me. Yet we find them in the work of one and
the same author. Thus, to which view did he really subscribe?
The depiction of scripture as having the special status as an infer‐
ence, but an inference that is not without problems, leads Dunne to an
interesting description of Dharmakīrtiʼs approach to it.
A significant feature of this argument is the tension that it evidences be‐
tween an appeal to oneʼs own empirical knowledge as opposed to trust‐
ing in anotherʼs (i.e., the Buddhaʼs) transempirical knowledge.... Never‐
theless, despite this claimʼs empiricist tones, Dharmakīrti is apparently
troubled by his own appeal to the Buddhaʼs transempirical knowledge
and concomitant extraordinary qualities. How else can we explain the fact
that he immediately backpedals by denying that scriptural inference is
really an inference at all (6)? Moreover, he then proceeds to reject explic‐
itly any appeal to credibility (7). (Dunne 2004: 241f.)
Tillemans, on the other hand, who is well aware of the passages ad‐
duced above in §3, considers scripture as depicted in the PV(SV)
108,2–109,19 to be an “exception.” In his summary of PV 1.216 he
writes,
(d) v. 216: Dignāgaʼs point in saying that authoritative words were an
inference was that when an authorityʼs words (= scripture) have been
found to be non‐belying on rationally decidable matters, then we are jus‐
tified to understand radically inaccessible matters based on that scrip‐
ture. In this particular case, contrary to what was said in v. 213, we do in‐
fer something more than just the speakerʼs intention from his words: we
also infer that the state of affairs obtains. This one ‘exception’ to v. 213
must be allowed because otherwise there would be no way (agatyā) for
us to come to know radically inaccessible things. (Tillemans 1999a: 399–
400 = 1999b: 41–42)
epithets and the general strategy employed to prove the validity of the Buddhist
teachings owe more to Vātsyāyana than to Dignāga.” On Pakṣilasvāmin, see below,
p. 110 n. 49.
And in his introduction to the same paper, which summarizes Dharma‐

kīrtiʼs position on āgama, we read,
It is well known that Dharmakīrti commented upon the phrase āptavādā‐
visaṃvādasāmānyād anumānatā in Dignāga as showing that quotations
from authorities could be used to prove certain propositions inferentially.
In particular, use of scripture or scriptural tradition (āgama) was suppos‐
edly not a separate means of valid cognition (pramāṇa), but was said to
be an inference because similar to other inferences in reliably represent‐
ing its object, i.e. in ‘not belying’ (avisaṃvāda). This then is āgamāśritānu‐
māna, ‘scripturally based inference’. Of course, not just any scripture can
be used as a basis for such an inference, nor can such inferences pertain
to any and all objects. The main criterion separating reliable from bogus
scriptures is the threefold analysis whereby it is determined that ...
There follows a description of the threefold analysis. Then Tillemans

goes on to say,
... Now, if the scripture passes this triple test, it is fit to be used in ‘scrip‐
turally based inferences’, but with the all‐important stipulation that such
inferences are only to be used in the case of radically inaccessible matters
(atyantaparokṣa), ones which are not in the domain of observation or
objective reasoning, but are only accessible once we have accepted (abhy‐
upagata) scripture. In short, āgamāśritānumāna works where objective
inference and observation leave off. ... This interpretation of Dhar‐
makīrtiʼs account of scripturally based inferences, which we shall term
for short ‘inference‐like‐any‐other’, is not just a hypothetical possibility.
(Tillemans 1999a: 395–396; not reprinted in 1999b)
The questions I want to address here are:

 Was Dharmakīrti “troubled by his own appeal to the Buddhaʼs trans‐
empirical knowledge”?
 Does Dharmakīrti hold the position that this “one ‘exceptionʼ to v.

213 must be allowed because otherwise there would be no way
(agatyā) for us to come to know radically inaccessible things”?
108 HELMUT KRASSER
 Is this “inference‐like‐any‐other”45 more than “just a hypothetical

possibility” and can it “be used to prove certain propositions inferen‐
tially”?
Let me begin with the second question by examining the meaning of

agatyā.
4.1. On agatyā
agatyā in the context of the āgama discussion has been used by Dhar‐
makīrti three times: once in PV 1.216, once in PVSV 109,19, and once in
PVSV 168,1. It has been translated variously as follows (the passages
are given approximately in chronological order):46
PV 1.216 (see above p. 88)
 “for [otherwise] there would be no way [to know such objects]” (Til‐
lemans 1986: 32, 1990: I.25, 1993: 11; and similarly Dunne 2004:
364)
 “for there is no [other] possibility (agatyā) [to explain it]” (Yaita

1987: 8)
 “since there is no [other] possibility [but to also regard the latter

kind of statement as an inference...]” (van Bijlert 1989: 124)
 “[... mais cela, Dignāga ne lʼa dit quʼ]en raison de lʼimpossibilité [où
lʼon se trouve sinon] dʼaccé der [aux objets radicalement impercepti‐
bles]” (Eltschinger 2007a: 224)
PVSV 109,19: tad etad agatyobhayathāpy anumānatvam āgamasyopa‐

varṇitam, varam āgamāt pravṛttāv evaṃ pravṛttir iti
 “Thus [the fact that] āgama is anumāna has been described in [differ‐
ent] two ways [in v.216 & 217], owing to impossibility [of explaining
in any other manner the āgamaʼs authoritativeness, i.e. by consider‐
ing our ordinary conception:] ‘When [some act] is going to be done, it
45 If one accepts the account of sāmānyāt as proposed above in §2.4, then Tillemansʼ
expression “inference‐like‐any‐other” for this kind of inference might not be the

best choice.
46 See also McClintock 2010: 259.
is certainly (evam) better to act on the basis of āgamaʼ.” (Yaita 1987:

10)
 “This fact that scripture is an inference is asserted in both cases [i.e.

in vv. 216 and 217] because of the lack of any [other] way. If one en‐
gages oneself on the basis of scripture, it is better to engage oneself
in this fashion [on the basis of a correct scripture rather than on the
basis of one which belies].” (Tillemans 1999a: 400)
 “[Mais de chacune] des deux manières [dont on lʼa fait, ce nʼest quʼ]en
raison de lʼimpossiblité dʼaccé der [sinon aux objets radicalement im‐
perceptibles que nous avons] exposé le caractère inférentiel de
lʼEcriture, [jugeant] que tant quʼà agir à partir dʼune Ecriture, mieux
vaut agir ainsi.” (Eltschinger 2007a: 227)
 “Thus, in both cases, that the scripture is a valid inference is ex‐

plained because there is no [other] way.” (Kataoka 2011b: 256)47
PVSV 168,1–2: agatyā cedam āgamalakṣaṇam iṣṭam

 “Now, we accept this defining character of scripture for lack of any
[other] way.” (Tillemans 1999a: 400)
 “And this definition of scripture [as an inference] is accepted [by us]

because there is no other way.” (Kataoka 2011b: 256 n. 10)
Of these possibilities I prefer the one according to which this definition
was given or accepted for lack of any other way. I understand agatyā as
referring back to the initial explanation I proposed in §2.1 of why
Dignāga gave a definition at all, namely, because otherwise activity
with regard to transcendent matters is not possible. Thus, he had no
other choice than to give this definition, knowing that, though it is
faulty, it is the best definition available. But he gave, of course, the best
one that is possible. Thus, I prefer agatyā to be read with the past
participles in these phrases: agatyābhihitā (k. 216), agatyā ... upa‐
47 Kataoka is of the opinion that agatyā should exclude such possibilities as the Bud‐
dhaʼs being a liar or his teaching having no aim and so forth, by relating it to the Mī‐
māṃsaka proof of Manuʼs reliability.
110 HELMUT KRASSER
varṇitam (PVSV 109,19–20), and agatyā ... iṣṭam (PVSV 168,1–2). This
is also the approach taken by the commentators.48
If we understand agatyā as “for [otherwise] there would be no way
[to know such objects]” or something similar, then we would have to
supply the phrase “knowledge of transcendent matters,” which, though
implied in his explanation of k. 216 (arthāvisaṃvādād anumānam PVSV
109,11), Dharmakīrti tries to avoid. By introducing agatyā, understood
in this way, Dharmakīrti is better able to down‐play the obvious con‐
tradiction between his attitude towards scripture and Dignāgaʼs de i‐
nition and to arrive at the “solution” that it is still better to act towards
transcendent matters on account of doubt, for sometimes it might
work (tatra kadācid avisaṃvādasambhavāt, cf. above n. 41). Moreover,
in this way Dharmakīrti can avoid committing a śeṣavadanumāna him‐
self.49
I also do not follow Kataokaʼs hypothesis, though it is tempting, that
agatyā excludes such possibilities as the Buddhaʼs being a liar or his
teaching having no aim and so forth. Here, the context is to explain
away as much as possible the contradiction between Dignāga and
Dharmakīrti, not at all to prove that the Buddha is a pramāṇa.
Returning to the question under consideration, namely whether this
“one ‘exceptionʼ to v. 213 must be allowed because otherwise there
would be no way (agatyā) for us to come to know radically inaccessible
things” (Tillemans), I cannot see that scripture as defined by Dignāga in
the way we have seen, so that it amounts to a śeṣavadanumāna, would
48 See, e.g., Śākyabuddhi and Karṇakagomin on PV 1.216. PVṬ Je D245b2/P289b8–
290a1: de yaṅ go skabs med de rnam pa gźan gyis śin tu lkog tu gyur pa la ʼjug pa
med paʼi phyir ro // aʼjug pa yod na de ltar ʼjug paʼoa [D : ’jug par bya’o P] // ≈ PVSVṬ
393,30–394,8: sā cāgatyābhihitānyena prakāreṇātyantaparokṣe pravṛttyasambha‐
vāt, asatyāṃ pravṛttau varam evaṃ pravṛtta itia.
a cf. PVSV 108,5–6: sati pravartitavye varam evaṃ pravṛtta iti, and PVSV 109,20–
21: varaṃ pravṛttāv āgamād evaṃ pravṛttir iti.

49 This kind of śeṣavadanumāna we find implied by Pakṣilasvāmin. NBh 97,2 on NSū
2.1.68: teṣāṃ khalu vai prāṇabhṛtāṃ svayam anavabudhyamānānāṃ nānyad upa‐

deśād avabodhakāraṇam asti. “Now, when these [ordinary] living beings do not
know for themselves [through their own power of cognition those things that should
be avoided or obtained], then there is no other cause of [correct] knowledge [left for
them] than the [valid] instruction [given by the expert speaker].” Translation van
Bijlert 1989: 161.
constitute an exception to verbal knowledge for Dharmakīrti in

guaranteeing knowledge of or access to external objects such as
heaven.50 It is only when it comes to religious practice that, acting on
the basis of a purified scripture, the probability of success is higher, but
there is no certainty at all.
Connected to this problem is also the question whether such an in‐
ference based on scripture is more than “just a hypothetical possibility”
and whether it can “be used to prove certain propositions inferen‐
tially.”
4.2. On āgamāpekṣānumāna, āgamāśraya‐anumāna

From the description by Tillemans cited above (pp. 106–107) we learn
that “if the scripture passes this triple test, it is fit to be used in ‘scrip‐
turally based inferences,ʼ” also called āgamāśritānumāna. If I am cor‐
rect in assuming that what Tillemans means by āgamāśritānumāna is
what Dharmakīrti calls āgamāpekṣānumāna (PVSV 109,1), āgamāśra‐
yam anumānam (NB 3.114, PVin 3 128,4),51 or āgamāpekṣam anumā‐
nam (PV 4.48 = PVin 3.9; see Tillemans 2000: 78),52 then I must confess
I have not been able to find any passage in Dharmakīrti, where he calls
an inference based on a purified scripture that has passed the triple
test a “scripturally based inference.” In PV 1 and PVSV Dharmakīrti
uses scripturally based inferences only in the context of testing a scrip‐
50 As nicely shown by Tillemans (1993: Intro. §§ 2–4, 1999a: 396–399), later Tibetans
such as Tsoṅ kha pa or Śākya mchog ldan took such an inference based on scripture
to be a full‐fledged probative inference in which the hetu fulfils the trairūpya condi‐
tion. See, e.g., Śākya mchog ldanʼs account in the translation of Tillemans (1999a:
398 n. 5): “As for proving the [reasonʼs three] characteristics, the pakṣadharma(tā)
is established by perception, for this reason is presented to an opponent who per‐
ceptually observes the statement, ‘From giving comes wealth and from morality,
happinessʼ [i.e. he sees that the statement is indeed present in the text of the
Ratnāvalī]. As for the proof of the pervasion (vyāpti), there are the following two
[subsections]: proving anvaya and proving vyatireka. We now take up the first [i.e.
anvaya]. Take as the subject the Ratnāvalī; it is non‐belying with regard to the
propositions which it teaches, because it is a scripture [judged] immaculate through
the three [kinds] of analysis.” For Indian forerunners, see Eltschinger 2007a: 107.
51 For the text and a translation of NB 3.114, see Tillemans 2000: 95 n. 336.
52 In PVin 2.42 this kind of inference is also referred to as āgamārthāśrayā yuktiḥ, see
above, p. 55 n. 97.
112 HELMUT KRASSER
ture. He does so when first explaining the triple test in his comments
on PV 1.215.
āgamāpekṣānumāne ʼpi, yathā rāgādirūpaṃ tatprabhavaṃ cādharmam
abhyupagamya tatprahāṇāya snānāgnihotrāder anupadeśaḥ. PVSV 109,1–3
Also in the case of a scripturally based inference53 [there should be no in‐

validation (abādhana). Non‐invalidation is] as follows (yathā): Having ac‐
cepted (abhyupagamya) that demerit (adharma) consists in [defilements]
such as desire (rāgādirūpa) and [the actions (karman)] born of them (tat‐
53 The text of Karṇakagomin (PVSVṬ 393,13–17) explaining the locative (āgamāpekṣā‐
numāne ʼpi) seems to be problematic: yathātmādīnām (/) ādiśabdāt pradhāneśva‐

rādiparigrahaḥ / na hy eṣāṃ kiñcil liṅgam asti yenānumeyāḥ syuḥ / etad api pratipā‐
dayiṣyati / viśuddhe viṣayadvaye ( ) ʼtyantaparokṣe cāgamaviṣaye paurvāparyaviro‐
dhena yasmin cintāṃ pravartayati tasminn āgamāpekṣam anumānam api / abā‐
dhanam iti prakṛtam /. ātman, etc., in ātmādīnām (PVSV 109,1) constitute exam‐
ples of entities that should not be taught to be inferable in a treatise. By the word
ādi, according to Karṇakagomin, primordial matter (pradhāna) of the Sāṅkhyas and
a creator god (īśvara) are included, because there is no inferential mark on the basis
of which they could be inferred. The problem now is, whether viśuddhe viṣayadvaye
should be read with etad api pratipādayiṣyati or with the next part beginning with
atyantaparokṣe. The editor indicated viśuddhe viṣayadvaye as an unidentified quota‐
tion and read it with atyantaparokṣe. In that case viśuddhe viṣayadvaye has to be un‐
derstood together with atyantaparokṣe as the object of the investigation by means
of the scripturally based inference, as was done by Dunne (2004: 363 n. 9). Dunne,
who also provides a translation of Śākyabuddhiʼs explanation, realized that this con‐
stitutes a problem and thus introduced his translation with the statement that
Karṇakagomin “offers a somewhat less helpful comment.” Thus, I propose to read
viśuddhe viṣayadvaye with etad api pratipādayiṣyati and to correct the text to etad
api pratipādayiṣyati – viśuddhe viṣayadvaya iti (or: ityādinā). viśuddhe viṣayadvaye
would then be a reference to PV 4.50 where Dharmakīrti will explain that one can
adduce passages from a scripture as inferential marks only after one has purified
that scripture in terms of perceptible and empirically inferable objects according to
the method explained in the comments on PV 1.214, but not before – this is the
meaning of viśuddhe viṣayadvaye, etc. (see Tillemans 2000: 80). And as the oppo‐
nents can prove ātman, pradhāna, etc., only with the help of inferential marks (liṅ‐
ga) that are taught in their treatises, they have no liṅgas available as long as their
treatises have not passed the test. The treatises, however, will not pass the test as
they teach ātman, pradhāna, etc., to be inferable without there being a liṅga availa‐
ble. Thus, Dharmakīrti in PV 4.50 will explain that there is no liṅga whatsoever, on
account of which ātman, etc., could be inferred (na hy eṣāṃ kiñcil liṅgam asti yenā‐
numeyāḥ syuḥ. etad api pratipādayiṣyati).
prabhava),54 [a treatise] cannot prescribe [practices] such as the Agni‐

hotra and ablutions (snāna) as means of eliminating it.
Later, this abhyupagama is presupposed in the comments on PV 1.333

(in the verse the formulation āgamāpekṣenānumānena is used).55
anyad api pratyakṣānumānābhyāṃ aprasiddhaviparyayam āgamāśrayeṇa
cānumānena bādhitam agnihotrādeḥ pāpaśodhanasāmarthyādikam. PVSV
174,21–2356
a prasiddha em. (PVSVṬ 612,7) : prasiddhi PVSV : pratisiddha PVSVms
39b8.
[The Veda states] yet other [things] which are contrary to what is estab‐
lished by perception and inference and which are negated by inference
based on scripture, such as the capacity of the Agnihotra and [ablutions]
to purify one of sin, etc.57
Thus, if I have not overlooked a passage in which Dharmakīrti uses the

terms āgamāśritānumāna, āgamāpekṣānumāna, and the like for an
inference that is based on a purified scripture, āgamāpekṣānumana for
Dharmakīrti is not the same as the inference that is called by later Ti‐
betans “inference based on authority” (yid ches paʼi rjes dpag, see Tille‐
mans 1993: 12).
Coming back to the second and third questions posed above (see pp.
107–108), Is inference based on a purified scripture more than “just a
hypothetical possibility,” as Tillemans alleges, and can it “be used to
prove certain propositions inferentially”? I would answer “no” to both.
For Dharmakīrti such an inference is a śeṣavadanumāna, and thus I
cannot see how it should be more than “just a hypothetical possibility.”
If such inferences nevertheless are used as in the Tibetan tradition,
then the “certain propositions” that can be inferred based on a purified
scripture would be any of the propositions of that scripture regarding
transcendent matters. Once the scripture has been established to be
54 Cf.
PVSVṬ 393,19: tatprabhavaṃ rāgādisamutthāpitaṃ kāyavākkarma cādhar‐
mam abhyupagamya.
55 This abhyupagama is also presupposed in PV 4.107; see Tillemans 2000: 150–152.
56 For Śākyabuddhiʼs explanation, see above, pp. 69–70 n. 142.
57 This āgama is attributed by Kamalaśīla to Kapila, the founder of the Sāṅkhya, and
others; see McClintock 2010: 319–320.

114 HELMUT KRASSER
purified by the triple test, then any statement of that scripture regard‐
ing transcendent matters has to be non‐belying (avisaṃvāda), because
it is a statement of that scripture. The logical reason is a svabhāvahetu,
as in the inferences of Śākyabuddhi referred to above (p. 99), at least in
the various passages discussed by Tillemans.
Now we can consider the first question posed above (see p. 107).
Was Dharmakīrti “troubled by his own appeal to the Buddhaʼs trans‐
empirical knowledge”? As we have seen so far, Dharmakīrti has a clear
concept of āgama and of Dignāgaʼs de inition: āgama is not a pramāṇa
and there can be no certainty from its statements regarding transcen‐
dent matters; Dignāgaʼs de inition is faulty, constituting a śeṣavadanu‐
māna. Thus, I think Dharmakīrti was not troubled by his own appeal to
the Buddhaʼs transcendent knowledge, but he was troubled by Ku‐
mārila. For Dignāgaʼs de inition came under ire in the Ślokavārttika,
where Kumārila in the broader context of his refutation of omniscience
has a section (ŚV codanā 121–132), which is titled “Denial of ekadeśa‐
saṃvādānumāna” by Kataoka (2011a: 348). Already the introductory
one‐and‐a‐half verses make it clear that the argument at stake is the
same as the one propounded by Dignāga in his definition of āgama in
PS 2.5ab.
Even a man who (yo ʼpi), after having seen [i.e., ascertained] (dṛṣṭvā) a
person to be reliable (satyavāditām) with regard to [perceptible] objects
that have connections with the sense‐faculties and so on (indriyādisam‐
bandhaviṣaye), supposes (kalpayet) [that a statement of that person
must] also (api) [be true] with regard to an [imperceptible and religious]
matter that one can only believe (śraddheye ʼrthe), because it is a state‐
ment of the same person (tadvacanatvena), he too (tenāpi) would have
proved (sādhitā syāt) validity (pramāṇatā) through dependence [on so‐
mething external] (pāratantryeṇa).58 (Kataoka 2011a: 348–354)
Although Kumārila refers to objects that are connected with the sense‐
faculties (indriyādisambandhaviṣaye), I think there is little risk in in‐
terpreting this as a reference to the objects of perception and empirical
inference. This is at least the way Kumārilaʼs commentators under‐
58 ŚV codanā 121–122ab: yo ʼpīndriyārthasambandhaviṣaye satyavāditām / dṛṣṭvā tad‐
vacanatvena śraddheye ʼrthe ʼpi kalpayet // tenāpi pāratantryeṇa bādhitā syāt pramā‐
ṇatā /.
stand it, alluding to the Buddhaʼs teaching of momentariness (kṣaṇi‐

katva) as the basis for establishing that his teaching is (generally) true:
Umbeka – pramāṇāntaragocarārthapratipādake hi kṣaṇikādivākye; Su‐
caritamiśra – indriyādisambandhaviṣaye hi jñānamātrakṣaṇikatvādau;
Pārthasārathimiśra – asmadādipramāṇagocarārthaṃ kṣaṇikaṃ sarva‐
saṃskṛtam ityādi buddhavākyam (see Kataoka 2011a: 348–349 n. 374).
This is exactly the example that is used by Śākyabuddhi and Karṇaka‐
gomin when explaining PS 2.5ab as presented in PVSV 108,1 (dper na
ʼdus byas thams cad ni skad cig ma źes bya ba la sogs pa lta buʼo //;
yathā kṣaṇikāḥ sarve saṃskārā ityādikaḥ; see above, n. 35). In the
continuation of his argument Kumārilaʼs tone becomes quite sarcastic
as he shows what else one could prove by such an argument.
This statement of mine (vaco mama) “the Buddha, etc. (buddhādīnām) are
not omniscient” (asarvajñam iti) is true, because it is stated by me (mad‐
uktatvāt), just like (yathaiva) [my statements] “fire is hot” (agnir uṣṇaḥ)
and (api) “[fire is] bright” (bhāsvara iti).59, 60
And (ca) it is a perceived fact (pratyakṣam) that I uttered (maduktatvam)

[this sentence], whereas you have to prove (tvayā sādhyā) that he [i.e., the
Buddha and so on] stated (taduktatā) [these teachings]. Therefore (tena)
mine (madīyaḥ) should be (syāt) a [correct] reason (hetuḥ), whereas
yours (tava) is not established because it is doubtful (saṃdigdhāsiddha‐
tā).61 (Kataoka 2011a: 364)
I think that being faced with such mocking criticism and knowing that
the back‐bone of his own apoha theory is completely contrary to what
Dignāga proposed with his āgama definition, it fits the context better if
we assume Dharmakīrti to be troubled by Kumārila rather than by his
own appeal to the Buddhaʼs transempirical knowledge. Dharmakīrti
could not but bite the bullet. And he did it quite skilfully.
59 On the possibility of a Mīmāṃsaka using such a kind of śeṣavadanumāna to derive
the validity of the Veda in all its parts from a statement such as “Fire is the remedy
for cold,” see above pp. 102–103 with n. 41.
60 ŚV codanā 130: buddhādīnām asarvajñam iti satyaṃ vaco mama / maduktatvād ya‐
thaivāgnir uṣṇo bhāsvara ity api //.

61 ŚV codanā 131: pratyakṣaṃ ca maduktatvaṃ tvayā sādhyā taduktatā / tena hetur ma‐
dīyaḥ syāt sandigdhāsiddhatā tava //.

116 HELMUT KRASSER
5. SUMMARY
Putting now together these various bits and pieces of information scat‐
tered throughout the PV(SV), we can sketch the following picture.
Being confronted with the fact that Dignāgaʼs de inition of āgama as
well as the cognition that arises from it as an anumāna is not only
faulty but also incompatible with his apoha theory, the only possibility
left for Dharmakīrti was to resort to damage control. But he tried not
only to minimize the damage as much as possible but also to benefit
from it.
First of all, this definition of Dignāga is not to be understood on a
logical level.62 It was presented by Dignāga only because humans like
us cannot live without religion. This is due to the fact that credible per‐
sons report disastrous consequences of bad karma as well as the possi‐
bility of final release. Since we do not see anything that contradicts
their reports, we are better off following their advice when we act. It is
only under these circumstances that Dignāga gave this account;63 being
himself a specialist in logic (nyāyavid, PV 1.331), he knew very well
that such a definition, logically speaking, is as deficient as the one pro‐
posed by the Naiyāyikas.64 Thus, Dignāga set the standard for scripture
as high as possible on a rational level – the entire body of āgama has to
62 PV 1.216 (anumānatā ... agatyābhihitā), see above, p. 88; PVSV 109,19–20 (agatyā ...
anumānatvam ... upavarṇitam), see above, p. 101; and PVSV 168,1–2 (agatyā cedam
āgamalakṣaṇam iṣṭam), see above, p. 104 n. 42. See also §4.1.
63 PVSV 108,2–6; see above, p. 85 n. 11.
64 PVSV 173,22–25 (on PV 1.331); see above, p. 103 n. 41. See also Tillemans 1999a:
401: “Karṇakagomin and Śākyabuddhi ad v. 216 are clear on the implications: scrip‐
tural inference is an inference because of the thought of people (pumso ʼbhiprāyava‐
śāt) who want to engage themselves (pravṛttikāma) on the spiritual path: it is not
an inference objectively (vastutas).” Tillemans cites, in n. 11, Karṇakagomin (PVSVṬ
394,20–22): kiṃ tarhīṣṭasya pratyakṣānumānāgamyasyārthasyānantaroktena nyā‐
yenāvisaṃvādād anumānam api pravṛttikāmasya puṃso ʼbhiprāyavaśāt / vastutas tv
ananumānam, śabdānām arthaiḥ saha sambandhābhāvāt /. Already in their com‐
ments on nāyam (PVSV 108,2), etc., Śākyabuddhi and Karṇakagomin had explained
that Dignāga did not call āgama an inference in terms of real validity (bhāvikaṃ prā‐
māṇyam). PVṬ D242b5–6/P286a2–3: ʼdi skad du slob dpon gyis dṅos su tshad ma ñid
du brjod pas luṅ rjes su dpag pa ñid du gsuṅs pa ni ma yin gyi / ʼon kyaṅ skyes buʼi ʼjug
pa la ltos nas bstan pa yin no // = PVSVṬ 390,21–22: nācāryeṇa bhāvikaṃ prāmā‐
ṇyaṃ kathayatānumānatvam āgamasyoktam, api tu puruṣapravṛttim apekṣya.
pass the triple test.65 If it does, the probability that one will be success‐
ful in attaining one’s religious goals is much higher.66 But this triple
test is not only helpful in religious matters, it also is an effective tool
when dealing with the scriptures of the opposing schools. It is exactly
this triple test that is consistently applied by Dharmakīrti throughout
his works in order to refute the scriptures of his opponents; namely,
they have to pass the test (1) in terms of perceptible objects, (2) in
terms of objects that are inferable empirically, and then, finally, (3) in
terms of internal contradictions which can be checked with the help of
scripturally based inferences. And because āgama is not an independ‐
ent pramāṇa, it cannot be used as long as it has not passed the first two
steps of the triple test. Thus, many of the central doctrines of the scrip‐
tures of opposing schools, such as a permanent soul (ātman), primor‐
dial matter (pradhāna/prakṛti), or a creator god (īśvara), which in fact
can only be proven by means of scripture (if at all), fail to pass the se‐
cond step of the test, i.e., ātman and so on, which are taught to be
inferable, in fact cannot be established by empirical inference.67
The price for escaping Kumārilaʼs critique, for keeping his theory
that there is no relation between words and external objects as stated
in PV(SV) 1.213 – one of the cornerstones in Dharmakīrtiʼs refutation
of the validity of the Veda – and for dragging down all scriptures to the
realm where they can be tested by perception or empirical inference,
was very high. Dharmakīrti had to abandon the validity of all scrip‐
tures, including that of the Buddhists. This is quite consistent with the
rest of his logical system which builds on pervasion (vyāpti). There is
no way whatsoever to establish a pervasion between the act of speak‐
ing of a credible person and his telling the truth.68 In accordance with
this theory Dharmakīrti established a different method to prove the
authoritativeness of the Buddha in the Pramāṇasiddhi chapter, basing
himself on empirical inference and thus avoiding a śeṣavadanumāna.69
65 PV 1.214–215 with PVSV; see above §2.2.
66 PVSV 174,1: tatra kadācid avisaṃvādasambhavāt; see above, p. 103 n. 41.
67 See PV 4.2 and 4.48–52 (especially 4.48 and 50) in Tillemans 2000: 11, 78–83; see
also above, p. 112 n. 53.

68 PVSV 167,26–168,1: tatpravṛtter avisaṃvādena vyāptyasiddheś ca; see above, p. 104
n. 42.
69 See above, pp. 105–106 and nn. 43 and 44.
118 HELMUT KRASSER
While Dharmakīrtiʼs courageous reform of Dignāgaʼs theory of

scripture, which he was forced to undertake not only by his own sys‐
tem but also by Kumārila, turned out to be quite successful within the
framework of his own system, it could very well have pleased Kumārila
and other non‐Buddhists to see their worst critic giving up the validity
of the buddhavacana. It was also very much welcomed by his Buddhist
adversary Bhāviveka, who immediately sent his greetings:
atrocyate pramāṇaṃ naḥ sarvaṃ tāthāgataṃ vacaḥ /
āptopadeśaprāmāṇyād bhadro hi pratipadyate //
nāgamāntarasandigdhaviparyastamatiḥ paraḥ /
tasmāt tatpratipattyarthaṃ tanmṛgyo yuktimannayaḥ // MHK 5.8–9
To this [pūrvapakṣa of the Yogācāra] we reply: All the words of the Tathā‐
gata are authoritative (pramāṇa) for us, because the teachings of a relia‐
ble person are authoritative. A good one puts [these] into practice. The
other one, whose mind is in doubt and confused by other scriptures, does
not. Therefore the path of reasoning (yuktimannaya) should be followed
by him in order to put these into practice.70
And it might well be the case that Bhāviveka was not the only one to
accuse Dharmakīrti of not being a good (bhadra) Buddhist by being
taken in by Kumārila’s arguments against scripture, the arguments of
the opponent (āgamāntara), and, in his confusion, thinking them to be
his own arguments. There might have been other Buddhists as well
who thought he had thrown the baby out with the bathwater.

70 For the interpretation of these two verses and the commentary on the latter, see
Section 1.3.1.1, “Who is the opponent in MHK 5.8–9?” in Krasser 2012: 545–546. For
the text and different translations, see Hoornaert 2000: 78 and 90, and Eckel 2008:
225–227 and 394–395.

JOHN TABER
Dharmakīrti and the Mīmāṃsakas in Conflict

1. INTRODUCTION: DHARMAKĪRTI’S RELATION TO MĪMĀṂSĀ

It is well known that Dharmakīrti in his writings is often preoccupied
with Mīmāṃsā theories and arguments. Frequently the Mīmāṃsaka is
an adversary and target of criticism, but occasionally he may have been
a positive stimulus for Dharmakīrti’s own thinking. For instance, Dhar‐
makīrti’s criticism of the śeṣavadanumāna, “inference with a remain‐
der,”1 as a fallacy at the beginning of Pramāṇavārttika 1, seems primar‐
ily intended to dispell a stock Mīmāṃsā argument2 against the Bud‐
dha’s omniscience, namely, the Buddha could not have been without
desire (hence, he could not have been omniscient; for no omniscient
person is possessed of desire), because he spoke.3 Yet at the beginning
 A version of this essay was presented as a talk at Ryukoku University on January 14,
2011. I would like to thank Prof. Shoryu Katsura for inviting me and the distingui‐
shed scholars present who contributed to a lively discussion and who made some
very helpful comments, including Prof. Noritoshi Aramaki, Prof. Kei Kataoka, Prof.
Yusho Wakahara, and Prof. Kiyotaka Yoshimizu.
1 Which includes inferences from cause to effect as well as inferences from effect to
cause and other inferences based merely on the non‐observation of the hetu in the
vipakṣa. On the śeṣavadanumāna, see PV 1.331/PVSV 173,22–26 above, pp. 62–63
nn. 126–127.
2 It is implicit in Bhāviveka’s presentation of Mīmāṃsā in MHK 9. MHK 9.3ab: rāgādi‐
doṣaduṣṭatvāt puruṣasya vaco mṛṣā. Cf. MHK 9.15 and 16.

3 See PV(SV) 1.12 and 14. I am inclined to resist the suggestion by Kataoka (2003a:
60–62 n. 35) that Dharmakīrti is specifically attacking Kumārila here. In ŚV codanā

137, Kumārila only points out that, being without activity, because he is without
desire, the teachings of an omniscient person would have to have been composed by
others (deśanā anyapraṇītaiva). He does not, to my knowledge, appeal to merely
speaking, which is the reason given by Dharmakīrti PV 1.12c for erroneously infer‐
ring that one is possessed of desire (vacanād rāgitādivat), as a reason for denying
an omniscient being. I think this is a significant difference. At ŚV codanā 132,
moreover, and in his Bṛhaṭṭīkā (TSK 3157/TSŚ 3156) Kumārila mentions several
120 JOHN TABER
of Pramāṇavārttika 2, vv. 10–16, Dharmakīrti points out several falla‐

cies in arguments allegedly proving an eternal creator of the universe
that more or less match fallacies identified by Kumārila in his refuta‐
tions of arguments for the existence of God in his Ślokavārttika – as if
Dharmakīrti actually borrowed from Kumārila.4 And perhaps most
strikingly, at PV 2.5c Dharmakīrti introduces an alternative definition
of pramāṇa, ajñātārthaprakāśo vā, which is strongly reminiscent of the
Mīmāṃsā definition of pramāṇa that specifies that it must, among
other things, present that which has not been previously grasped (that
is, it must be anadhigatārthaviṣaya or an agṛhītagrāhi‐ or apūrvārtha‐
jñāna, etc.), which requirement may also have been originally intro‐
duced by Kumārila.5
It is at the end of Pramāṇavārttika 1, however, beginning with verse
224 and extending to the end of the text – about a third of the treatise –
where Dharmakīrti launches an all‐out attack on Mīmāṃsā, focusing on
the central claim of the Mīmāṃsakas that the Veda is eternal and
reasons – prameyatva (ŚV codanā 132a), jñeyatva, vastutva, sattva (TSK

3157a2b/TSŚ 3156a2b) – but “speaking” or “being a speaker” (vaktṛtva; cf. SS 23,11–
14) is not among them. Kataoka (forthcoming, n. 2) gives a whole list of passages
from Dharmakīrti’s works that, he maintains, “[seem] to presuppose (or criticize)
ŚV or TV.” Some of the passages Kataoka has discussed in published articles. As
always, the devil is in the details and their interpretation. I would add to Kataoka’s
list the passage that begins PV 3.25ff., where Dharmakīrti refutes the reality of
universals. Some of the ideas about universals he attacks are strikingly similar to
ones that find expression in Kumārila’s discussions.
4 See Krasser 1999.
5 Krasser 2001 traces this definition back to a verse quoted by Ratnakīrti that pre‐
sumably comes from Kumārila’s Bṛhaṭṭīkā: tatrāpūrvārthavijñānaṃ niścitaṃ bādha‐

varjitam / aduṣṭakāraṇārabdhaṃ pramāṇaṃ lokasammatam //. It should be noted,
however, that the verse mentions four criteria that must be met for a cognition to be
a pramāṇa, namely, (1) it presents a new object, (2) it is definite, (3) it is free of
sublation, and (4) it is produced by non‐defective causes, and that only the first of
these is alluded to by PV 2.5a. Moreover, it is unlikely, pace Krasser 2001: 195, that
Dharmakīrti would have proposed this alternative definition “in order to prove to
the Mīmāṃsakas that, even according to their own definition when understood
properly, not the Veda, but the Buddha is to be regarded as a pramāṇa,” since he
would have been well aware that the Mīmāṃsakas considered the teachings of the
Buddha to have been contradicted by other pramāṇas in many respects. (Thus, at
the very least, the cognition of the Buddha was not bādhavarjita!) And so Dharma‐
kīrti’s motive for introducing this alternative definition remains rather mysterious.
DHARMAKĪRTI AND THE MĪMĀṂSAKAS IN CONFLICT 121
authorless. The passage that is translated in this volume, PV(SV)

1.312–340, represents the culmination of that critique. In this essay I
would like to examine certain aspects of Dharmakīrti’s treatment of
Mīmāṃsā in this passage.
My main concern will be, How well do Dharmakīrti’s criticisms in
our text actually tell against the Mīmāṃsā position of the authorless‐
ness of scripture? This may strike some readers as a rather unusual
question to ask. Usually, as historians of Indian philosophy we restrict
ourselves to reconstructing the ideas of Indian philosophers and un‐
derstanding them in their historical context. We do not normally con‐
cern ourselves with their validity or cogency. I do believe, however,
that this question is of relevance in assessing the depth and sophistica‐
tion of Dharmakīrti’s knowledge of Mīmāṃsā. Buddhist legends about
Dharmakīrti tell us that he was born into a Brahmin family.6 If that is
true, then he would have received a Brahmin’s education, which very
well could have included instruction in, or at least exposure to, Mī‐
māṃsā. How extensive was his training in Mīmāṃsā? Was he really
steeped in it – so that perhaps Mīmāṃsā had a more profound influence
on his thought than we realize? His preoccupation with mantras, in
particular, gives this impression. Or did he have merely a superficial
acquaintance with it, just enough to arouse a strong aversion in him
toward it and provide him with enough information to be able to
devise clever objections against its doctrines?7 (As they say, a little
knowledge is a dangerous thing.) Or was his expertise in Mīmāṃsā
somewhere in between?
6 Thus, Tāranātha (GCh 229): “Having a very sharp intellect, he [Dharmakīrti] tho‐
roughly studied from his early childhood the fine arts, the Vedas with all their
aṅgas, medicine, grammar and all the tīrthika philosophies.”
7 In my experience in the United States, at least, the most outspoken atheists are
those who attended Catholic schools when they were young.

122 JOHN TABER
2. DHARMAKĪRTI’S TREATMENT OF MĪMĀṂSĀ
2.1. Overview of Dharmakīrti’s critique of Mīmāṃsā, PV(SV)

1.312–340
For the Mīmāṃsaka the validity (prāmāṇya) of the Veda is grounded
on its authorlessness (apauruṣeyatva). Dharmakīrti has the Mīmāṃsa‐
ka declare at PV 1.224 that the falsehood of statements can derive only
from the defects of the humans who utter them (i.e., principally, their
ignorance or dishonesty); an authorless statement therefore cannot be
false. Up to the section of the text we have translated Dharmakīrti has
already challenged this doctrine on several fronts.8 The authorlessness
of the Veda would require that there be an eternal connection between
word and meaning; but the connection between word and meaning is
established by convention; indeed, in general, any relation between
independent entities is not real but only mentally constructed.9 It
would also require that words and sentences are themselves uncreated
and permanent; but we know them to be transient, like everything else.
(Here, Dharmakīrti presents his vināśitvānumāna.)10 We know, more‐
over, that humans are capable of devising mantras that are causally
efficacious; thus the fact that the Veda contains mantras does not speak
against its having a human author, either.11 Nor does the fact that there
is no memory of an author of the Veda; for a phenomenon of a certain
9 PV(SV) 1.226–238. See Eltschinger 2007a: 138–142.
10 PV(SV) 1.269–283ab. Extensive arguments specifically against the eternality of

words and the sentence (whether conceived of as a sphoṭa or a succession of phone‐
mes, varṇānupūrvī) are found at PV(SV) 1.247–268. For detailed summaries see
Eltschinger 2007a: Chapters 5 and 6. Much of what Dharmakīrti says in his critique
of the Mīmāṃsā doctrine of the eternality of language, starting with PV(SV) 1.225,
seems to presuppose ideas of Kumārila. Yet there are at least two very important
views of Kumārila that have an obvious bearing on his discussion that, as far as I am
able to discern, he ignores: (1) the intrinsic validity (svataḥprāmāṇya) of all cogni‐
tions and (2) the impossibility of fixing the relation between word and meaning by
convention (expounded in ŚV sambandhākṣepaparihāra). It is these sorts of lapses
(there are others) that still make me reluctant to accept the widely held thesis that
when Dharmakīrti is attacking Mīmāṃsā he is specifically attacking Kumārila, des‐
pite all the good work done by other scholars to point out evidence for it.
11 PVSV 123,14–124,23 and PV(SV) 1.292–311. See Eltschinger 2001 and 2008.
type – in this case, a linguistic corpus – will always have the same cause
as other things of that type, whether or not it is immediately known,
just like fire and fuel.12
Now, in this final section, beginning with v. 312, Dharmakīrti pre‐
sents his final reductio ad absurdum of the Mīmāṃsā position – as if
saving his most devastating criticism for last. Even if the Veda were
authorless, he points out, we could never know what it means! For the
Veda itself does not tell us what it means; humans must surmise it. And
there is, according to the Mīmāṃsakas’ own assumptions, no human
qualified to interpret the Veda; for they believe that humans are
incapable of knowing the supersensible things of which the Veda
speaks.13
artho ’yam nāyam artho na iti śabdā vadanti na /
kalpyo ’yam arthaḥ puruṣais te ca rāgādisaṃyutāḥ // PV 1.312
[Vedic] words do not [themselves] declare, “This is our meaning, not

this.” The meaning [which Vedic words have] must be postulated by
humans. The latter are possessed, however, of [moral defects] like desire.
Moreover, it is doubtful that there is any unbroken, uncorrupted tradi‐

tion of Vedic interpretation. Even if there were an “authorless” tradi‐
tion of interpretation, we could not be confident we knew what it
meant any more than we are in knowing what the Veda means, since it
would refer to the same supersensible matters. Finally, mundane usage
(lokavāda, prasiddhi) cannot be resorted to as a criterion for determin‐
ing the meaning of Vedic statements, least of all by the Mīmāṃsakas,
who like to point out that what ordinary humans say is for the most
part untrue – hence, surely, how they commonly employ words cannot
serve as any kind of pramāṇa – and who also routinely deviate from
common usage themselves in their Vedic interpretations.
Dharmakīrti thus presents the spectacle of a community of people
diligently following the prescriptions and prohibitions of texts the
meaning of which, according to their own teachings, they could not
12 PV(SV) 1.242; see the entire section PV(SV) 1.239–247.
13 MīSū 1.1.4. See ŚV pratyakṣasūtra, esp. 17–37; Taber 2005: 51–58.

124 JOHN TABER
possibly understand. No wonder that the final statement of his critique

is an expression of utter contempt for the pious Brahmin!
vedaprāmāṇyaṃ kasyacit kartṛvādaḥ
snāne dharmecchā jātivādāvalepaḥ /
santāpārambhaḥ pāpahānāya ceti
dhvastaprajñāne pañca liṅgāni jāḍye // PV 1.340
[Believing in the] authority of the Veda, claiming something [permanent]

to be the agent [of actions], seeking merit in ablutions, taking pride in
one’s caste14 and undertaking penance to remove sin – these are the five
signs of complete stupidity devoid of any discrimination.
2.2. Dharmakīrti’s discussion, PV(SV) 1.319–320ab

Let us look a little more closely at some of the things Dharmakīrti says
about “ordinary parlance” (lokavāda) or common usage (prasiddhi) as
a criterion for determining the meaning of Vedic sentences and the
possibility of an eternal, authorless tradition of exegesis. He takes up
both of these matters together in vv. 319–320ab and his extensive Sva‐
vṛtti thereon. The account that follows is essentially identical to our
synopsis of this section of the text. Because of its significance for
assessing Dharmakīrti’s knowledge of Mīmāṃsā, in particular, it seems
worthwhile to highlight it here.
In raising the question of the criterion of common usage Dharma‐
kīrti may have had in mind the principle stated at Mīmāṃsāsūtra 1.3.30
that the words of the Veda and ordinary language have the same mean‐
ings. In any case, Mīmāṃsakas routinely cite prasiddhi, common usage,
as a reason for interpreting Vedic passages in certain ways; at other
times, however, they justify deviating from common usage.
Is it possible, then, that we needn’t rely on anyone with a special gift
for knowing the supersensible to interpret for us what the Veda means;
rather, we can just construe Vedic sentences ourselves according to the
ordinary, everyday meanings of their words? Dharmakīrti immediately
points out that, even if we construe Vedic words as ordinary ones, they
will still in many instances be polysemic, as indeed ordinary words
14 Lit., “pride in declaring/speaking about one’s caste.”

often are. Who will be able to determine which of the many possible
meanings of a word in the Veda is the right one? (v. 319) The ordinary
meaning of everyday discourse, moreover, is established by con‐
vention, which is accessible to instruction, while the Veda is suppos‐
edly eternal. Who could, in the case of words which supposedly have a
beginningless, authorless relation with supersensible meanings, be
able to tell us what they mean?
If, on the other hand, one held that there is an explanation of the
meaning of the Vedic word that is beginningless and authorless as well
– that is to say, an exegetical tradition that extends forever back into
the past – then the problem is just moved back a step. How is the
meaning of this beginningless explanation known? Moreover, how do
we know the explanation has never been corrupted? We are aware of
various factors that introduce errors into traditions – enmity, pride,
and so forth. And why would the Mīmāṃsaka, of all people, put confi‐
dence in a supposedly unbroken tradition of explanation, since he is
the one who emphasizes that humans afflicted with moral faults are
not to be trusted?15 That, in fact, is his most characteristic point – “the
color of his own face” – says Dharmakīrti.
In the continuation of his Svavṛtti to v. 319 Dharmakīrti goes further
into how we hear of Vedic schools recovering after nearly dying out –
even today some have only a few reciters – so that one might suspect
that even those schools that have many adherents today could have
been nearly extinct at one time but were restored, and that in the proc‐
ess of restoration errors could have crept into the recitation of the
Veda in various ways. And the same could be the case for any “begin‐
ningless” tradition of Vedic interpretation.
In summary, it would seem that one cannot establish the meaning of
the Veda either through a beginningless, authorless explanation or
ordinary linguistic practice. Returning to the latter, Dharmakīrti points
out that even if the relation of word and meaning weren’t conventional
but eternal, ordinary parlance still shows us that words in general are
15 Thus, Kumārila’s famous statement (ŚV codanā 144ab), “At all times, humans for the
most part speak what is untrue” (sarvadā cāpi puruṣāḥ prāyeṇānṛtavādinaḥ /). I am
not convinced that anṛtavādin necessarily means here, or in the Vedic passage cited
by Śabara (ŚBh II.4,4), anṛtavādinī vāk, intentionally speaking falsehood, hence
lying.
126 JOHN TABER
polysemic, so that there would always be doubt about the meanings of

Vedic words; indeed, there are lots of Vedic words whose meanings are
unknown or known but used in unusual ways. But mainly what ordi‐
nary parlance teaches us is that a word can mean anything; it is actu‐
ally only by convention that it is assigned to a specific meaning. So
again in the case of Vedic words, it seems that a person capable of
knowing supersensible things would be required to tell us the super‐
sensible meanings to which they are assigned.
Yet even the Mīmāṃsaka, continues Dharmakīrti in PV(SV) 1.220ab,
does not always follow common usage when it comes to explaining the
meaning of Vedic words. He offers as examples the words svarga and
urvaśī. svarga commonly means “heaven,” but the Mīmāṃsaka con‐
strues it as “delight.”16 urvaśī, meanwhile, is usually the name of a
nymph who resides in heaven, but typically the Mīmāṃsakas do not
interpret proper names in the Veda as referring to individuals – which
would impugn its eternality if they were themselves “non‐eternal” –
but offer etymological analyses instead.17 Moreover, ritualists referred
to the two kindling sticks that are rubbed together to start the fire in
the Soma sacrifice as Urvaśī (the lower) and Purūravas (the upper
one), who according to a widespread myth was her consort.18 (Learned
Buddhists who knew the myth and the ritual must have thought this
was a fine joke!) The Mīmāṃsaka, moreover, cannot claim that one
must sometimes resort to an uncommon meaning because the common
one is blocked, i.e., it does not fit the context, for how could one ever
know in the case of a Vedic statement, which refers to a supersensible
state of affairs, that the common meaning is blocked? And if we accept
uncommon meanings in the case of such words as svarga and urvaśī,
how do we know we shouldn’t accept one for agnihotraṃ juhuyāt
svargakāmaḥ, say, “One should eat dog meat”? The meaning of this
sentence cannot be resolved by other Vedic passages, because the
meaning of those other passages are in doubt as well – for all we know,
the occurrence of agnihotraṃ juhuyāt svargakāmaḥ there could also
mean “One should eat dog meat”! Thus Dharmakīrti repeats the slur,
16 See ŚBh V.72,6–7 ad MīSū 4.3.15.
17 See ŚBh I.121,7–10 ad MīSū 1.1.28 (pūrvapakṣa) and I.123,7–124,5 ad 1.1.31 (sid‐
dhānta). For etymological derivations of urvaśī see Nir 5.14.

18 See TaitS 1.3.7.1 and 6.3.5.2–3; for versions of the myth see ṚV 10.95 and ŚB 11.5.1.
made initially in v. 318, for which this part of Pramāṇavārttika 1 is

famous.
The criticisms that Dharmakīrti directs against Mīmāṃsā interpre‐
tive practices in our text, and especially in PV(SV) 1.319–320ab, reflect
a knowledge of Mīmāṃsā that goes considerably beyond just the
awareness that the Mīmāṃsakas held the Veda to be authorless, that
even the relation between word and meaning is authorless, and that
humans are incapable of cognizing the supersensible, because they are
corrupted by desire, etc. They refer to specific exegetical principles,
such as that the meaning of the words of the Veda and those of ordi‐
nary parlance are (for the most part) the same, that nevertheless in
certain circumstances the common meaning must be abandoned, and
that in many instances other passages of the Veda itself clarify what a
particularly problematic passage means. And, also quite significantly,
they allude to the Mīmāṃsā belief that recensions of the Veda are liable
to extinction or at least periodic decline. Kumārila appeals in his Tan‐
travārttika to lost Vedic texts as the basis of the authority of the dhar‐
maśāstras.19 Do these references indicate that Dharmakīrti had a really
in‐depth knowledge of Mīmāṃsā, indeed, that perhaps at one point in
time he had even been trained in Mīmāṃsā?
A way to approach this matter, I think, is to ask, What would a Mī‐
māṃsaka make of these kinds of criticisms? Would they have seemed
adequate, or even be seen as posing a serious challenge, to someone
thoroughly trained in Mīmāṃsā? Do they really get at the heart of Mī‐
māṃsā exegetical practice? Could they have been posed by someone
who really knew what Mīmāṃsā was about? Or are they the objections
of someone who was brilliant and perhaps much more knowledgeable
of Mīmāṃsā than most, but who was fundamentally an outsider? That
is what I would like to consider in the rest of my essay. I believe the
short answer to the question, What would a Mīmāṃsaka make of these
criticisms?, is something like the following. Contrary to what Dharma‐
kīrti claims, the Veda itself tells us what it means. Mīmāṃsā, drawing on
an ancient tradition of Vedic exegesis, provides us with a system of
rules and principles, as clear and objective as the rules of logic, for
interpreting it. One need only learn how to apply the proper methodo‐
19 See Kataoka, forthcoming.

128 JOHN TABER
logy – which it is in fact one of the main tasks of Mīmāṃsā to elucidate

– and the meaning of the Veda will unfold. But the full answer to this
question is quite complicated, and I can only begin to give a hint of it
here.
3. THE MĪMĀṂSĀ SCIENCE OF EXEGESIS
3.1. The interpretation of arthavādas in Mīmāṃsā

Pādas 2–4 of the first adhyāya of the Mīmāṃsāsūtra and its commen‐
taries (to which I shall refer, for the sake of convenience, as the Mī‐
māṃsādarśana) are specifically concerned with the meaning of the
Veda – with whether it has any meaning, whether its meaning is cohe‐
rent, and how to ascertain its meaning. The question of the meaning of
Vedic passages is of course a theme that recurs throughout the Mīmāṃ‐
sādarśana, but it constitutes the main preoccupation of the adhikara‐
ṇas that come immediately after the general discussion of the means of
knowing Dharma in the tarkapāda. In the first adhikaraṇa of the se‐
cond pāda (sūtras 1–18) the question is raised whether those passages
of the Veda that do not directly relate to action are without artha.20
Although “without artha” (ānarthakya) here means without purpose, it
is closely related to being without meaning. Certain sentences of the
Veda appear not to have a purpose because they cannot be construed
as enjoining actions or accessories of actions. Thus, we encounter sta‐
tements such as “He [Agni, frightened by the Asuras] cried (arodīt).
That he [viz., Rudra] cried constitutes the rudra‐ness of Rudra” (TaitS
1.5.1). This occurs in a Brāhmaṇa passage that discusses the punar‐
ādheya, the rite for rekindling the sacrificial fire. Later in the same
Saṃhitā, in a discussion of the kāmyeṣṭi for offspring and cattle, we
meet with the sentence “Prajāpati cut out his own fat” (TaitS 2.1.1).
Now crying, the pūrvapakṣin of Śabara’s commentary points out, is not
something one can do at will; and having cut out one’s own fat, one
could not (presumably, completely incapacitated!) proceed to perform
the sacrifice, as specified, “with a hornless goat.”21 Construed as injunc‐
20 MīSū 1.2.1: āmnāyasya kriyārthatvād ānarthakyam atadarthānāṃ tasmād anityam
ucyate.
21 ŚBh II.3,1–3.
tions, such passages do not make sense. Other passages, meanwhile,

are contradicted by scripture itself, or even by perception. Thus, “The
mind is a thief, speech is a liar” (MaitS 4.5.2)22 which Śabara associates
with a discussion of the dakṣinā given to the priests;23 and “During the
day it is only the smoke of the fire that is seen, not its flame, and during
the night only the flame of the fire is seen, not the smoke” (parallel:
TaitB 2.1.2.10),24 which Śabara connects with the injunction “In the
evening one offers [by saying] agnir jyotir jyotir agniḥ, in the morning
[by saying] sūryo jyotir jyotiḥ sūryaḥ” (TaitB 2.1.9.2).25 The first two
sentences cannot be injunctions because they are contradicted else‐
where by prohibitions against stealing and lying;26 the last is contra‐
dicted by ordinary experience. These kinds of sentences, the pūrva‐
pakṣin concludes, “do not effect a permanent purpose,”27 and so may be
excluded from consideration in interpreting the Veda. This point may
have been raised within Mīmāṃsā or ritualist circles, that is to say, not
necessarily by Buddhists or other heterodox skeptics, but by experts in
the sacrifice who simply thought that the meaning of certain portions
of the Veda may be ignored, or that they may not even be intended as
having meaning at all.28
The solution to the apparent purposelessness of such sentences, as
developed in the siddhānta of the first adhikaraṇa (sūtras 7ff.), is to
interpret them properly as arthavādas which praise the actions of the
22 ŚBh II.4,4.
23 hiraṇyaṃ haste bhavaty atha gṛhṇāti, parallel to MaitS 4.8.3, which reads nayati
instead of gṛhṇāti (see Garge 1952: 108). ŚBh II.28,2–3.

24 ŚBh II.5,4–5.
25 ŚBh II.28,9–12.
26 See ŚBh IV.322,3–378,3 ad MīSū 3.4.12–13, where Śabara argues, remarkably, that
TaitS 2.5.5.6, nānṛtaṃ vadet, has as its scope, not human behavior in general, but
only the Darśapūrṇamāsa!
27 ŚBh II.3,5–6: tasmād evaṃjātīyakāni vākyāny anityāny ucyante. yady api ca nityāni
tathāpi na nityam arthaṃ kurvantīti.

28 In the continuation of the passage the pūrvapakṣin explains that the objection is not
intended to impugn the entire Veda: sa eṣa vākyaikadeśasyākṣepo na kṛtsnasya vā‐

kyasya (ŚBh II.3,7–8). Concerns about the meaningfulness of mantras are attributed
by Yāska to Kautsa in Nir 1.15, who argues that if the Veda is without meaning then
a science of etymology is unnecessary. Renou (1960: 68), meanwhile, notes that a
Prātiśākhya of the Atharva Veda school is ascribed to a certain Kautsa.
130 JOHN TABER
injunctions they are associated with.29 Also in TaitS 2.1.1, for example,
the sentence “Vāyu is the swiftest deity” comes immediately after the
injunction “One who desires wealth should sacrifice (ālabheta) a white
[victim] dedicated to Vāyu.” This sentence, then, supports the injunc‐
tion by eulogizing the god to whom the victim is sacrificed; it is an
arthavāda. When an arthavāda occurs together with an injunction, it
motivates the sacrificer to carry out the rite. Yet the connection be‐
tween vidhi and arthavāda is not always clear, nor is the manner in
which the arthavāda supports the vidhi. Often an arthavāda must be
interpreted figuratively. When in the same passage it is said, “Prajāpati
cut out his own fat,” it is not intended that this ever happened.30 From
the story told in this section of the Brāhmaṇa, however – of Prajāpati
creating livestock and offspring by cutting out his own fat and placing
it in the fire, which moreover can be given a perfectly reasonable sense
if one grasps the secondary meanings of its words31 – it becomes evi‐
dent that a praising of the efficacy of the act enjoined, in this instance,
the sacrificing of a hornless goat dedicated to Prajāpati by one desiring
offspring and cattle, is to be understood.32
Similarly, the mentioning of tears, which Śabara associates with the
injunction not to place silver on the barhis for the priest – one should,
rather, always give gold! – is to strengthen the prohibition; for, the pas‐
sage continues, “The tear that was shed became silver; therefore silver
is not a suitable gift, for it is born of tears” (TaitS 1.5.1.1–2). The artha‐
vāda does not really report that Rudra wept, but connects silver with
weeping and unhappiness. And so for the other false statements, e.g.,
“The mind is a thief, speech is a liar” – this, too, strengthens the injunc‐
29 MīSū 1.2.7: vidhinā tv ekavākyatvāt stutyarthena vidhīnāṃ syuḥ.
30 For, Śabara explains (ŚBh II.26,11–12), if the mentioning of an actual occurrence
(vṛttāntānvākhyāna) were asserted (vidhīyamāna) then the fault of the Veda’s

having a beginning would ensue.
31 Śabara (ŚBh II.27,1–4) offers such a reading: “Prajāpati would be some eternal thing
– wind, space, or the sun. ‘He cut out his own fat’ [would mean] rain, the wind, a ray
of light. ‘He placed it in the fire’ [would mean,] in lightning, the light [of the heavens]
[read: varcasi?], or the mundane [fire].”
32 ŚBh II.26,1–27,4.
tion to give the priests gold, according to Śabara and Kumārila.33 The
mind is a thief in a secondary sense, “due to its form,”34 because, as Śa‐
bara explains, “as thieves have a concealed form so does the mind.”35
Speech is figuratively a liar, “because of what is predominant,” i.e., for
the most part people say what is untrue.36
All of these sentences, then, can be given a coherent meaning that
associates them with action, first, by construing them together with the
appropriate injunctions in the same or other passages37 and, second, by
ascertaining their figurative meaning when they have one.38 Mīmāṃsā
is largely dedicated to showing how to make such determinations.39
3.2. The construal of mantras

The interpretation of arthavādas is just one of many areas in which
Mīmāṃsā developed a scientific methodology for solving problems of
interpretation. Another very important topic in Mīmāṃsā of course is
33 Kumārila (TV II.28,13–14) explains – rather implausibly – that mind and speech are
being deprecated because, though related to the action to be performed, they are of
much less significance than the gold: tadatyantāntaraṅgabhūtayor apy anayor dūre‐
ṇa hiraṇyād ūnatvaṃ steyānṛtavādayogād iti.
34 MīSū 1.2.11: rūpāt prāyāt.
35 ŚBh II.28,6–7.
36 See above, p. 125 n. 15.
37 That is, together with which they form “a single sentence” (ekavākyatā). See McCrea
2000: 436–7.
38 Some of the most notable cases where one is called upon to resort to figurative
meaning are those passages that attribute physical features to deities, as if they
were embodied. Mīmāṃsā in general rejects the popular notion of deities as super‐
natural beings. See MīSū 9.1.6–10 and 10.4.23. According to Garge 1952: 152, “It is
clear … that the Mīmāṃsakas do not regard the Deities as objects of worship nor be‐
lieve that they exist anywhere except in the words of the Vedic texts. They are only
eternal concepts.”
39 According to Garge 1952: 261, “arthavāda passages appear in the Vedic Texts in
numerous forms, giving rise to a number of complicated questions regarding their

syntactical interpretations. Jaimini composed as many as sixty‐four sūtras and
eleven adhikaraṇas in which he exhaustively explains all the principal types in
which the arthavāda texts appear in Vedic Texts.”
132 JOHN TABER
the construal of mantras, the principal question in their case being

whether they are meaningful at all. As Śabara poses it,
Now, is it the case that mantras express something intended or do not ex‐
press something intended? Do they assist the sacrifice by revealing a
meaning or by their mere utterance?40
The pūrvapakṣin notes various mantras that appear to be nonsensical

in different ways. They mention things that do not exist (ṚV 4.58.3 re‐
fers to a bull [vṛṣabha] with four horns, three feet, two heads, and
seven hands);41 they assign purposes to inanimate objects (“O plant,
protect this one!” TaitS 1.2.1.b);42 they are self‐contradictory (“Aditi is
the heaven, Aditi is the atmosphere” ṚV 1.89.10);43 or their meanings
are simply unknown (sṛṇy eva jarbharī turpharītū ṚV 10.106.6).44 Thus,
it would seem that their meanings are not intended, but just the utter‐
ing of the (nonsensical) mantra is what is important.
Śabara’s siddhāntin responds, however, that the meanings of man‐
tras are fully intended and that they are purposeful in that they “bring
to light the subsidiary parts of the sacrifice during the sacrifice,”45 that
is, as later treatises will explain, they serve as a kind of script for the
priest to follow in performing the ritual. 46 Thus, they are meaningful,
but in many instances their meaning is figurative. The mantra mention‐
40 ŚBh II.49,1–2. See Taber 1989.
41 MīSū 1.2.31a: avidyamānatvāt.
42 MīSū 1.2.31b: acetane ’rthe khalv arthaṃ nibandhanāt.
43 MīSū 1.2.31c: arthavipratiṣedhāt.
44 MīSū 1.2.31e: avijñeyāt. In the Jośī edition of Mīmāṃsādarśana 1.2–2.1 MīSū 1.2.31
is broken into nine parts, numbered 31–39, “for ease of explanation” (p. 48, note).
The beginning of the siddhānta, aviśiṣṭas tu vākyārthaḥ, however, restarts the num‐
bering with 32. Jha’s translation also breaks up sūtra 1.2.31 but begins the siddhā‐
nta with number 40! All of Śabara’s examples of unintelligible Ṛgvedic mantras in
his discussion of this reason appear to be taken from Yāska’s Nirukta. See Nir 13.5
for an explanation of ṚV 10.106.6; Nir 6.15 for an explanation of amyak (in ṚV
1.169.3: amyak sā); and Nir 5.11 for kāṇukā (in ṚV 8.77.4: indraḥ somasya kāṇukā).
See Garge 1952: 143–144.
45 ŚBh
II.57,12–15: yajñe yajñāṅgaprakāśanam eva prayojanam … na hy aprakāśite
yajñe yajñāṅge ca yāgaḥ śakyo ’bhinirvartayitum.
46 See AS 17: prayogasamavetārthasmārakā mantrāḥ. teṣāṃ ca tādṛśārthasmārakatve‐
naivārthavattvam. See Taber 1989: 149–50.

ing a bull with four horns, etc., Śabara explains, should be taken as an
arthavāda, construed figuratively as follows: the four horns are the
hotṛs; the three feet are the savanas; the two heads are the sacrificer
and his wife; the seven hands the meters; the bull is the sacrifice, and
so on. It is like describing a river by saying, “It has Cakravāka birds as
its breasts, a row of swans as its teeth, reeds as its clothes, and mosses
as its hair.”47 “Aditi is heaven, Aditi is the atmosphere,” similarly,
should not be taken literally. It is like saying, “You are my mother, you
are my father.”48
In fact, a meaning always exists for a mantra, but it may not be evi‐
dent to everyone.49 In difficult cases, Śabara explains, the meaning
must be determined “from the roots of words by means of illustrative
passages, etymology, and grammar.”50 Kumārila expands this list to
include the purpose (artha) of the mantra, i.e., its application to a par‐
ticular ritual action; its context (prakaraṇa), i.e., all the relevant pas‐
sages specifying the various factors of the sacrifice to which it belongs;
the hymn in which it occurs; the deity to whom it is addressed; and the
ṛṣi to whom it is attributed51 – Kumārila of course explains that this
means, not the ṛṣi who composed the mantra, but the ṛṣi who, according
to legend, employed it on an important occasion.52 And he goes on to
47 ŚBh II.64,6–12.
48 ŚBh II.65,7–8. Cf. Nir 4.23. Indeed, ṚV 1.89.10 reads in its entirety: aditir dyaur aditir
antarikṣam aditir mātā sa pitā sa putraḥ / viśve devā aditiḥ pañca janā aditir jātam
aditir janitvam.
49 MīSū 1.2.41: sataḥ param avijñānam.
50 ŚBh II.66,1–67,1: nigamaniruktavyākaraṇavaśena dhātuto ’rthaḥ kalpayitavyaḥ. The
Vedic passages cited in the Nirukta, e.g., as revealing the meanings of words are
called nigamas. As suggested by Kullūkabhaṭṭa ad MS 4.19 (paryāyakathanena vedā‐
rthabodhakān nigamākhyāṃś ca granthān), compilations of such examples with ex‐
planations may have existed. Kumārila and Śabara generally recommend resorting
to etymology in explaining unknown Vedic words, yet they also recognize the prin‐
ciple that the established conventional meaning (rūḍhārtha), when it is known, is
more authoritative. See, e.g., TV II.149,2–154,4 ad MīSū 1.3.10. As for grammar, the
eighth adhikaraṇa of the Mīmāṃsādarśana, MīSū 1.3.24–29, is devoted to establish‐
ing its importance for Vedic study. See below, p. 141–143 nn. 86–87.
51 TV II.66,2–3: tatra cārthaprakaraṇasūktadevatārṣanigamaniruktavyākaraṇajñānāny
adhigamopāyāḥ, teṣāṃ hy evamartham eva paripālanam.

52 TV II.66,10–14.
134 JOHN TABER
offer detailed explanations of the pūrvapakṣin’s examples of mantras

whose meaning is unknown.53 Note that Śabara and Kumārila do not
appeal here to an “authorless explanation that has come down through
an unbroken tradition,” as Dharmakīrti has his Mīmāṃsaka suggest,54
not to mention expositors endowed with a special capacity for per‐
ceiving supersensible things! Although we may recognize certain au‐
thorities, they are simply those who have proven themselves the most
skilled in applying the above‐mentioned considerations.55
3.3. The determination of figurative meaning

As we have seen, one must often resort to figurative or secondary
meaning in order to make sense of arthavādas and mantras, yet there
is a method for doing that, as well.56 MīSū 1.4.2357 gives the grounds for
construing a word figuratively – literally, as a “statement of a secon‐
dary aspect or feature” (guṇavāda) – when its primary meaning is
blocked. As Śabara and Kumārila explain at length, a word can be em‐
ployed in a secondary sense when it refers to a quality related to its
primary meaning.58 (The Mīmāṃsā theory of metaphor is quite
53 E.g., of sṛṇy eva jarbharī turpharītū (ṚV 10.106.6), TV II.66,18–67,9. Cf. Nir 13.5. Cf.
Sāyaṇa’s explanation (ṚVBh 4.695), which is quite similar. See Garge 1941–42.
54 See PVSV 168,15–16 above, pp. 45–46. But see also below, pp. 141–142 n. 86.
55 Kumārila (TV II.67,25–27) concludes his discussion of MīSū 1.2.41: tad evaṃ sarva‐
tra kenacit prakāreṇābhiyuktānām arthotprekṣopapatteḥ prasiddhatarārthābhāve

’pi vedasya tadabhyupagamāt siddham arthavattvam. “Thus, in this way, since it is
possible in every case for those who are versed in [the Veda] to reflect on its mean‐
ing in some way, it is established that the Veda is meaningful, even though a com‐
monly known meaning is absent, because [those experts] accept this (?).”
56 Cf. the complaint of the pūrvapakṣin at the beginnning of the discussion of artha‐
vādas, TV II.2,7–14.
57 Taken as one sūtra in the Jośī edition of Mīmāṃsādarśana 1.2–2.1: tatsiddhir jātiḥ
sārūpyaṃ praśaṃsā bhūmā liṅgasamavāyaḥ. Jha breaks it into six sūtras, 1.4.23–28.
58 ŚBh II.315,5–321,2 and TV ad loc. The word “lion,” e.g., indicates as its primary
meaning something in which courage predominates. One uses “lion” figuratively

when one intends to refer to the quality of courage in something, e.g., a man, which
is related to the primary meaning of the word. Śabara (ŚBh II.315,5) formally ety‐
mologizes guṇavāda as guṇād eṣa vādaḥ, “This statement is due to a quality.” Never‐
theless, he uses the expression in his discussion as meaning “the statement of a
quality/secondary aspect” (e.g., ŚBh II.315,6: katham aguṇavacano guṇaṃ brūyāt).
sophisticated and merits detailed treatment by someone.) The quality

in question can be “accomplishing that [function]” (tatsiddhi), as when
it is said, “The grass bundle (prastara) is the sacrificer” (TaitS 2.6.5.3).
The sacrificer is the most important of all the factors of the sacrifice;
when the prastara is placed upon the barhis59 and thereby, as it were,
given a place of honor, it assumes a role similar to that of the sacrificer;
it accomplishes the same function.60 Or the quality in question may be
that of “origin” (jāti) – that is to say, having the same origin – as in the
case of the sentence “The Brahmin is of fire” (TaitB 2.7.3.1); for both
the Brahmin and fire are said to have come out of the mouth of Prajā‐
pati. Or it can be visual similarity (sārūpya), as in the case of the sen‐
tence “The post is the sacrificer” (par. ŚB 13.2.6.9); for both stand erect.
And so on. In sum, the determination of figurative meaning in Mīmāṃ‐
sā is not a subjective matter, either, but follows established rules.
3.4. The appeal to common usage (lokaprasiddhi)

Of course, the main guiding principle for determining the meanings of
Vedic words is the fact, as Mīmāṃsā sees it, that the words of the Veda
are the same as the words of ordinary parlance and have the same
meanings. As I mentioned above, this principle is stated in MīSū 1.3.30
and developed in the commentaries thereto.61 Dharmakīrti criticizes
the Mīmāṃsakas, in PV(SV) 1.319–324, for employing this principle
arbitrarily. They insist on it, he implies, when it suits their purposes
and abandon it when it doesn’t. The Mīmāṃsādarśana, however –
though actual practice of course may have been different – is always
careful to justify when it applies and when it doesn’t.
On the face of it, according to Śabara’s pūrvapakṣin in ŚBh 1.3.30,
the words of the Veda and ordinary parlance seem to be different. We
have such sentences as “The cows of the gods move on their backs”
(ĀpŚSū 11.7.6) and “May Agni kill the Vṛtras (vṛtrāṇi)” (par. ṚV
59 Namely, in the course of the Darśapūrṇamāsa sacrifice. See TaitS 1.1.11.
60 TV II.315,21. Cf. however McCrea 2000: 438, who offers a different analysis.
61 Again, the basic idea is already found articulated by Yāska, Nir 1.16: arthavantaḥ
[mantrāḥ] śabdasāmānyāt. Cf. MīSū 1.2.32 (aviśiṣṭas tu vākyārthaḥ) in the discus‐

sion of the meaningfulness of mantras, and ŚBh (ad loc.) II.57,5: aviśiṣṭas tu loke pra‐
yujyamānānāṃ vede ca padānām arthaḥ.
136 JOHN TABER
6.16.34). Whatever “cows” and “Agni” refer to in these sentences, they

cannot be the same things we mean when we use those words in com‐
mon speech.62 But the siddhāntin insists that we recognize the words
themselves as familiar; we don’t perceive any difference between them
and common words.63 It is just a question of how they are being em‐
ployed in these sentences. The word “cows” in the above mantra, e.g., is
used to refer to divine cows, and we would expect divine cows to have
rather different properties from ordinary ones.64 But the core meaning
of the word is the same. As Kumārila puts it, just as men with snub
noses and hooded ears (? karṇaprāvaraṇa) are no less men, so cows
with different properties from ordinary ones, e.g., dwarf cows, are still
cows.65 The cows that are referred to in Vedic texts, moreover, often
are just ordinary ones, especially when it is a matter of the dakṣinā to
be given to priests! In any case, when we proceed on the assumption
that the meanings of words are the same as those of ordinary parlance,
from which they are in fact phonetically indistinguishable, then the
Veda for the most part makes sense; it consists of intelligible injunc‐
tions, and the application of its mantras to ritual acts is clear.66
62 In the case of Agni, it is well known that Indra, not Agni, is the slayer of the Vṛtras.
Meanwhile, vṛtra in common parlance is masculine, not neuter. See the various
other reasons for holding Vedic words to be different cited by Kumārila, TV
II.231,15–19.
63 ŚBh II.232,5: na teṣām eṣāṃ ca vibhāgam upalabhāmahe. See TV II.233,21–25 for
Kumārila’s expansion of this point.

64 TV II.233,8–9: yathāśrutagavādīnāṃ yāpi vācyāntare śrutiḥ / arthaikatvāvirodhena
guṇamātrānyatāparā //. Or else, as Kumārila suggests (TV II.233,23–234,1), the

sentence could mean that from the perspective of the gods in heaven cows on earth
appear to be moving on their backs!
65 TV II.233,10–13.
66 This seems to be the point being made by MīSū 1.3.30 itself: prayogacodanābhāvād
arthaikatvam avibhāgāt. “There is sameness of meaning [of Vedic and ordinary

words], because there is [the comprehension of] injunctions and applications,
because there is no [perceived] difference.” Śabara and Kumārila interpret the two
ablatives of the sūtra as providing separate reasons. The first reason for sameness
of meaning – thereby, implicitly, sameness of the words themselves, Kumārila
clarifies (TV II.232,18–19) – is that it makes possible the comprehension of
injunctions from the Veda and the applications of its mantras. Śabara and Kumārila,
however, both consider this an inadequate response, since it indicates merely a pra‐
yojana for considering the words and meanings the same, not a reason why they
In fact, the vast majority of words in the Veda are words we recog‐
nize. It is because we are confident that uttānā vahanti means “move
on their backs” that we suspect that gavāḥ, “cows,” is being used in an
unusual sense. When we hear the sentence agnihotraṃ juhuyāt svarga‐
kāmaḥ,67 the words juhuyāt and svargakāmaḥ, at least, are already
quite familiar to us. We know that someone desirous of heaven is being
told to make some sacrifice. The only real question can be, what exactly
does agnihotra mean? Yet even then, the various possibilities for ana‐
lyzing the word etymologically are clearly given. Those are to be
weighed, moreover, in light of the fact that it is not just the Vedic texts
that have been handed down, but also all the rituals that go with them,
and the associations of certain rituals with certain texts.
Indeed, the pertinent question in regard to agnihotraṃ juhuyāt svar‐
gakāmaḥ for the Mīmāṃsaka is not, Could it mean something com‐
pletely different from what scholars versed in the Veda have always
taken it to mean? Such a suggestion, for those who are otherwise able
to follow the Veda as a coherent body of prescriptions and prohibi‐
tions, would be simply absurd. The pertinent question is rather, ac‐
cording to MīSū 1.4.4 and the commentaries thereon, Does the word
agnihotra enjoin a guṇa, an accessory of the sacrifice – that is to say,
one of the complex of items that comprise the sacrificial act: the sub‐
stance offered, the instrument used to offer it, the deity to whom it is
offered, and so forth – or is it merely the name of a sacrifice?
The answer to this question is provided, then, according to the Mī‐
māṃsā method of analysis, as follows. If we take agnihotra as a bahu‐
vrīhi meaning that in which an offering (hotra) is made to Agni (agnaye
must be so considered (ŚBh II.232,3–5: ucyate prayojanam idam. hetur vyapadiśya‐

tām iti. tato hetur ucyate. avibhāgād iti). A second reason for sameness of meaning is
therefore required: namely, essentially, because we recognize the words as being
the same. It does seem possible, however, to read the two ablatives as nested: the
meanings of Vedic and ordinary words must be the same, because we comprehend
coherent injunctions from the Veda, due to the fact that we recognize the words
themselves as the same.
67 Which, however, is not precisely matched by any Vedic text. The texts that come
closest to it are TaitS 1.5.9.1: agnihotraṃ juhoti; KāṭhS 6.3: payasāgnihotraṃ juhoti;
and MaitS 1.8.6: ya evaṃ vidvān agnihotraṃ juhoti. Garge takes the Kāṭhaka and
Maitrāyaṇīya passages as providing the originative injunctions for the agnihotra;
see Garge 1952: 102 and 266. See also above, p. 40 n. 35.
138 JOHN TABER
hotram asminn iti), maintains the pūrvapakṣin, then it would appear to

enjoin one of the accessories of the sacrifice, specifically, the deity for
whom it is intended. “In this way,” Śabara states, “common usage will
be upheld.”68 The siddhāntin, however, points out that the deity of the
agnihotra is already indicated by another text.69 Since an injunction can
enjoin only that which is previously unknown,70 the mentioning of
agnihotra in agnihotraṃ juhuyāt svargakāmaḥ could not have injunc‐
tive force in reference to the deity. It must therefore function merely as
the name of a sacrifice to be offered to Agni; i.e., the sentence should be
taken to be enjoining the sacrifice itself, not one of its accessories.
Thus, the correct analysis of the word, still as a bahuvrīhi, will draw
attention to the action to which it refers: “that in which there is an of‐
fering to Agni, which is a sacrifice (homa), that is an agnihotra.”71 Ku‐
mārila even argues that the compound that is the basis of the bahuvrīhi
should be read, not as a dative tatpuruṣa – which would directly enjoin
an offering for Agni – but as a genitive tatpuruṣa, which merely
mentions Agni (“an offering of Agni”).72
In the cases of other names of sacrifices, discussed in other adhika‐
raṇas in the first part of MīSū 1.4, one must diverge even further from
common usage. In the injunction citrayā yajeta paśukāmaḥ (TaitS
2.4.6.1) one would normally assume citrayā to refer to a speckled
cow;73 thus, the injunction should be taken as enjoining a particular
kind of victim. But this would entail a “syntactic split” (vākyabheda) –
68 ŚBh II.285,4–5: prasiddhir evam anugrahīṣyate.
69 Namely, yad agnaye prajāpataye ca sāyaṃ juhoti, par. MaitS 1.8.7.
70 ŚBh II.285,12–13: aviditavedanaṃ ca vidhir ity ucyate. See Kataoka 2003b. Kataoka
argues that the idea that a vidhi can only enjoin something not previously known
was the basis for the well‐known Mīmāṃsā doctrine that a pramāṇa is “the appre‐
hension of an unapprehended object” (agṛhītārthagrahaṇa). His thesis is plausible,
but I do not think one can exclude the possibility that the derivation went in the
other direction.
71 ŚBh II.286,1–2: yasminn agnaye hotraṃ homo bhavati tad agnihotram.
72 Similarly, we speak of aśvaghāsa, “grass of horses” (TV II.287,23–288,11).
73 For a discussion of the problems associated with this citation, see Garge 1952: 29–
30, 83.
more than one thing would be enjoined in a single statement.74 Thus, in

this instance, too, we must interpret citrā as the name of the sacrifice.
Similarly, for udbhidā yajeta (Tāṇḍya Brāhmaṇa 19.7.275), one would
normally, on the basis of the known meanings of the components of the
word udbhid, take it to be referring to the axe.76 In that case, however,
since yajeta actually means, according to the Mīmāṃsā analysis of op‐
tative verbs, yāgena kuryāt, one would have to construe udbhidā figu‐
ratively as udbhidvatā; that is to say, one would read the sentence as
udbhidvatā yāgena kuryāt. But it is preferable to avoid resorting to a
secondary meaning whenever possible – another basic Mīmāṃsā rule.
Therefore, one should take udbhid as the name of the sacrifice being
enjoined. The sentence, then, is a primary injunction of a certain sacri‐
fice.77
In sum, although the meaning of the word in ordinary parlance is in
most cases the key to understanding a Vedic word, there are circum‐
stances where it has to be abandoned. Yet this is never to be done arbi‐
trarily, without concrete justification, as Dharmakīrti seems to sug‐
gest.78 One of the main purposes of Mīmāṃsā, especially in the section
74 If citrā referred to a property of the victim, then the sentence would specify that a
sacrifice previously mentioned should be carried out by means of a cow that is

female and speckled, and that the sacrifice will have cattle as its result. All that
cannot be enjoined by a single sentence! (ŚBh II.280,1–281,1 ad MīSū 1.4.3). Cf.,
however, the treatment of aruṇayā piṅgākṣyā ekahāyanyā somaṃ krīṇāti (par. TaitS
6.1.6.7) in ŚBh IV.33,2–35,5 ad 3.1.12, which one would expect to be parallel, but
which differs for subtle reasons. See McCrea 2000: 442–446.
75 According to Jha; not confirmed by Garge 1952.
76 TV II.271,5–8.
77 ŚBh II.274,5–277,1 ad MīSū 1.4.2.
78 Even when it comes to svarga, one of Dharmakīrti’s examples of a word that Mī‐
māṃsakas do not construe according to prasiddhi, Śabara offers an elaborate justifi‐

cation, ad MīSū 6.1.1–3 (ŚBh V.173,13–184,4), for why it is not to be taken in its cus‐
tomary sense. There, the topic of the adhikaraṇa is whether svarga, in injunctions
such as darśapūrṇamāsābhyāṃ svargakāmo yajeta, “One who desires heaven should
sacrifice with the new and full moon ceremonies,” refers to the principal factor
(pradhāna) of the ritual or to a secondary element (guṇa), with the implication that
if it refers to the latter, then the purport of the injunction is that only one who de‐
sires heaven is qualified to carry out the sacrifice (the full argument is too complex
to relate here). Such a view is maintained by the pūrvapakṣin, who supports it by
alleging that according to ordinary usage (laukikaḥ prayogaḥ), heaven is a sub‐
140 JOHN TABER
of the Mīmāṃsādarśana under discussion (MīSū 1.2–1.4), is to provide

guidelines for when it is to be abandoned and when not.
3.5. The resolution of polysemy

Yet often words in ordinary parlance are polysemic. Dharmakīrti asks,
Who can determine which of several meanings of the ordinary word
the Vedic word is supposed to have?79 Now, the problem of how to de‐
cide between multiple meanings of a word is taken up in the fourth
(Yavavarāha) adhikaraṇa of Mīmāṃsāsūtra 1.3 (sūtras 8 and 9).80 As
Śabara introduces the topic, we find the words yava, varāha, and vetasa
in the Veda. Yet these words are used differently among different
speakers in different locales (deśa). In some places yava refers to bar‐
ley (dīrghaśūka), in others to mustard seed (priyaṅgu); for some speak‐
ers varāha means a pig, for others, a kind of bird; and in some places
vetasa means cane (? vañjula), in others, the rose‐apple tree (jambu).81
How do we decide in which senses these, and other apparently
polysemic words, are being used in the Veda? The pūrvapakṣin main‐
tains that, since different meanings are established among different
communities of speakers in each case, there should be an option.82 The
stance (dravya) – as we gather when people say, “Fine silken clothes are heaven,”
“Sixteen‐year‐old girls are heaven,” etc. – in general, “a substance that yields delight
(prītimad dravyam)” (ŚBh V.176,4). He rejects the common employment of svarga
for a certain place, since there is no evidence for the existence of such a place (ŚBh
V.177,9–16)! Since substances are commonly intended as being for the purpose of
actions, hence as subordinate factors, heaven would therefore be a subordinate
factor. In the end this interpretation is rejected by the siddhāntin, who argues for
deviating from the common usage cited by the pūrvapakṣin, yet apparently agreeing
with him that we should not take svarga as referring to a particular place, and
interpreting heaven rather simply as “delight” or “happiness” (prīti), i.e., the
principal thing to be brought about by such sacrifices.
79 See PV(SV) 1.319cd and 323 above, pp. 44 and 54–55.
80 This is the fifth adhikaraṇa in Jha’s translations of the Śābarabhāṣya and Tantra‐
vārttika.
81 ŚBh II.139,2–140,2.
82 MīSū 1.3.8: teṣv adarśanād virodhasya samā vipratipattiḥ. “Because of not seeing any
opposition in their case, a different view is equal in force.” Kumārila (TV II.140,7–
10) presents the leading idea of the pūrvapakṣa as follows: yatra deśe hi yaḥ śabdo
yasminn arthe prayujyate / śaktis tadgocarā tasya vācikākhyā pramīyate // tasyāś ca
siddhānta, however, as stated by MīSū 1.3.9, is that “[those meanings],

rather, which are based on scripture [are to be chosen]; for that is the
basis [of Dharma].”83 Or, as Śabara somewhat freely explains the sūtra,
“The meaning of a word [in the Veda] is what it is for those who base
themselves on scripture,”84 i.e., those “learned” (śiṣṭa) in the Veda. In‐
deed, they are the basis (nimitta) for ascertaining śruti and smṛti.85
What this means, once again, is not that such people have some capac‐
ity others lack, or access to an eternal, authorless tradition of exegesis,
even though Śabara speaks of their commanding “an unbroken mem‐
ory of words and the Veda.”86 It means, rather – as I understand it –
sarvagāmitvaṃ tannyāyatvāt pramīyate / naikeṣām eva sā hy asti keṣāṃ cid vā na

vidyate //. “If a word is used in a particular meaning in a particular place, it is cogni‐
zed that it has a śakti, called expressive, which has that [meaning] as its scope. And
it is cognized that that [śakti] pervades everywhere, since it is suitable for that.
Indeed, it does not exist for only some people but not exist for others.”
83 MīSū 1.3.9: śāstrasthā vā tannimittatvāt.
84 ŚBh II.141,4: yaḥ śāstrasthānāṃ sa śabdārthaḥ.
85 ŚBh II.141,5–142,1: śiṣṭā nimittaṃ śrutismṛtyavadhāraṇe. I take śiṣṭa as having a
stronger connotation than merely “cultured,” which it seems to have, e.g., in the
Dharmaśāstras.
86 ŚBh II.141,5: teṣām avicchinnā smṛtiḥ śabdeṣu vedeṣu ca. We encounter other locu‐
tions like this in the pūrvapakṣa of the eighth adhikaraṇa (1.3.24–29) of Kumārila’s
TV. This adhikaraṇa concerns whether the Veda makes any restriction regarding
the correct use of words, hence, whether the study of the grammatical smṛti texts
(such as Pāṇini) pertains to Dharma. (The Grammarian tradition of course held that
the study of vyākaraṇa does pertain to Dharma; see, e.g., MBh I.8,3–22.) The pūrva‐
pakṣin argues to the contrary that grammar is of no relevance, maintaining at one
juncture in his long discourse (TV II.200,3–202,14) that it is not necessary or even
helpful in resolving doubts about the meaning of the Veda (asandeha is one of the
purposes of grammar listed by Patañjali, MBh I.13,14); for other means suffice –
first and foremost, “the usage of elders” (vṛddhavyavahāra), but also nigama, niruk‐
ta, kalpasūtra, and tarka. See TV II.200,3–6: asandehaś ca vedārthe yad apy prayoja‐
nam / tad apy asad yato nāsmāt padavākyārthanirṇayau // yataḥ padārthasandehās
tāvad bahavo vṛddhavyavahārād eva nivartante. śeṣāś ca nigamaniruktakalpasūtra‐
tarkābhiyuktebhyaḥ sarveṣām arthapratipādanaparatvāt. Yet in this passage the
pūrvapakṣin also refers to “the tradition of knowers of the meaning of the Veda”
(vedārthavitparamparā) and “the tradition of expositors” (vyākhyātṛparamparā),
and even suggests at one point that the explanation of the Veda is as eternal as the
Veda itself (TV II.201,18–21): na cāgṛhītaśabdārthaiḥ kaiś cid vyākaraṇāśrayaṇāt /
vyākhyātuṃ śakyate vedo yataḥ syāt tena niścayaḥ // yathaivāvasthito vedas tathā
vyākhyāpi sarvadā / ataḥ sthūlapṛṣatyādivyākhyā vyākaraṇād ṛte //. Kumārila does
142 JOHN TABER
that their judgment is informed by lifelong study of the Veda; they are
steeped in its meaning. They know, in particular, the entire Veda, the
contexts in which words occur and how they are used; they have also
studied grammar and the kalpasūtras. Just as one would want to con‐
sult someone who has devoted his life to studying Dharmakīrti and
Buddhist literature when trying to figure out a difficult passage in
Dharmakīrti, so would one want to consult one who has devoted his
life to studying the Veda and its auxiliary sciences when trying to
ascertain the meanings of Vedic words.87
not endorse this view in his siddhānta, but neither does he explicitly reject it. This
passage suggests in any case that there very well could have been some Mīmāṃsa‐
kas who believed that the meaning of Vedic terms is secured in part by an unbroken,
authorless tradition of exegesis. It is well known that, elsewhere, Kumārila is deci‐
dedly guarded toward any tradition that is not ultimately grounded on the Veda.
This pertains especially to heterodox practices, such as those of the Saṃsāramoca‐
kas and the Buddhists (Halbfass 1983: 15–16). But at TV II.75,16–23 he rejects a
mere tradition of remembering an injunction to perform the aṣṭakā (mentioned MS
4.150), without any Vedic basis, as insufficient for establishing its authority, becau‐
se it would be like a tradition of the blind (andhaparamparā). Cf. ŚBh II.73,1–74,2
(pūrvapakṣa).
87 Cf. however, TV II.141,14–15: rasavīryavipākānāṃ bhedād vaidyair yavādayaḥ / nir‐
dhāryāḥ svārthatattvena dharmasiddhyaiva yājñikaiḥ //. “[The meaning of] yava and
so forth are ascertained by experts in medicine from differences in taste, strength,
and ripening, according to the nature of their meanings [i.e., the things themselves],
whereas they are [ascertained] by experts in the sacrifice by virtue of their accom‐
plishing Dharma.” This, however, need not be taken as implying a supernormal abili‐
ty of perceiving Dharma on the part of experts in sacrifice, but a profound know‐
ledge of all aspects of Dharma and its ramifications as defined by the Veda. Thus,
Kumārila will say later, in the siddhānta of the eighth adhikaraṇa (on grammar),
that those thoroughly versed in the rules of grammar can perceive correct and incor‐
rect words (sādhutvam indriyagrāhyam) – just as those versed in the Dharmaśāstras
can tell if someone is a Brahmin just by looking at him (TV II.217,13–218,4)! There,
he will employ the analogy of the expert jeweler (TV II.219,16–21): yady apy ana‐
bhiyuktānāṃ prayogo ’sti sasaṅkaraḥ / abhiyuktā vivekṣyante tathāpi brāhmaṇādi‐
vat // … yathā ratnaparīkṣāyāṃ sādhvasādhutvalakṣaṇam / tathā vyākaraṇāt sid‐
dhaṃ sādhuśabdanirūpaṇam //. By contrast, other ancient authorities explicitly at‐
tributed the ability to directly perceive Dharma to ṛṣis, who then in turn taught the
mantras to lesser beings who lack any supernormal ability. Thus, Yāska, Nir 1.20:
sākṣātkṛtadharmāṇa ṛṣayo babhūvuḥ. te ’varebhyo ’sākṣātkṛtadharmabhya upade‐
śena mantrān samprāduḥ. Cf. MBh I.11,11–12. Such statements are conspicuously
missing in Kumārila. Indeed, such a view is rejected by Bhartṛhari, VPR 1.150
Thus, those versed in the Veda will know, in particular, the

continuation of the passage (vākyaśeṣa), which of itself often clarifies
which meaning is intended. They are able to recite (samāmananti), for
example, the following text which occurs after “vessels full of yava”
have been enjoined:88 “When other plants wither, these still stand, as if
rejoicing.”89 Since barley matures in early spring when other plants
have lost their leaves, this indicates that there is a practice of using
yava to refer to barley in the Veda. Similarly, there is the text, “There‐
fore the cows run after the varāhas,”90 which clarifies that in the Veda
varāha refers to pigs. Finally, those learned in the Veda will know the
text “The vetasa is water‐born,”91 which indicates that vetasa refers to
the vañjula plant, which grows in the water. The general principle that
applies to all these cases is enunciated at MīSū 1.4.29, “In doubtful
cases [the meaning is determined] from the continuation of the pas‐
sage.”92 Dharmakīrti indeed may be alluding to this principle when he
says, PVSV 170,5–11, that one cannot resolve the meaning of a
problematic Vedic passage by resorting to other Vedic passages, be‐
cause the meaning of those other passages are in doubt as well.93 This
criticism, however, assigns all Vedic sentences to the same level of in‐
comprehensibility and makes no distinction between those the mean‐
ing of which is more or less obvious and those whose meaning is ob‐
scure indeed.
(Cardona 2007: 697 and n. 25) and, as I interpret him, by Kumārila as well, ŚV
codanā 143–151.
88 ŚBh II.142,1: yavamayeṣu karambhapātreṣu vihiteṣu. It is uncertain which Vedic text
Śabara is referring to.

89 yatrānyā oṣadhayo mlāyante ’thaite modamānā ivottiṣṭhanti. Source unknown.
There are of course many passages cited by Śabara that have not been traced. Of
approximately 1700 passages cited, from a variety of śākhās, less than half have
been identified. Śabara only a few times indicates the source of a citation himself.
Many passages he quotes inexactly, or deliberately modifies, or combines with other
passages. Moreover, he may well have cited from Vedic texts that are now complete‐
ly lost or used compendia that no longer exist. For a comprehensive discussion see
Garge 1952, esp. pp. 39–50.
90 ŚB 4.4.3.19.
91 TaitS 5.3.12.2.
92 MīSū 1.4.29: sandigdheṣu vākyaśeṣāt.

144 JOHN TABER
Precisely because the meanings of these words become clear from

their context, however, and also disputing that these examples really
are polysemic to begin with,94 Kumārila suggests another interpreta‐
tion of the adhikaraṇa. How does one decide on the meaning of a word
used in the Veda when it has one meaning among Āryas and another
among Mlecchas? pilu, for example, among the Āryas means a particu‐
lar tree and among the Mlecchas an elephant. Here, the pūrvapakṣin
suggests again – rather provocatively, it would seem – that there
should be an option; for both words are expressive within the commu‐
nities that employ them.
Just as there is the notion of the beginninglessness of the connection95 for
the meaning comprehended by Āryas, so for the meaning comprehended
by Mlecchas, for the reason is the same.96 For there is no interruption of
the use of it among the Mlecchas; and how could one comprehend a
difference of beginningless śaktis?97
The siddhāntin responds: The notion that Mleccha words are expres‐
sive is erroneous; one is misled by their similarity to Ārya words. “Just
as there are corruptions of [the forms of] words due to metaphorical
and incorrect employment, etc., so there are corruptions of meanings
(arthāpabhraṃśa).”98 And it is those versed in the śāstras who are able
to recognize them, “like [false coins] can be distinguished from among
both true and false coins by those who are expert in them.”99
94 TV II.142,16–19.
95 Read sambandhānāditāmatiḥ, pace TV and TVH 387,15.
96 Namely, there is no memory of the founder of the convention, and so forth.
97 TV
II.143,16–19: yathaiva hy āryagamye ’rthe sambandhānāditāmatiḥ / mleccha‐
gamye tathaiva syād aviśiṣṭaṃ hi kāraṇam // na prayogāvadhis tasya mleccheṣv api
hi dṛśyate / anādyor arthaśaktyoś ca viśeṣo gamyatāṃ katham //. Cf. ŚBh II.183,2–5
ad MīSū 1.3.24.
98 TV II.144,3: śabdāpabhraṃśavad eva gauṇabhrāntyādiprayoganimittā arthāpabhra‐
ṃśā bhavanti. A similar point is made by the pūrvapakṣin in adhikaraṇa 5 ad MīSū

1.3.10 (II.150,22): dharme cānādarāt teṣāṃ bhraṃśyetārtho ’pi śabdavat. The pūrva‐
pakṣin goes on to explain how Mleccha words only seem to have denotations by
virtue of their similarity of Sanskrit words. Kumārila apparently follows the Gram‐
marians in this matter; see VPR 1.175–183.
99 TV II.144,4–5: sādhvasādhukarṣāpaṇamadhyād iva tatparīkṣibhir vivektum.
Thus, because they are expert in the [various] recensions, the view of those
who reside in Āryāvarta is thought to be valid knowledge (pramā) of
what is a means of Dharma or the meaning of a word.100
And those who know more śāstras are accorded more authority when
it comes to what is the true meaning of a word in the Veda.101, 102
4. SUMMARY AND OBSERVATIONS

This will have to suffice as a very brief and inadequate survey of ways
in which Mīmāṃsakas believed they were able to ascertain meaning of
100TV II.144,7–8: ataḥ śākhābhiyuktatvād āryāvartanivāsinām / yā matiḥ saiva dharmā‐
ṅgaśabdārthatvapramā matā //. My italics, of course, to emphasize that the reason

is not “because they can perceive Dharma”!
101 TV II.144,10–11: abhiyuktatarā ye ye bahuśāstrārthavedinaḥ / te te yatra prayuñji‐
ran so so ’rthas tattvato bhavet //.

102 Kumārila nevertheless goes on to argue in his interpretation of the fifth adhikaraṇa
(according to Jośī’s enumeration) of MīSū 1.3 (sūtra 10) that words that are current
only among Mlecchas should be construed as they are used conventionally by them
and should not be given artificial designations by means of grammar and etymology.
Kumārila goes on to consider yet another interpretation of the fourth adhikaraṇa
according to which it concerns what to do when words have different meanings in
ordinary parlance and the Veda (TV II.146,3ff.). For example, caru means a dish in
ordinary language, but in the Veda it means a kind of porridge, usually of rice (oda‐
na) which is mixed with curds or milk (despite the phrase cited by Śabara at the
beginning of the adhikaraṇa ad MīSū 1.3.8 [ŚBh II.139,2]: yavamayaś caruḥ [source
unknown]). Here too, however, Kumārila is inclined to set aside common usage
when the text of the Veda clearly demands construing a word according to a diffe‐
rent meaning. This is, as before, generally clear from the entire passage. After it is
enjoined, “The caru is to be offered to Aditi in the milk,” it is said, “[One pleases]
Aditi by cooked rice (odana).” Thus, “the word caru, by virtue of its śakti being ex‐
cluded from the illogical designation of multiple meanings, is restricted to just one
meaning according to the common usage of the Veda and those versed in sacrifice,
and refers to cooked rice” (TV II.148,11–12: tathā caruśabdo ’py anyāyānekārthā‐
bhidhānapratibaddhaśaktitvād ekatra niyamyamāno yājñikavedaprasiddhibhyām
odanaviṣaya eva bhavati; cf. TV II.146,12: yājñikānāṃ ca vede ca prasiddhis tv oda‐
naṃ prati //). From this and other statements one sees that prasiddhi evidently
means for Kumārila established usage relative to a literature or a community. It is
not necessarily equivalent to lokavāda.
146 JOHN TABER
the Veda by objective methods. Many more examples could be given.103

These examples show that Mīmāṃsā consists in a system of estab‐
lished rules and procedures, as objective as those of logic, for inter‐
preting scriptural passages. Starting from the entirely reasonable as‐
sumption that most of the words in the Veda are ordinary Sanskrit
words, then resorting to grammatical analysis, etymology, the examina‐
tion of context, the collation of other occurrences of words and parallel
passages, and knowledge of the rituals with which Vedic passages are
associated, Mīmāṃsakas were confident that they could come up with
plausible hypotheses about the meanings of Vedic words and sen‐
tences that are otherwise unknown or obscure. They – or at least the
Mīmāṃsakas we have been considering, Śabara and Kumārila – make
no appeal to the authority of individual teachers with supernatural
ability – such persons are never mentioned, as far as I can tell, though
to be sure the existence of bona fide experts in Vedic interpretation
was acknowledged – or any eternal authorless tradition of interpreta‐
tion. Indeed, their methods were not altogether unlike the “objective”
methods employed by modern philologists today in interpreting Vedic
texts.
Thus, when Dharmakīrti asserts, “Vedic words do not cry out,
‘Come, you revered Brahmins, this is to be grasped as our meaning, not
something else,’” we can imagine the Mīmāṃsaka responding, “To the
contrary, the Veda can be made to yield up its secrets if one rigorously
applies the proper methodology.” The Mīmāṃsaka would have been just
as supremely confident that he could make sense of the Veda as is the
modern Vedic scholar Michael Witzel, who writes,
If we follow these rules and use all the other tools mentioned earlier [in
discussing the editing of the Kaṭha Āraṇyaka], we can achieve in Vedic
Studies a certainty that approaches that of the natural sciences. In fact,
we can proceed in a similar fashion, by trial and error, and by proposing a
theory [about the exact meaning of a text] and actually testing it. Only
when the word, concept, or custom is hapax or is attested too infre‐
quently to allow a proper investigation of the whole range of meanings,
103 For further discussion of rules for interpreting the meaning of words, see Garge
1952: 252–256. Much of the second kāṇḍa of Bhartṛhari’s Vākyapadīya is devoted to

how to determine the meanings of words in their various contexts. See, e.g., VPR
2.303ff.
must we remain content with a merely probable answer, or a mere guess.

In all other cases, of course, only after painstaking study, we can conclude
that yes the theory was right [, thereby arriving confidently at the mean‐
ing of the passage], or no it was not.104
Of course, the Mīmāṃsaka would never suggest, as Prof. Witzel does,

that one is able “to enter the Vedic mind,” i.e., discern “the original in‐
tent of the composers of the text.”105 He would, rather, claim to be able
to grasp the “intent” of the Veda itself.106
It is the appreciation of Mīmāṃsā as a fixed system of rules, indeed,
as a system of “logic” independent of human invention, which may be
applied correctly or incorrectly, that is conspicuously lacking in Dhar‐
makīrti.107 More specifically: He knows that the Mīmāṃsakas appeal to
prasiddhi and vākyaśeṣa, but he seems not to have any idea what that
entails in specific instances, that the application of these concepts is
constrained in all kinds of ways. He knows that the Mīmāṃsakas
generally distrust what ordinary people say – hence, it would seem,
one cannot rely on common usage as a pramāṇa in determining the
meanings of words; but he ignores the fact that the ordinary meaning
of a word is something that is agreed upon by everyone, hence it is evi‐
104 Witzel 1996: 174–5, with bold type removed.
105 Witzel 1996: 167.
106 See Yoshimizu 2008.
107 Indeed, Bhartṛhari refers to a list of exegetical principles in VPR 1.152 (and Vṛtti),
similar to the sorts of rules invoked by Śabara in construing problematic Vedic pas‐
sages, as the kind of “reasoning that is in conformity with Vedic scripture, which
[serves as] the eye for those who do not see” (VPR 1.151ab: vedaśāstrāvirodhī ca tar‐
kaś cakṣur apaśyatām /), i.e., for those unable to comprehend the Veda through
their own insight. He goes on to state that such “reasoning, which resides in men, is
the capacity of the words themselves” (VPR 1.153ab: śabdānām eva sā śaktis tarko
yaḥ puruṣāśrayaḥ /); see Cardona 2007: 699–700. That is to say, the principles one
follows both in using language to express one’s intention and in interpreting the
meaning of an utterance when it is spoken are determined by the reality of language
itself – one could say, they are objective. “Language alone is the teacher” (śabda
evopadeṣṭā, VPV 209,1). Cf. Bronkhorst 1997: 368: “[According to Mīmāṃsā] the
Veda … is not produced by a human mind … nor should it be interpreted by a mind.
Or rather, only those interpretations which reduce the interference of the mind to a
minimum can be accepted as correct.” Bronkhorst in that essay argues along much
the same lines as I have argued here.
148 JOHN TABER
dently not a matter of opinion or judgement (where the reliability of

one’s informants would actually come into play).108 He knows that Mī‐
māṃsakas interpret certain words in strange ways (svarga, urvaśī), yet
he seems hardly aware of the extensive roles played by figurative
meaning and etymological meaning in Mīmāṃsā discussions. He knows
there are problems with polysemic words in the Veda, but he does not
appreciate the variety of tools the Mīmāṃsaka has at his disposal for
solving them. He knows that the Mīmāṃsakas recognize certain au‐
thorities, but does not seem to be aware that their status for the Mī‐
māṃsakas is based on their learning, not personal charisma – which
Mīmāṃsakas are outspokenly suspicious of – or supernormal ability.
He knows that the Mīmāṃsakas were aware that the transmission of
the Veda was subject to the vicissitudes of time,109 but he doesn’t seem
to understand that that would be moot for most Mīmāṃsakas – cer‐
tainly the ones we have been considering here, Śabara and Kumārila –
when it comes to the problem of interpreting the Veda, since they do
not appeal to any authorless tradition of Vedic exegesis. He imagines
that agnihotraṃ juhuyāt svargakāmaḥ might mean “One should eat dog
meat,” but he doesn’t realize that there was a legitimate question about
what kind of injunction it is, and, more generally, he is oblivious to the
question of the nature of vidhi and all the other issues concerning vidhi
with which not just Mīmāṃsaka but most other Brahmanical schools as
well were deeply preoccupied.
Thus, in the end, we arrive at the not very surprising result that,
while Dharmakīrti exhibits broad knowledge of problems of Mīmāṃsā
exegesis in PV(SV) 1.312–340, he displays nothing like an expert’s
command. It is possible he received some training in Mīmāṃsā at an
early stage, but it would have been an introductory course. The sorts of
criticisms he directs against Mīmāṃsā exegetical practices do not hold
up very well, and do not really address the essence of Mīmāṃsā, which
is, again, that those practices are dictated by, and in turn testify to –
insofar as they work – the inherent meaningfulness of the Vedic corpus;
it is possible to ascertain the objective meaning of a text. I believe that
most modern scholars working on textual materials believe this – that
108 Cf. ŚV sambandhākṣepaparihāra 14cd.
109 Though, interestingly, he does not mention the fact that Mīmāṃsakas thought that
some Vedic texts have been completely lost.

it is possible to get at the meaning of a text, removed from any

consideration of the author’s intention110 – and so, to that extent, they
too are Mīmāṃsakas!
It would have been surprising, however, if matters turned out any
differently. Dharmakīrti, after all, is addressing his own community –
other Buddhists. He is not trying to convince Mīmāṃsakas or, it would
appear, even get at the truth. In general, philosophical debate in classi‐
cal India was characterized by jalpa, “disputation,” not saṃvāda,
“discussion.” It was acceptable to raise prima facie objections to the
views of one’s adversaries, as a means of protecting one’s own position,
without necessarily having to represent those views sympathetically or
completely accurately, placing the burden on one’s opponent to set the
record straight; nor did one go very far considering the best responses
to one’s objections.111 In short, a classical Indian philosopher was not
obliged to become an expert in the systems he was criticizing, and
Dharmakīrti was no exception. I must say, as a philosopher, I find this
final stage of his critique of Mīmāṃsā, for all its cleverness, rather
disappointing. The real strengths of the Mīmāṃsā position are
unappreciated, the deepest issues are not addressed. Even though his
criticisms of Mīmāṃsā go farther than those of other Buddhists,112
Dharmakīrti in the end must be seen, in his critique of Mīmāṃsā, as
another example of the dominant polemical tendency of classical In‐
dian thought.
110 From Wimsatt and Beardsley to Roland Barthes, Jacques Derrida, and Paul Ricœur,
modern critics and philosophers have pointed out various ways in which texts
transcend their authors and so must be interpreted without regard to “author’s
intention.”
111 One
may contrast Socrates’ consideration of the views of Protagoras in Plato’s
Theaetetus 165d–168c.
112 E.g., Bhāviveka, and even his successors Śāntarakṣita, Kamalaśīla, and Śubhagupta,
whose critiques of Mīmāṃsā exegesis are not as sophisticated as Dharmakīrti’s and

were probably derived from his.

Abbreviations and Bibliography

A — Aṣṭādhyāyī (Pāṇini). Otto Böhtlingk: Pâṇini’s Grammatik (herausgegeben, über‐
setzt, erläutert und mit verschiedenen Indices versehen). Hildesheim 1964 (Leipzig
1887): Georg Olms Verlagsbuchhandlung.
AASP — Austrian Academy of Sciences Press = Verlag der Österreichischen Akademie
der Wissenschaften.
AJP — H.R. Kāpadīā: Anekāntajayapatākā by Haribhadra Sūri with his own commentary
and Municandra Sūri’s supercommentary. Baroda 1940–1947: Oriental Institute
(GOS 88, 105).
Akamatsu 2010 — Akihiko Akamatsu: Anumāna in Bhartṛhari’s Vākyapadīya. Pp. 183–
190 in Brendan Gillon (ed.): Logic in Earliest Classical India. Delhi 2010: Motilal
Banarsidass.
AK(Bh) — Prahlad Pradhan: Abhidharmakośabhāṣyam of Vasubandhu. Patna 1975
(1967): KPJRI (TSWS 8).
Akimoto 1993 — Masaru Akimoto: Seiten ron, Dharmottara no ‘Pramāṇaviniścayaṭīkā’
(2). 聖典論 — ダルモッタラの『プラマーナ・ヴィニシュチャヤ・ティーカー
』(2). [On Āgama – An annotated translation from the Pramāṇaviniścayaṭīkā of
Dharmottara (2).] Pp. 137–149 in Egaku Mayeda (ed.): Studies in Original Bud‐
dhism and Mahāyāna Buddhism in commemoration of late Professor Dr. Fumimaro
Watanabe. Vol. II. Kyoto 1993: Nagata Bunshodo.
Aklujkar 1989 — Ashok Aklujkar: The number of pramāṇas according to Bhartṛ‐hari.
WSTB 33 (1989), pp. 151–158.
AKVy — Unrai Wogihara: Sphuṭārthā Abhidharmakośavyākhyā, the work of Yaśomitra.
Tokyo 1989 (1936): Sankibo Buddhist Book Store (The Publishing Association of
Abhidharmakośavyākhyā).
ĀpŚSū — Richard Garbe: The Śrauta Sūtra of Āpastamba belonging to the Taittirīya
Saṃhitā with commentary of Rudradatta. 3 vols. Calcutta 1882–1902 (Bibliotheca
Indica).
AS — Georges Thibaut: Arthasaṃgraha of Laugākṣī Bhāskara. Benares 1882 (Benares
Sanskrit Series 4).
152 ABBREVIATIONS AND BIBLIOGRAPHY
ASBh — Nathmal Tatia: Abhidharmasamuccaya‐bhāṣyam. Patna 1976: KPJRI (TSWS

17).
ĀśvŚSū — Rāmanārāyaṇa Vidyāratna: The Śrauta Sūtra of Āśvalāyana, with the com‐
mentary of Gārgya Nārāyaṇa. 2 vols. Calcutta 1864–1874 (Bibliotheca Indica 49).
ATBS — Arbeitskreis für tibetische und buddhistische Studien.
AVP — George Melville Bolling/Julius von Negelein: The Pariśiṣṭas of the Atharvaveda.
Leipzig 1909–1910: Otto Harrassowitz.
BaudhŚSū — Willem Caland: Baudhāyana Śrauta Sūtra. 3 vols. Calcutta 1904–1923:
Asiatic Society.
BCAPSh — Swami Dwarika Das Shastri: Bodhicaryāvatāra of Ārya Śāntideva with the
commentary Pañjikā of Shri Prajñākaramati and Hindi translation. Varanasi 1988:
Bauddha Bharati (Bauddha Bharati Series 21).
BCAPLVP — Pp. 234–388 in Louis de La Vallée Poussin: Bouddhisme. Études et maté‐
riaux. Ādikarmapradīpa. Bodhicaryāvatāraṭīkā. London 1898: Luzac & Co.
BhārŚSū — C.G. Kashikar: The Śrauta, Paitṛmedhika and Pariśeṣa Sūtras of Bharadvāja.
Poona 1964: Vaidika Saṃśodhana Maṇḍala.
Bhāskarī — K.A.S Iyer/K.C. Pandey: Īśvarapratyabhijñāvimarśinī of Abhinavagupta, doc‐
trine of divine recognition. 3 vols. Delhi 1986 (Allahabad 1938/1950/1954): Mo‐
tilal Banarsidass.
BHSD — Franklin Edgerton: Buddhist hybrid Sanskrit grammar and dictionary. Vol. II:
Dictionary. Delhi 1993 (New Haven 1953): Motilal Banarsidass.
Biardeau 1964 — Madeleine Biardeau: Théorie de la connaissance et philosophie de la
parole dans le brahmanisme classique. Paris/La Haye 1964: Mouton & Co (Le
monde d’outre‐mer passé et présent, Études 23).
BKGA — Beiträge zur Kultur‐ und Geistesgeschichte Asiens.
BoBhū — Nalinaksha Dutt: Bodhisattvabhūmi (Being the XVth section of Asaṅgapādaʼs
Yogācārabhūmi). Patna 1978: KPJRI (TSWS 7).
Bronkhorst 1994 — Johannes Bronkhorst: Studies on Bhartṛhari, 2: Bhartṛhari and
Mīmāṃsā. Pp. 371–388 in R.C. Dwivedi (ed.): Studies in Mīmāṃsā, Dr. Mandan
Mishra felicitation volume. Delhi 1994: Motilal Banarsidass.
Bronkhorst 1997 — Johannes Bronkhorst: Philosophy and Vedic Exegesis in the Mī‐
māṃsā. Pp. 359–371 in Franco/Preisendanz 1997.
Cabezón 1992 — José Ignacio Cabezón: Vasubandhu’s Vyākhyāyukti on the authenticity
of the Mahāyāna Sūtras. Pp. 221–243 in Jeffrey R. Timm (ed.): Texts in context.
Traditional hermeneutics in South Asia. Albany 1992: SUNY.
ABBREVIATIONS AND BIBLIOGRAPHY 153
Cardona 1980 — George Cardona: Pāṇini. A survey of research. Delhi 1980 (The Hague
1976): Motilal Banarsidass.
CŚ — Karen Lang: Āryadeva’s Catuḥśataka on the bodhisattva’s cultivation of merit and
knowledge. Copenhagen 1986: Akademisk Forlag (Indiske Studier 7).
Cardona 2007 — George Cardona: On the position of Vyākaraṇa and Pāṇini. Pp. 693–
710 in Karin Preisendanz (ed.): Expanding and merging horizons: Contributions to
South Asian and cross‐cultural studies in commemoration of Wilhelm Halbfass.
Vienna 2007: AASP (BKGA 53).
CTPH — China Tibetology Publishing House.
D — J. Takasaki/Z. Yamaguchi/Y. Ejima: sDe dge Tibetan Tripiṭaka bsTan ’gyur pre‐
served at the Faculty of Letters, University of Tokyo. Tokyo 1981ff.
Devasthali 1969 — G.V. Devasthali: Paribhāṣā (Introduction and general survey).
Indian Antiquary 3 (3rd Series) (1969), pp. 1–13.
DPŚ — *Daśapadārthaśāstra (*Candramati). See Miyamoto 2007.
Dreyfus 1997 — Georges Dreyfus: Recognizing reality, Dharmakīrti’s philosophy and its
Tibetan interpretations. Albany 1997: SUNY.
D’Sa 1980 — Francis X. D’Sa: Śabdaprāmāṇyam in Śabara and Kumārila. Towards a
study of the Mīmāṃsā experience of language. Vienna 1980 (Publications of the De
Nobili Research Library 7).
Dunne 2004 — John D. Dunne: Foundations of Dharmakīrti’s philosophy. Boston 2004:
Wisdom Publications (SITB).
Eckel 2008 — Malcolm David Eckel: Bhāviveka and his opponents. Cambridge, Mass.
2008: Harvard University Press (Harvard Oriental Series 70).
Eltschinger 2000 — Vincent Eltschinger: “Caste” et philosophie bouddhique. Continuité
de quelques arguments bouddhiques contre le traitement réaliste des dénominations
sociales. Vienna 2000: ATBS (WSTB 47).
Eltschinger 2001 — Vincent Eltschinger: Dharmakīrti sur les mantra et la perception du
suprasensible. Vienna 2001: ATBS (WSTB 51).
Eltschinger 2007a — Vincent Eltschinger: Penser l’autorité des Écritures: La polémique
de Dharmakīrti contre la notion brahmanique orthodoxe d’un Veda sans auteur.
Eltschinger 2007b — Vincent Eltschinger: On 7th and 8th century Buddhist accounts of
human action, practical rationality and soteriology. Pp. 135–162 in B. Kellner/H.
Krasser/H. Lasic/M.T. Much/H. Tauscher (eds.): Pramāṇakīrtiḥ. Papers dedicated
to Ernst Steinkellner on the occasion of his 70th birthday. Part 1. Vienna 2007:
ATBS (WSTB 70.1).
Eltschinger 2008 — Vincent Eltschinger: Dharmakīrti on mantras and their efficiency.
Pp. 273–289 in ICEBS (= Executive Committee of the International Conference on
Esoteric Buddhist Studies, Koyasan University, Sept. 2006): Esoteric Buddhist stu‐
dies: Identity in diversity. Koyasan 2008: Koyasan University.
Eltschinger 2010 — Vincent Eltschinger: Dharmakīrti. Revue Internationale de Philoso‐
phie 64/3 [Buddhist philosophy] (2010), pp. 397–440.
Eltschinger, forthcoming — Vincent Eltschinger: Ignorance, epistemology and soterio‐
logy – Part 2. Journal of the International Association of Buddhist Studies 33 (2010)
[forthcoming].
Franco 1987 — Eli Franco: Perception, knowledge and disbelief. A study of Jayarāśi’s
scepticism. Stuttgart 1987: Franz Steiner Verlag.
Franco 1997 — Eli Franco: Dharmakīrti on compassion and rebirth. Vienna 1997: ATBS
(WSTB 38).
Franco/Preisendanz 1997 — Eli Franco/Karin Preisendanz (eds.): Beyond orientalism.
The work of Wilhelm Halbfass and its impact on Indian and cross‐cultural studies.
Amsterdam/Atlanta 1997: Rodopi (Poznań Studies in the Philosophy of the
Sciences and the Humanities 59).
Frauwallner 1932 — Erich Frauwallner: Beiträge zur Apohalehre. 1. Dharmakīrti.
Übersetzung. WZKM 39 (1932), pp. 247–285.
Frauwallner 1933 — Erich Frauwallner: Beiträge zur Apohalehre. 1. Dharmakīrti.
Übersetzung (Fortsetzung). WZKM 40 (1933), pp. 51–94.
Frauwallner 1934 — Erich Frauwallner: Dharmakīrtis Saṃbandhaparīkṣā. Text und
Übersetzung. WZKM 41 (1934), pp. 261–300.
Frauwallner 1954 — Erich Frauwallner: Die Reihenfolge und Entstehung der Werke
Dharmakīrti’s. Pp. 142–154 in: Asiatica. Festschrift Friedrich Weller. Leipzig 1954:
Otto Harrassowitz.
Frauwallner 1959 — Erich Frauwallner: Dignāga, sein Werk und seine Entwicklung.
Wiener Zeitschrift für die Kunde Süd‐ und Ostasiens 3 (1959), pp. 83–164.
Frauwallner 1968 — Erich Frauwallner: Materialien zur ältesten Erkenntnislehre der
Karmamīmāṃsā. Vienna 1968: Hermann Böhlaus Nachf./Kommissionsverlag der
Österreichischen Akademie der Wissenschaften (Veröffentlichungen der Kom‐
mission für Sprachen und Kulturen Süd‐ und Ostasiens 6).
Garge 1941–1942 — D.V. Garge: The contribution of the Śabara‐Bhāṣya to Ṛgveda

exegesis. Or the treatment of Śabara of the Ṛgveda passages cited in his Bhāṣya.
Bulletin of the Deccan College Research Institute 3 (1941–1942), pp. 531–546.
Garge 1952 — D.V. Garge: Citations in Śābara‐Bhāṣya. Poona 1952: Deccan College.
GCh — D. Chattopadhyaya: Tāranātha’s history of Buddhism in India. (rGya gar chos
’byuṅ). Simla 1970: Indian Institute of Advanced Study.
Gnoli 1960 — See PVSV.
Gonda 1988 — Jan Gonda: Mantra interpretation in the Śatapatha‐Brāhmaṇa. Leiden
1988: E.J. Brill.
GOS — Gaekwad’s Oriental Series.
Halbfass 1983 — Wilhelm Halbfass: Studies in Kumārila and Śaṅkara. Reinbek 1983:
Verlag für Orientalistische Fachpublikationen.
Hara 1991 — Minoru Hara: A Note on the ancient Indian oath (III). Pp. 51–65 in Li
Zheng/ Jiang Zhongxin et al.: Papers in honour of Prof. Dr. Ji Xianlin on the occasion
of his 80th birthday. Vol. I. Nanchang 1991: Popular Press of the Province Jiangxi.
Hattori 1968 — Masaaki Hattori: Dignāga, on perception, being the Pratyakṣapari‐
ccheda of Dignāga’s Pramāṇasamuccaya. Cambridge, Mass. 1968: Harvard Uni‐
versity Press.
Hayes 1980 — Richard P. Hayes: Diṅnāgaʼs views on reasoning (svārthānumāna). JIP 8
(1980), pp. 219–277.
Hayes 1984 — Richard P. Hayes: The question of doctrinalism in the Buddhist episte‐
mologists. Journal of the American Academy of Religion 52/4 (1984), pp. 645–670.
Hayes 1988 — Richard P. Hayes: Dignāga on the interpretation of signs. Dordrecht
1988: Kluwer (Studies in Classical India, 9).
HBṬĀ — Hetubinduṭīkāloka (Durvekamiśra). Sukhlaji Sanghavi/Muni Shri Jinavijayaji:
Hetubinduṭīkā of Bhaṭṭa Arcaṭa with the sub‐commentary entitled Āloka of Durveka
Miśra. Baroda 1949: Oriental Institute.
Hoornaert 2000 — Paul Hoornaert: An annotated translation of Madhyamakahṛdaya‐
kārikā / Tarkajvālā V.8–26. Kanazawa Daigaku Bungakubu Ronshū – Kōdōka‐
tetsugakuhen 金沢大学文学部論集行動科・哲学編 [Studies and Essays, Behav‐
ioral Sciences and Philosophy, Faculty of Letters, Kanazawa University] 20 (2000),
pp. 75–111.
Hopkins 1974 — E. Washburn Hopkins: Epic mythology. Delhi 1974 (Strasbourg 1915):
Motilal Banarsidass.
Houben 1997 — Jan E.M. Houben: Bhartṛhari’s perspectivism (1): The Vṛtti and Bhartṛ‐
hari’s perspectivism in the first kāṇḍa of the Vākyapadīya. Pp. 317–358 in
Franco/Preisendanz 1997.
HV — Harivaṃśa. See MBh (vol. V).
ĪPV — M. R. Shāstrī: Īśvarapratyabhijñāvimarśinī. 2 vols. Srinagar 1918–1921: Nirnaya‐
Sagar Press (Kashmir Series of Texts and Studies 22, 33).
ĪPVV — M. K. Shāstrī: Īśvarapratyabhijñāvivṛtivimarśinī by Abhinavagupta. 3 vols.
Bombay 1938–1943: Nirnaya Sagar Press (Kashmir Series of Texts and Studies 60,
62, 65).
Iyer 1965 — K.A. Subramania Iyer: The Vākyapadīya of Bhartṛhari with the Vṛtti. Chap‐
ter I. Poona 1965: Deccan College.
Iyer 1992 — K.A. Subramania Iyer: Bhartṛhari. A study of the Vākyapadīya in the light of
the ancient commentaries. Poona 1992 (1969): Deccan College, Post Graduate and
Research Institute.
Jha — Ganganatha Jha: Śābara‐Bhāṣya. 3 vols. Baroda 1973: Oriental Institute (GOS
66).
JIP — Journal of Indian Philosophy.
Kataoka 2003a — Kei Kataoka: Kumārila’s critique of omniscience. Indo‐Sisōshi Kenkyū
インド思想史研究 (Studies in the History of Indian Thought) 15 (2003), pp. 35–69.
Kataoka 2003b — Kei Kataoka: The Mīmāṃsā definition of pramāṇa as a source of new
information. JIP 31 (2003), pp. 89–103.
Kataoka 2007 — Kei Kataoka: Kumārila’s notion of pauruṣeyavacana. Rivista di Studi
Sudasiatici 2 (2007), pp. 39–55.
Kataoka 2011a — Kei Kataoka: Kumārila on truth, omniscience, and killing. Part 2. A
critical edition of Mīmāṃsā‐Ślokavārttika ad 1.1.2 (codanāsūtra). Vienna 2011:
AASP. [For part 1, see ŚV codanā.]
Kataoka 2011b — Kei Kataoka: Manu and the Buddha for Kumārila and Dharmakīrti.
Pp. 255–269 in H. Krasser/H. Lasic/E. Franco/B. Kellner (eds.): Religion and logic
in Buddhist philosophical analysis. Proceedings of the Fourth International Dhar‐
makīrti Conference. Vienna, August 23–27, 2005. Vienna 2011: AASP.
Kataoka, forthcoming — Kei Kataoka: Transmission of scripture: Exegetical problems
for Kumārila and Dharmakīrti. Forthcoming in V. Eltschinger/H. Krasser (eds.):
Scriptural authority and apologetics in the Indian religio‐philosophical environment
(Proceedings of a panel organized at the 14th World Sanskrit Conference, Kyoto, Sep‐
tember 1–5, 2009).
KāṭhS — Leopold von Schroeder: Kāṭhaka. Die Saṃhitā der Kaṭha‐Śākhā. 3 vols. Leipzig
1900–1910: E.A. Brockhaus.
Katsura 1984 — Shoryu Katsura: Dharmakīrti’s theory of truth. JIP 12 (1984), pp. 215–
235.
Katsura 1999 — Shoryu Katsura (ed.): Dharmakīrtiʼs thought and its impact on Indian
and Tibetan philosophy. Vienna 1999: AASP (BKGA 32).
Kawasaki 1976 — Shinjo Kawasaki: The Mīmāṃsā chapter of Bhavya’s Madhyamaka‐
hṛdaya‐kārikā: Text and translation (1) Pūrva‐pakṣa. Studies (Institute of Philoso‐
phy, University of Tsukuba) (1976), pp. 1–15.
Keira 2006 — Ryusei Keira: The proof of voidness and scriptural authority –
Kamalaśīlaʼs way of adopting scriptures. Pp. 177(638)–192(623) in Mochizuki
Kaishuku 望月海淑 (ed.): Hokkekyō to Daijōkyōten no Kenkyū 法華経と大乗経典
の研究 [The Lotus Sutra and related systems]. Tokyo 2006: Sankibō Busshorin.
KPJRI — Kashi Prasad Jayaswal Research Institute.
Krasser 1999 — Helmut Krasser: Dharmakīrti’s and Kumārila’s refutations of the exis‐
tence of god: A consideration of their chronological order. Pp. 215–223 in Katsura
1999.
Krasser 2001 — Helmut Krasser: On Dharmakīrti’s understanding of pramāṇabhūta
and his definition of pramāṇa. WZKS 45 (2001), pp. 173–199.
Krasser 2004 — Helmut Krasser: Are Buddhist pramāṇavādins non‐Buddhistic? Dignā‐
ga and Dharmakīrti on the impact of logic and epistemology on emancipation.
Hōrin 11 (2004), pp. 129–146.
Krasser 2012 — Helmut Krasser: Bhāviveka, Dharmakīrti and Kumārila. Pp. 535–594
in François Voegeli, Vincent Eltschinger, Danielle Feller, Maria Piera Candotti,
Bogdan Diaconescu & Malhar Kulkarni (eds), Devadattīyam, Johannes Bronkhorst
Felicitation Volume. Bern, etc. 2012: Peter Lang:
KSP — Karmasiddhiprakaraṇa (Vasubandhu).
Kunjunni Raja 1963 — K.K. Kunjunni Raja: Indian theories of meaning. Madras 1963:
The Adyar Library and Research Centre (The Adyar Library Series 91).
La Vallée Poussin 1980 — Louis de La Vallée Poussin: L’Abhidharmakośa de Vasuban‐
dhu. 6 vols. Bruxelles 1980 (Paris/Louvain 1923–1931): Institut Belge des Hautes
Études Chinoises (Mélanges Chinois et Bouddhiques 16).
Lasic 2010 — Horst Lasic: A hot dispute about lukewarm air: Dignāga on āptavāda. Pp.
509–525 in Piotr Balcerowicz (ed.): Logic and belief in Indian philosophy. Delhi
2010: Motilal Banarsidass.
MĀ — Madhyamakāloka (Kamalaśīla). D no. 3887, Sa 133b4–244a7.

MaitS — Leopold von Schroeder: Maitrāyaṇī Saṃhitā. Die Saṃhitā der Maitrāyaṇīya‐
Śākhā. 4 vols. Leipzig 1881–1886: E.A. Brockhaus.
MBh — Franz Kielhorn: The Vyākaraṇa‐Mahābhāṣya of Patañjali. 3rd ed. revised and
furnished with additional readings, references, and select critical notes by K.V.
Abhyankar. 3 vols. Poona 1962–1972 (1880): Dandekar Oriental Research Insti‐
tute.
MBhār — Mahābhārata‐saṃhitā: cikitsitapāṭhatimikā (The Mahābhārata text as consti‐
tuted in its critical edition). 5 vols. Poona 1971–1976: Bhandarkar Oriental Re‐
search Institute.
McClintock 2010 — Sara L. McClintock: Omniscience and the rhetoric of reason. Śānta‐
rakṣita and Kamalaśīla on rationality, argumentation, and religious authority. Bos‐
ton 2010: Wisdom Publications (SITB).
McCrea 2000 — Lawrence McCrea: The hierarchical organization of language in Mī‐
māṃsā interpretive theory. JIP 28 (2000), pp. 429–459.
McCrea 2009 — Lawrence McCrea: ‘Just like us, just like now’: The tactical implications
of the Mīmāṃsā rejection of yogic perception. Pp. 55–70 in Eli Franco (ed.): Yogic
perception, meditation and altered states of consciousness. Vienna 2009: AASP
(BKGA 65).
Mīmāṃsādarśana — Kāśīnātha Vāsudevaśāstrī Abhyaṅkar/Gaṇeśaśāstrī Jośī: Śrīmajjai‐
minipraṇītaṃ Mīmāṃsādarśanam. 2nd edition (with 2nd pt., MīSū 1.2–2.1, shown as
edited only by Gaṇeśaśāstrī Jośī and dated 1981). 7 pts. Vārāṇasī 1972–1981:
Ānandāśrama Press (Ānandāśrama Sanskrit Series 97).
MNP — Mahamahopadhyaya Vasudev Shastri Abhyankar: Mīmāṃsā‐Nyāya‐Prakāśa by
Āpadeva. Poona 1972: Bhandarkar Oriental Research Institute (Government
Oriental Series‐Class A 3).
MīSū — Mīmāṃsāsūtra (Jaimini). See Mīmāṃsādarśana.
Miyamoto 2007 — Keiichi Miyamoto: Daśapadārthī. An ancient literature of thoroughly
metaphysical realism. Kyoto 2007: Rinsen Book Co.
Monier‐Williams 1963 = Monier Monier‐Williams: A Sanskrit‐English dictionary (Ety‐
mologically and philologically arranged with special references to cognate Indo‐
European languages). Delhi 1963 (Oxford 1899): Motilal Banarsidass.
Moriyama, forthcoming — Shinya Moriyama: On the role of abhyupagama in Dharma‐
kīrti’s scripturally based inference. Forthcoming in V. Eltschinger/H. Krasser
(eds.): Scriptural authority and apologetics in the Indian religio‐philosophical envir‐
onment (Proceedings of a panel organized at the 14th World Sanskrit Conference,

Kyoto, September 1–5, 2009).
MS — J.L. Shastri: Manusmṛti, with the Sanskrit commentary Manvartha Muktāvalī of
Kullūka Bhaṭṭa. Delhi 1983: Motilal Banarsidass.
ms(s) — Manuscript(s).
NBh — See NS.
NBṬ — F.I. Ščerbatskoj: Nyāyabindu БУДДІЙСКІЙ УЧЕЂНИКЂ ЛОГИКИ СОЧИНЕНІЕ
ДАРМАКИРТИ И ТОЛКОВАНІЕ НА НЕГО Nyāyabinduṭīkā СОЧИНЕНІЕ ДАРМОТ‐
ТАРЫ. Osnabrück 1970 (Saint‐Petersbourg 1904): Biblio Verlag.
NBṬV — Louis de La Vallée Poussin: Tibetan translation of the Nyāyabindu of Dharma‐
kīrti (A treatise on Buddhist logic) with the commentary of Vinītadeva. Calcutta
1984 (1907): The Asiatic Society (Bibliotheca Indica 171).
Nir — Lakshman Sarup: The Nighaṇṭu and the Nirukta. The oldest Indian treatise on
etymology, philology and semantics. Delhi 1967 (1920–1927): Motilal Banarsidass.
NM — K.S. Varadacharya: Nyāyamañjarī of Jayantabhaṭṭa with Ṭippaṇī‐Nyāyasaurabha
by the editor. 2 vols. Mysore 1969: Oriental Research Institute (University of
Mysore, Oriental Research Institute Series 116).
NSū — Anantalal Thakur: Gautamīyanyāyadarśana with Bhāṣya of Vātsyāyana (Nyā‐
yacaturgranthikā, vol. I). Delhi 1997: Indian Council of Philosophical Research.
NV — Anantalal Thakur: Nyāyabhāṣyavārttika of Bhāradvāja Uddyotakara (Nyāya‐
caturgranthikā, vol. II). Delhi 1997: Indian Council of Philosophical Research.
Oberhammer 1974 — Gerhard Oberhammer: Die Überlieferungsautorität im Hinduis‐
mus. Pp. 41–92 in Gerhard Oberhammer (ed.): Offenbarung, geistige Realität des
Menschen. Arbeitsdokumentation eines Symposiums zum Offenbarungsbegriff in
Indien. Vienna 1974 (Publications of the De Nobili Research Library 2).
Oberhammer et al. 1991 — Gerhard Oberhammer/Joachim Prandstetter/Ernst Prets:
Terminologie der frühen philosophischen Scholastik in Indien. Ein Begriffs‐wörter‐
buch zur altindischen Dialektik, Erkenntnislehre und Methodologie. Vol. I: A–I.
Oetke 1988 — Claus Oetke: “Ich” und das Ich. Analytische Untersuchungen zur buddhis‐
tisch‐brahmanischen Ātmankontroverse. Stuttgart 1988: Franz Steiner Verlag.
Ōmae 1999 — Futoshi Ōmae: Dharmakīrti as a varṇavādin. Pp. 295–300 in Katsura
1999.
Ono et al. — M. Ono/J. Oda/J. Takashima: KWIC index to the Sanskrit texts of Dharma‐
kīrti. Tokyo 1996: Institute for the Study of Languages and Cultures of Asia and
Africa, Tokyo University of Foreign Studies (Lexicological Studies 8).
P — Daisetz T. Suzuki: The Tibetan Tripiṭaka: Peking edition. Tokyo/Kyoto 1957: Tibe‐
tan Tripitaka Research Institute.
PDhS — Johannes Bronkhorst/Yves Ramseier: Word index to the Praśastapādabhāṣya.
A complete word index to the printed editions of the Praśastapādabhāṣya. Delhi
1994: Motilal Banarsidass.
Perry 1997 — Bruce M. Perry: Early Nyāya and Hindu orthodoxy: ānvīkṣikī and adhi‐
kāra. Pp. 449–470 in Franco/Preisendanz 1997.
Preisendanz 1994 — Karin Preisendanz: Studien zu Nyāyasūtra III.1 mit dem Nyāyatat‐
tvāloka Vācaspati Miśras II. 2 vols. Stuttgart 1994: Franz Steiner Verlag.
Pruden 1988–1990 — Leo M. Pruden: Abhidharmakośabhāṣyam by Louis de La Vallée
Poussin. English Translation by Leo M. Pruden. 4 vols. Berkeley 1988–1990: Asian
Humanities Press.
PS 2 — See PSV
PSṬms — Manuscript B of Jinendrabuddhiʼs Pramāṇasamuccayaṭīkā. See the description
by A. MacDonald in E. Steinkellner/H. Krasser/H. Lasic: Jinendrabuddhiʼs Viśālāmala‐
vatī Pramāṇasamuccayaṭīkā, chapter 1. Part II: Diplomatic edition with a manuscript
description by Anne MacDonald. Beijing/Vienna 2005: CTPH/AASP (STTAR 1).
PSṬTib — Pramāṇasamuccayaṭīkā (Jinendrabuddhi). D no. 4268, Ye 1–314a7/P no. 5766,
Re 1–355a8.
PSVK — Pramāṇasamuccayavṛtti (Dignāga), Tibetan translation of PSV by Kanakavar‐
man and (Mar thuṅ ) Dad pa(ʼi) ś es rab. P no. 5702, Ce 93b4–177a7. No equivalent
in D.
PSVV — Pramāṇasamuccayavṛtti (Dignāga), Tibetan translation of PSV by Vasudharara‐
kṣita and (Źa ma) Seṅ (ge) rgyal (mtshan). D no. 4204, Ce 14b1–85b7/P no. 5701,
Ce 13a6–93b4.
PV 1 — See PVSV.
PV 2–4 — Yūsho Miyasaka: Pramāṇa‐vārttika‐kārikā (Sanskrit and Tibetan). Acta Indo‐
logica 2 (1972), pp. 1–206. [PV 2 = pramāṇasiddhi; PV 3 = pratyakṣa; PV 4 = parār‐
thānumāna.]
PVin 2 — Ernst Steinkellner: Dharmakīrti’s Pramāṇaviniścaya, chapters 1 and 2. Bei‐
jing/Vienna 2007: CTPH/AASP (STTAR 2).
PVin 3 — Pascale Hugon/Toru Tomabechi: Dharmakīrtiʼs Pramāṇaviniścaya, chapter 3.

With a preface by Tom J.F. Tillemans. Beijing/Vienna 2011: CTPH/AASP (STTAR
8).
PVinṬms — (Incomplete) Sanskrit manuscript of PVinṬ ad PVin 2. For a description see
Steinkellner 2007 = PVin 2: xxx–xxxi.
PVinṬTib — Pramāṇaviniścayaṭīkā (Dharmottara). D no. 4229 and 4227, Dze 1–Tshe
178a3.
PVSV — Raniero Gnoli: The Pramāṇavārttikam of Dharmakīrti. The first chapter with
the autocommentary. Roma 1960: Istituto Italiano per il Medio ed Estremo Oriente
(Serie Orientale Roma 23).
PVSVms — Manuscript B used by Gnoli for his edition of PVSV. For a description see Gnoli
1960 = PVSV: XXXVII–XXXVIII.
PVSVTib — Pramāṇavārttikasvavṛtti (Tibetan version). D no. 4216, Ce 261b1–365a7/P
no. 5717a, Ce 404b3–535a4.
PVSVṬ — Rāhula Sāṅkṛtyāyana: Ācārya‐Dharmakīrteḥ Pramāṇavārttikam (Svārthānu‐
mānaparicchedaḥ) Svopajñavṛttyā Karṇakagomiviracitayā taṭṭīkayā ca sahitam.
Kyoto 1982 (Allahabad 1943): Rinsen Books.
PVSVṬms — Shoren Ihara: Sanskrit manuscripts of Karṇakagomin’s Pramāṇavārtti‐
ka(sva)vṛttiṭīkā, facsimile edition. The Sanskrit commentaries on the Pramāṇavārt‐
tikam from Rāhula Sāṃkṛtyāyana’s collection of negatives, II. Patna/Narita 1998:
Bihar Research Society/Naritasan Institute for Buddhist Studies.
PVṬ — Pramāṇavārttikaṭīkā (Śākyabuddhi). D no. 4220, Je 1–Ñe 282a7/P no. 5718, Je
1–Ñe 348a8.
PVV — Rāhula Sāṅkṛtyāyana: Dharmakīrti’s Pramāṇavārttika with commentary by
Manorathanandin. Published as an appendix to the Journal of the Bhandarkar Ori‐
ental Research Institute 24–26 (1938–1940).
Ratié 2006 — Isabelle Ratié: La mémoire et le soi dans l’Īśvarapratyabhijñāvimarśinī
d’Abhinavagupta. Indo‐Iranian Journal 49 (2006), pp. 39–103.
Renou 1960 — Louis Renou: Le destin du Veda dans L’Inde. Études védiques et pāṇinéen‐
nes. Vol. IV. Paris 1960: Institut de Civilisation Indienne.
ṚV — Theodor Aufrecht: Die Hymnen des Rigveda. 2 vols. Wiesbaden 1968 (1877): Otto
Harrassowitz.
ṚVBh — N.S. Sontakke and C.G. Kashikar: Ṛgveda‐Saṃhitā with the commentary of Sā‐
yaṇācārya. 5 vols. Poona 1946: Vaidika Saṃśodhana Maṇḍala.
Sakai 2011 — Masamichi Sakai: Śākyabuddhi and Dharmottara on the inference of

momentariness based on the absence of external causes of destruction. Pp. 407–
421 in H. Krasser/H. Lasic/E. Franco/B. Kellner (eds.): Religion and logic in Bud‐
dhist philosophical analysis. Proceedings of the Fourth International Dharmakīrti
Conference. Vienna, August 23–27, 2005. Vienna 2011: AASP.
ŚāṅkhŚSū — Alfred Hillebrandt: Śāṅkhāyana Śrauta Sūtra together with the commen‐
tary of Varadattasuta Ānartīya and Govinda. 4 vols. Calcutta 1885–1899: Asiatic
Society.
Sarup 1967 — See Nir, “Introduction.”
Saundarananda — Linda Covill: Handsome Nanda, by Aśvaghoṣa. New York 2007: The
Clay Sanskrit Library.
ŚB — Albrecht Weber: The Śatapatha‐Brāhmaṇa in the Mādhyandina‐Śākhā with ex‐
tracts from the commentaries of Sāyaṇa, Harisvāmin and Dvivedagaṅga. Ber‐
lin/London 1855.
ŚBh — Śābarabhāṣya (Mīmāṃsāsūtrabhāṣya) (Śabarasvāmin). See Mīmāṃsādarśana.
ŚBhF — Erich Frauwallner: Śabarasvāmi’s Bhāṣyam zu den Mīmāṃsāsūtren I.1,1–5. Pp.
7–61 in Frauwallner 1968.
Silk 2008a — Jonathan A. Silk: Incestuous ancestries: The family origins of Gautama
Siddhārtha, Abraham and Sarah in Genesis 20:12, and the status of scripture in
Buddhism. History of Religions 47/4 (2008), pp. 253–281.
Silk 2008b — Jonathan A. Silk: Putative Persian perversities: Indian Buddhist
condemnations of Zoroastrian close‐kin marriage in context. Bulletin of the School
of Oriental and African Studies 71/3 (2008), pp. 433–464.
SITB — Studies in Indian and Tibetan Buddhism.
SK — Har Dutt Sharma: The Sāṃkhya‐kārikā. Īśvara Kṛṣṇa’s Memorable Verses on
Sāṃkhya philosophy with the commentary of Gauḍapādācārya. Poona 1933: Orien‐
tal Book Agency (Poona Oriental Series 9).
Skilling 2000 — Peter Skilling: Vasubandhu and the Vyākhyāyukti literature. Journal of
the International Association of Buddhist Studies 23/2 (2000), pp. 297–350.
SP(V) — Sambandhaparīkṣā(vṛtti) (Dharmakīrti). See Frauwallner 1934.
SS — Sarvajñasiddhi (Ratnakīrti). Pp. 1–28 in Anantalal Thakur: Ratnakīrtinibandhā‐
valiḥ. Patna 1975: KPJRI (TSWS 3).
Stcherbatsky 1984 — Theodore Stcherbatsky: Buddhist logic. Vol. I. Delhi 1984 (Lenin‐
grad 1930): Munshiram Manoharlal.
Steinkellner 1971 — Ernst Steinkellner: Wirklichkeit und Begriff bei Dharmakīrti.

WZKS 15 (1971), pp. 179–211.
Steinkellner 1979 — Ernst Steinkellner: Dharmakīrti’s Pramāṇaviniścayaḥ. Zweites
Kapitel: Svārthānumānam. Teil II: Übersetzung und Anmerkungen. Vienna 1979:
AASP (Veröffentlichungen der Kommission für Sprachen und Kulturen Südasiens
15).
Steinkellner 2007 — See PVin 2.
Steinkellner/Much 1995 — Ernst Steinkellner/Michael Torsten Much: Texte der er‐
kenntnistheoretischen Schule des Buddhismus. Systematische Übersicht über die
buddhistische Sanskrit‐Literatur II. Göttingen 1995: Vandenhoeck&Ruprecht (Ab‐
handlungen der Akademie der Wissenschaften in Göttingen, philologisch‐histori‐
sche Klasse, dritte Folge 214).
STTAR — Sanskrit Texts from the Tibetan Autonomous Region.
SUNY — State University of New York Press.
ŚV — Svāmī Dvārikadāsa Śāstrī: Ślokavārttika of Śrī Kumārila Bhaṭṭa, with the commen‐
tary Nyāyaratnākara of Pārthasārathi Miśra. Varanasi 1978: Tara Publications.
ŚV codanā — Kei Kataoka: Kumārila on truth, omniscience, and killing. Part 1. A critical
edition of Mīmāṃsā‐Ślokavārttika ad 1.1.2 (codanāsūtra). Vienna 2011: AASP. [For
part 2, see Kataoka 2011a.]
Taber 1989 — John Taber: Are mantras speech acts? Pp. 144–164 in Harvey Alper
(ed.): Understanding mantras. Albany 1989: SUNY.
Taber 2005 — John Taber: A Hindu critique of Buddhist epistemology: Kumārila on per‐
ception. London 2005: RoutledgeCurzon.
TaitS — Albrecht Weber: Die Taittirīya‐Saṃhitā. 2 vols. In Albrecht Weber: Indische
Studien 11 (1871) and 12 (1872). Hildesheim/New York 1973: Georg Olms.
Tib — Tibetan.
Tillemans 1986 — Tom J.F. Tillemans: Dharmakīrti, Āryadeva and Dharmapāla on
scriptural authority. Tetsugaku 哲學 [The Journal of Hiroshima Philosophical
Society] 38 (1986), pp. 31–47.
Tillemans 1990 — Tom J.F. Tillemans: Materials for the study of Āryadeva, Dharmapāla
and Candrakīrti. 2 vols. Vienna 1990: ATBS (WSTB 24).
Tillemans 1993 — Tom J.F. Tillemans: Persons of authority. The sTon pa tshad maʼi
skyes bur grub paʼi gtam of A lag sha Ngag dbang bstan dar, a Tibetan work on the
central religious questions in Buddhist epistemology. Stuttgart 1993: Franz Steiner
Verlag (Tibetan and Indo‐Tibetan Studies 5).
Tillemans 1997 — Tom J.F. Tillemans: Dharmakīrti on prasiddha and yogyatā. Pp. 177–
194 in Agata Bareja‐Starzyńska/Marek Mejor (eds.): Aspects of Buddhism. Pro‐
ceedings of the international seminar on Buddhist studies, Liw, 25 June 1994. War‐
saw: Oriental Institute, Warsaw University.
Tillemans 1999a — Tom J.F. Tillemans: How much of a proof is scripturally based in‐
ference (āgamāśritānumāna)? Pp. 395–405 in Katsura 1999.
Tillemans 1999b — Tom J.F. Tillemans: Scripture, logic, language. Essays on Dharma‐
kīrti and his Tibetan successors. Boston 1999: Wisdom Publications (SITB).
Tillemans 2000 — Tom J.F. Tillemans: Dharmakīrti’s Pramāṇavārttika. An annotated
translation of the fourth chapter (parārthānumāna). Vol. I (k. 1–148). Vienna 2000:
AASP.
TS(P)K — Embar Krishnamacharya: Tattvasaṅgraha of Śāntarakṣita with the commen‐
tary of Kamalaśīla. 2 vols. Baroda 1984 (1926): Oriental Institute.
TS(P)Ś — Swami Dwarikadas Shastri: Tattvasaṅgraha of Ācārya Shāntarakṣita with the
commentary ‘Pañjikā’ of Shri Kamalshīla. 2 vols. Varanasi 1981–1982 (1968):
Bauddha Bharati (Bauddha Bharati Series 1).
TSWS — Tibetan Sanskrit Works Series.
TUS — Tattvopaplavasiṃha (Jayarāśi). See Franco 1987.
TV — Tantravārttika (Kumārila Bhaṭṭa). See Mīmāṃsādarśana.
TVH — Kunio Harikai: Sanskrit Text of the Tantravārttika, Adhyāya 1, Pāda 3, Adhika‐
raṇa 4–6, Collated with six manuscripts. Fukuoka 2009: Kyushu University (South
Asian Classical Studies 4).
VādhŚSū — Braj Bihari Chaubey: Vādhūla‐Śrautasūtram. Critically edited with introduc‐
tion and indices. Hoshiarpur 1993: Katyayan Vaidik Sahitya Prakashan.
VaikhŚSū — Willem Caland: Vaikhānasa‐Śrautasūtram. The description of Vedic rites
according to the Vaikhānasa school belonging to the Black Yajurveda, for the first
time critically edited. Calcutta 1941: Royal Asiatic Society of Bengal.
VājS — Albrecht Weber: The Vājasaneyi‐Saṃhitā in the Mādhyandina and the Kāṇva‐
Śākhā with the commentary of Mahīdhara. Berlin/London 1852.
van Bijlert 1989 — Vittorio A. van Bijlert: Epistemology and spiritual authority. The de‐
velopment of epistemology and logic in the old Nyāya and the Buddhist school of
epistemology with an annotated translation of Dharmakīrtiʼs Pramāṇavārttika II
(Pramāṇasiddhi) vv. 1–7. Vienna 1989: ATBS (WSTB 20).
Verhagen 2008 — Peter Verhagen: Studies in Indo‐Tibetan Buddhist hermeneutics (6)
– Validity and valid interpretation of scripture according to Vasubandhu’s Vyā‐
khyāyukti. Pp. 233–258 in R. Gombrich/C. Scherrer‐Schaub (eds.): Buddhist

studies: Papers of the 12th World Sanskrit Conference, vol. VIII. Delhi 2008: Motilal
Banarsidass.
Vibh — Vibhūticandra’s marginal notes on PVV. See PVV.
Voegeli 2009 — François Voegeli: The Vth prapāṭhaka of the Vādhūlaśrautasūtra. Un‐
published PhD Thesis. Lausanne 2009: University of Lausanne, Faculty of Letters.
VPI — K. A. Subramania Iyer: Vākyapadīya of Bhartṛhari with the Vṛtti and the Paddhati
of Vṛṣabhadeva. Kāṇḍa I. Poona 1966: Deccan College.
VPR — Wilhelm Rau: Bhartṛharis Vākyapadīya: Die Mūlakārikās nach den Handschriften
herausgegeben und mit einem Pāda‐Index versehen. Wiesbaden 1977: Deutsche
Morgenländische Gesellschaft.
VPV — Vākyapadīyavṛtti. See Iyer 1965.
VSū — Muni Śrī Jambuvijayaji: Vaiśeṣikasūtra of Kaṇāda with the commentary of Can‐
drānanda. Baroda 1961: Oriental Institute (GOS 136).
VY — Vyākhyāyukti (Vasubandhu). Jong Cheol Lee: The Tibetan text of the Vyākhyāyukti
of Vasubandhu. Tokyo 2001: The Sankibo Press (Bibliotheca Indologica et Bud‐
dhologica 8).
Wakahara 1988 — Yūshō Wakahara: Mantora no kōka to zenchisha, Pramāṇavārttika‐
svavṛtti kenkyū マントラの効果と全知者, Pramāṇavārttikasvavṛtti 研究 (1) (vv.
292–311). [Effect of mantras and the omniscient. Pramāṇavārttikasvavṛtti Studies
(1)] Bukkyō Shigaku Kenkyū 仏教史学研究 [The Journal of the History of
Buddhism] 31 (1988), pp. 1–27.
Wakahara 1990 — Yūshō Wakahara: Dharmakīrti no Veda seiten hihan, Pramāṇavārt‐
tikasvavṛtti kenkyūダルマキールティのヴェーダ聖典批判, Pramāṇavārttikasva‐
vṛtti 研究 (2) (vv. 312–340). [Dharmakīrti’s criticism of Vedic scriptures.
Pramāṇavārttikasvavṛtti Studies (2).] Ryūkoku Daigaku Daigakuin Kiyō 龍谷大学
大学院紀要 [Ryukoku University, The Bulletin of the Graduate School of Letters] 11
(1990), pp. 10–33.
Witzel 1996 — Michael Witzel: How to enter the Vedic mind. Pp. 163–175 in E. Garzilli
(ed.): Translating, translations, translators from India to the West. Cambridge
1996: Harvard College (Harvard Oriental Series, Opera Minora 1).
WSTB — Wiener Studien zur Tibetologie und Buddhismuskunde.
WZKM — Wiener Zeitschrift für die Kunde des Morgenlandes.
WZKS — Wiener Zeitschrift für die Kunde Südasiens.
Yaita 1985 — Hideomi Yaita: On anupalabdhi, annotated translation of Dharmakīrti’s

Pramāṇavārttikasvavṛtti (I). Taishō Daigaku Daigakuin Kenkyū Ronshū 大正大学大
学院研究論集 [Journal of the Graduate School, Taisho University] 9 (1985), pp.
216–199.
Yaita 1987 — Hideomi Yaita: Dharmakīrti on the authority of Buddhist scriptures
(āgama) – An annotated translation of the Pramāṇavārttika‐svavṛttiḥ ad v. 213–
217. Nanto Bukkyō 南都佛教 [Journal of the Nanto Society for Buddhist Studies] 58
(1987), pp. 1–17.
Yaita 1988 — Hideomi Yaita: Dharmakīrti on the person free from faults, annotated
translation of the Pramāṇavārttikasvavṛttiḥ ad v. 218–223. Journal of Naritasan
Institute for Buddhist Studies 11 (1988), pp. 433–445.
Yaita 2005 — Hideomi Yaita: Appendix 3 – Dharmakīrtiʼs Pramāṇavārttikasvavṛtti ad
vv. 198–223. Pp. 423–455 in Hideomi Yaita: Three Sanskrit texts from the Buddhist
pramāṇa‐tradition – The Hetuvidyā section in the Yogācārabhūmi, the Dharmotta‐
raṭippanaka, and the Tarkarahasya. Narita 2005: Naritasan Shinshoji.
YBhū — Vidhushekhara Bhattacharya: The Yogācārabhūmi of Ācārya Asaṅga: the San‐
skrit text compared with the Tibetan version. Calcutta 1957: University of Calcutta.
YD — Albrecht Wezler/Shujun Motegi: Yuktidīpikā. The most significant commentary on
the Sāṃkhyakārikā. Vol. I. Stuttgart 1998: Franz Steiner Verlag (Alt‐ und Neu‐In‐
dische Studien 44).
Yoshimizu 1999 — Chizuko Yoshimizu: The development of sattvānumāna from the
refutation of a permanent existent in the Sautrāntika tradition. WZKS 43 (1999),
pp. 231–254.
Yoshimizu 2008 — Kiyotaka Yoshimizu: The intention of expression (vivakṣā), the ex‐
pounding (vyākhyā) of a text, and the authorlessness of the Veda. Zeitschrift der
Deutschen Morgenländischen Gesellschaft 158 (2008), pp. 51–71.

Indices

Index locorum

A DPŚ
1.1.22 • 60 10 • 65, 66
1.1.68 • 60 91 • 65
AJP HBṬĀ
I.56,6 • 40 368,1 • 40
I.323,11 • 39 HV
AK 21.10 • 50
3.45 • 42 ĪPV
AKBh I.64,10 • 39
76,22–23 • 83 ĪPVV
80,28–81,3 • 44 I.62,16 • 39
282,8–9 • 77 KāṭhS
460,2–3 • 83 6.3 • 40, 137
AKVy 19.8 • 62
183,16–17 • 45 19.9 • 62
ĀpŚSū 29.8 • 62
7.12.13–14 • 50 MaitS
11.7.6 • 135 1.2.7 • 50
AS 1.8.6 • 40, 137
17 • 132 1.8.7 • 138
32,1–7 • 40 3.9.5 • 50
ASBh 4.5.2 • 129
153,5 • 86 4.8.3 • 129
ĀśvŚSū MBh
10.9/2b • 62 I.8,3–22 • 141
AVP I.11,11–12 • 142
49.2,1 • 47 I.13,14 • 141
BaudhŚSū II.37,17 • 40
15.28 • 62 MBhā r
20.27 • 50 1.84.17 • 50
BCAPSh/BCAPLVP 3.78.3 • 50
327,7–13/295,17–23 • 67 12.329.19 • 50
BhārŚSū MHK
7.9.13–15 • 50 5.8–9 • 118
Bhāskarī 9.3ab • 119
I.98,9 • 39 9.15 • 119
BoBhū 9.16 • 119
25,19 • 83
168 INDICES
MīSū 1.16 • 135

1.1.4 • 123 1.20 • 142
1.1.24 • 40 4.23 • 133
1.1.31 • 50 5.11 • 132
1.2.1 • 128 5.13 • 50
1.2.7 • 130 5.14 • 126
1.2.11 • 131 6.15 • 132
1.2.31 • 132 13.5 • 132, 134
1.2.32 • 135 NM
1.2.41 • 133, 134 I.664,6–7 • 19
1.3.8 • 140, 145 I.664,16 • 19
1.3.8–9 • 140 II.220,4–5 • 40
1.3.9 • 141 II.220,10–11 • 40
1.3.10 • 144 NSū
1.3.24 • 144 1.1.10 • 66
1.3.24–29 • 133, 141 2.1.68 • 18
1.3.30 • 25, 124, 135, 136 NV
1.4.2 • 139 204,12 • 87
1.4.3 • 139 254,1 • 62
1.4.4 • 40, 137 254,4 • 40
1.4.23 • 134 255,1 • 62
1.4.29 • 143 337,14–15 • 68
6.1.1 • 49 PDhS
6.1.1–3 • 139 15,8–10 • 66
9.1.6–10 • 131 16,3–4 • 66
10.4.23 • 131 16,7–8 • 66
MNP PS
16,3–17,6 • 40 2.5ab • 63, 64, 84, 85, 87, 88, 89, 90, 91,
MS 92, 93, 95, 97, 99, 100, 101, 114, 115
4.19 • 133 2.28b • 63, 102, 103
4.150 • 142 PSṬms
NB 62b7 = PSṬTib D92a2/P103b1–2 • 93
3.114 • 111 63a4 • 93, 94
NBh 63a5 • 93
3,11–12 • 36 63a6 • 93
3,12 • 86 63a7 = PSṬTib D92b2–3/P104a3–4 • 93
96,11–97,16 • 18 63b5–6 = PSṬTib D93a1–2/P104b2–4 •
97,2 • 110 94
97,8–9 • 18 PSVK
97,15–16 • 18 111a2–3 • 93
NBṬ 111a3–4 • 93
81,19 • 55 115b2–4 • 63, 103
NBṬV PSVV
130,1–3 • 55 29a2–3 • 93
Nir 33a1–2 • 63
1.15 • 129
INDICES 169
PV 1.330–335 • 11
1.12 • 119 1.331 • 116
1.12c • 119 1.333 • 113
1.13d • 62 1.339 • 9
1.14 • 63, 103, 119 1.340 • 124
1.40–185 • 9 2.1ab1 • 87, 105
1.166ab • 68 2.1c2–2 • 59
1.199 • 75 2.5a • 120
1.213 • 64, 73, 75, 84, 101, 106, 107, 110, 2.5c • 105
117 2.10–16 • 120
1.213–217 • 84 2.32 • 105
1.213–268 • 84 2.145–146ab • 105
1.213–338 • 9 2.145–279 • 20
1.214 • 70, 86, 96, 112 3.25ff. • 120
1.214–215 • 86, 117 3.94 • 87
1.214–223 • 84 4.2 • 117
1.215 • 43, 55, 64, 86, 88, 96, 112 4.48 • 111
1.215–216 • 21, 88 4.48–51 • 55
1.216 • 87, 88, 89, 90, 91, 92, 94, 95, 98, 4.48–52 • 117
99, 106, 108, 109, 110, 116 4.50 • 112
1.216–217 • 100 4.92 • 98
1.217 • 20, 21, 100, 105, 108, 109 4.101ab • 104
1.218 • 95, 101 4.106–108 • 55
1.218–219 • 75 4.107 • 113
1.224 • 120, 122 4.108 • 86
1.225 • 122 4.210 • 87
1.226–238 • 122 PVin
1.227cd • 61 2 65,1–9 • 75
1.229cd • 59 2 66,4–7 • 75
1.239–247 • 123 2 70,6–72,8 • 11
1.242 • 123 2 72,10–11 • 11, 19, 61, 63
1.247–268 • 122 2 92,4–6 • 62
1.269–283ab • 122 2 92,7–12 • 63
1.292–311 • 122 2.1–7 • 59
1.310cd–311 • 14 2.35 • 39
1.312 • 123 2.35–50 • 12
1.312–340 • 121, 148 2.36 • 40
1.314c • 87 2.37a • 44
1.316a • 87 2.37bd • 44
1.317–329 • 11 2.38 • 53
1.318 • 15, 127 2.38ab • 49
1.319–320ab • 124, 127 2.38cd • 53
1.319–324 • 135 2.39 • 12, 53
1.319cd • 140 2.40 • 53
1.323 • 140 2.41 • 54
1.327 • 15 2.42 • 55, 111
170 INDICES
2.43 • 12 39,30 • 96
2.43–44 • 55 84,5–6 • 68
2.44 • 12 101,23 • 75
2.44cd • 55 101,23–102,8 • 75
2.45 • 55, 56 102,2–8 • 76
2.46 • 56 107,14–17 • 75
2.47ab • 57 107,14–25 • 75
2.47cd • 57 107,14–109,22 • 10
2.48 • 59 107,14–112,6 • 10
2.49ab • 59 107,14–176,4 • 9
2.49c • 60 107,14–176,16 • 10
2.49d • 61 107,17–19 • 75
2.50 • 61 107,19–108,6 • 64
2.65d • 62 107,22–25 • 75
2.66 • 63 108,1 • 98, 115
3 31,11 • 86 108,1–2 • 84, 89, 90, 91, 92, 97, 98
3 128,4 • 111 108,1–109,22 • 106
3.9 • 111 108,2–3 • 64
PVinṬms 108,2–5 • 43
71a4–b3 = PVinṬTib Dze 227b4–228a2 • 108,2–6 • 85, 116
12 108,2–109,19 • 106
71b6 • 44 108,5–6 • 44, 110
72a3 • 44 108,6–16 • 70
72b1–2 = PVinṬTib Dze 228b6–229a1 • 108,15–16 • 85
53 108,16–109,4 • 64
73a1–6 = PVinṬTib Dze 229a6–b4 • 55 108,16–109,11 • 21
73a6–b3 = PVinṬTib Dze 229b4–230a1 • 108,24 • 86
55 109,1 • 86, 111, 112
73b1–2 = PVinṬTib Dze 229b6–7 • 55 109,1–3 • 55, 69, 77, 112
74a1 • 56 109,3–4 • 87
74b1 • 58 109,4 • 87
74b3–5 = PVinṬTib Dze 230b8–231a1 • 109,7–9 • 96
58 109,8 • 87
PVinṬTib 109,8–11 • 99
Dze 228b6–229a1 • 53 109,11 • 110
Dze 229a6–b4 • 55 109,11–19 • 20, 21
Dze 229b4–230a1 • 55 109,15–19 • 100
Dze 229b6–7 • 55 109,19 • 108
Dze 231a1–b3 • 62 109,19–20 • 110, 116
Dze 230b8–231a1 • 58 109,20–21 • 110
PVSV 109,20–22 • 101
10,15–17 • 62 109,23–110,15 • 10
10,19–23 • 63 109,24–110,15 • 76
24,16–93,5 • 9 110,15–111,11 • 10
37,27 • 93 111,11–112,5 • 10
39,6–8 • 39 112,6 • 61
INDICES 171
112,16–27 • 14 393,13–17 • 112

113,8–120,7 • 10 393,19 • 113
113,14–23 • 61 393,25–28 • 99
113,25–114,3 • 59 393,30–394,8 • 110
114,25–27 • 59 394,9–10 • 96
118,14–17 • 59 394,20–22 • 116
120,2–6 • 59 583,16–17 • 13
120,8–126,15 • 10 584,21 • 17
123,14–124,23 • 122 584,22–23 • 17
123,14–124,28 • 11 584,23–24 • 17
126,24–134,25 • 10 584,24–25 • 17
141,14–150,5 • 11 585,8 • 16
152,1–4 • 61 585,11–13 • 17
153,7 • 87 585,14–15 • 17
155,14 • 87 585,21 • 16
155,18–164,24 • 11 585,27–28 • 16
157,9–18 • 78 585,28–30 • 16
157,10–18 • 11 586,10 • 16
157,29–162,11 • 11 586,13–14 • 17
164,9–24 • 14 586,17–18 • 17
164,24–25 • 17 586,18–19 • 16
167,11–14 • 15 586,19–20 • 17
167,23–168,2 • 104 586,21–22 • 16
167,26–168,1 • 117 586,22–23 • 17
168,1 • 108 586,23–24 • 17
168,1–2 • 109, 110, 116 589,16 • 15
168,3–4 • 16 589,31–590,12 • 15
168,15–16 • 134 590,16–17 • 16
169,4–13 • 16 592,7 • 16
170,5–11 • 143 592,7–13 • 16
172,15–24 • 15 593,27 • 16
173,14–175,10 • 11 598,15 • 16
173,16–17 • 19 598,21 • 16
173,16–174,1 • 103 605,24–25 • 16
173,22–25 • 116 608,15–16 • 18
173,26–174,6 • 20 608,17 • 18
174,1 • 117 610,12 • 16, 19
174,21–23 • 113 610,15–16 • 19
174,22–23 • 86 612,7 • 113
175,3–4 • 87 PVṬ
176,6–12 • 9 Je D242b3–5/P285b6–286a1 • 98
PVSVṬ Je D242b5–6/P286a2–3 • 116
174,21 • 96 Je D245a7–b1/P289b5–7 • 99
390,15–19 • 98 Je D245b2/P289b8–290a1 • 110
390,21–22 • 116 Ñe D45b1/P51b6 • 17
390,30–391,9 • 85 Ñe D45b1–2/P51b6–7 • 17
172 INDICES
Ñe D45b2/P51b7–8 • 17 ŚāṅkhŚSū
Ñe D45b2–3/P51b8–52a1 • 17 16.5.4 • 62
Ñe D45b5/P52a3–4 • 16 ŚB
Ñe D45b6–7/P52a5–6 • 17 3.4.1.22 • 50
Ñe D46a1/P52a8 • 17 4.4.3.19 • 143
Ñe D46a4/P52b3–4 • 16 11.5.1 • 50, 126
Ñe D46a7/P52b8 • 16 13.2.6.9 • 135
Ñe D46b1/P52b8–53a1 • 16 13.2.6.12 • 62
Ñe D46b4/P53a5 • 16 ŚBh
Ñe D46b5/P53a6–7 • 17 I.63,5–66,2 • 45
Ñe D46b6–7/P53b1–2 • 17 I.110,5 • 40
Ñe D46b7–47a1/P53b2–3 • 16 I.121,7–10 • 126
Ñe D47a1–2/P53b3 • 17 I.123,7–124,5 • 126
Ñe D47a2/P53b4 • 17 II.3,1–3 • 128
Ñe D47a2–3/P53b4–5 • 16 II.3,5–6 • 129
Ñe D47a3/P53b5–6 • 17 II.3,7–8 • 129
Ñe D47a6/P54a1–2 • 15 II.4,4 • 125, 129
Ñe D49a4/P56b2 • 15 II.5,4–5 • 129
Ñe D49b5/P57a6 • 15 II.26,1–27,4 • 130
Ñe D50a1–2/P57b3–4 • 16 II.26,11–12 • 130
Ñe D51a6/P59a8–b1 • 16 II.27,1–4 • 130
Ñe D51a6–b1/P59b1–4 • 16 II.28,2–3 • 129
Ñe D52b2–3/P61a5–6 • 16 II.28,6–7 • 131
Ñe D56a3/P66a2–3 • 16 II.28,9–12 • 129
Ñe D61a1/P72b2 • 16 II.49,1–2 • 132
Ñe D62b2/P74b4–5 • 18 II.57,5 • 135
Ñe D62b2/P74b5 • 18 II.57,12–15 • 132
Ñe D62b7/P75a5 • 18 II.64,6–12 • 133
Ñe D63b7/P76b2 • 16 II.65,7–8 • 133
Ñe D64a2/P76b5–6 • 19 II.66,1–67,1 • 133
Ñe D66b7–67a5/P80b2–81a1 • 18 II.73,1–74,2 • 142
PVV II.139,2 • 145
409,8 • 18 II.139,2–140,2 • 140
410,18–411,1 • 55 II.141,4 • 141
411,4 • 18 II.141,5 • 141
ṚV II. 141,5–142,1 • 141
1.89.10 • 132, 133 II.142,1 • 143
1.169.3 • 132 II.183,2–5 • 144
4.58.3 • 132 II.232,3–5 • 137
6.16.34 • 136 II.232,5 • 136
8.77.4 • 132 II.274,5–277,1 • 139
10.95 • 50, 126 II.280,1–281,1 • 139
10.106.6 • 132, 134 II.285,1 • 40
ṚVBh II.285,4–5 • 138
4.695 • 134 II.285,12–13 • 138
II.286,1–2 • 138
INDICES 173
II.315,5 • 134 Tāṇḍya Brāhmaṇa

II.315,5–321,2 • 134 19.7.2 • 139
II.315,6 • 134 TSK/Ś
IV.33,2–65,5 • 139 171–172 • 67
IV.322,3–378,3 • 129 171–173 • 66
V.72,6–7 • 49, 126 2774/2773 • 41
V.173,13–184,4 • 139 2775/2774 • 19, 65
V.176,4 • 140 3157/3156 • 119
V.177,9–16 • 140 3527/3526 • 40
SK TSPK/Ś
17 • 68 735,19–736,3/891,20–892,12 • 41
SS 735,22–23/891,23–892,4 • 42
23,11–14 • 120 736,3–18/892,13–893,10 • 19
ŚV 736,4–18/892,13–893,10 • 65
ā tmavā da 76 • 68 877,24–25/1062,21–22 • 70
codanā 44ab • 87 TUS
codanā 121 • 19 148,12 • 40
codanā 121–122ab • 62, 114 TV
codanā 121–132 • 114 II.2,7–14 • 134
codanā 130 • 115 II.28,13–14 • 131
codanā 131 • 115 II.66,2–3 • 133
codanā 132 • 119 II.66,10–14 • 133
codanā 137 • 119 II.66,18–67,9 • 134
codanā 143–151 • 143 II.67,25–27 • 134
codanā 144ab • 26, 125 II.75,16–23 • 142
pratyakṣasūtra 17–37 • 123 II.140,7–10 • 140
sambandhākṣepaparihāra 14cd • 148 II.141,14–15 • 142
TaitB II.142,16–19 • 144
2.1.2.10 • 129 II.143,16–19 • 144
2.1.9.2 • 129 II.144,3 • 144
2.7.3.1 • 135 II.144,4–5 • 144
TaitS II.144,7–8 • 145
1.1.11 • 135 II.144,10–11 • 145
1.2.1 • 132 II.146,3 • 145
1.3.7.1 • 50, 126 II.146,12 • 145
1.5.1 • 128 II.148,11–12 • 145
1.5.1.1–2 • 130 II.149,2–154,4 • 133
1.5.9.1 • 40, 137 II.150,22 • 144
2.1.1 • 128, 130 II.200,3–6 • 141
2.4.6.1 • 138 II.200,3–202,14 • 141
2.5.5.6 • 129 II.201,18–21 • 141
2.6.5.3 • 135 II.217,13–218,4 • 142
5.3.12.2 • 143 II.219,16–21 • 142
6.1.6.7 • 139 II.231,15–19 • 136
6.3.5.2–3 • 50, 126 II.232,18–19 • 136
7.4.18.2 • 62 II.233,8–9 • 136
174 INDICES
II.233,10–13 • 136 Hetuprakaraṇa • 8

II.233,21–25 • 136 Kalpasūtra • 141, 142
II.233,23–234,1 • 136 Kaṭhāraṇyaka • 146
II.271,5–8 • 139 Mahābhārata • 50
II.287,23–288,11 • 138 Mahāyānasūtra • 46
II.315,21 • 135 Mīmāṃsāsūtra • 47
VādhŚSū Nigama • 133, 141
5.2.1.4 • 50 Nirukta • 132, 133, 141
VaikhŚSū Pramāṇasamuccaya • 7, 62
Pramāṇasiddhi • 101, 105, 117
8.5 • 50
Pramāṇavārttika • 12, 119
10.10 • 50
Pramāṇaviniścaya • 12
VājS
Purāṇa • 10, 50
5.2c • 50
Ratnāvalī • 111
Vibh Sā barabhā ṣ ya • 140
409 n. 8 • 19 Slokavā rttika • 120
VPR Smṛti • 10
1.30–32 • 35 Srautasū tra • 62
1.34 • 35 Sūtra • 70
1.150 • 142 Tantravārttika • 127, 140
1.151ab • 147 Tarkapāda • 128
1.152 • 147 Vādavidhi • 7
1.153ab • 147 Vākyapadīya • 146
1.175–183 • 144 Vārttika • 12
2.303 • 146 Vārttikakārikā • 12
VPV Veda • 12, 18, 19, 29, 33, 34, 38, 39, 40,
89,2–6 • 36 42, 44, 45, 46, 49, 50, 56, 70, 71, 72,
209,1 • 147 77, 83, 102, 103, 105, 113, 115, 117,
VSū 120, 121, 122, 123, 124, 125, 126,
1.1.5 • 66 127, 128, 129, 130, 134, 135, 136,
1.1.14 • 65, 66 137, 140, 141, 142, 143, 144, 145,
3.2.4 • 66 146, 147, 148
10.1 • 65
Vedāṅga • 121
YD Vikramorvaśīya • 50
101,7 • 87
Vyākaraṇa • 60, 133, 141, 142
Yājñavalkyasmṛti • 47
70,15 • 87
Yajurveda • 47
99,12 • 87

Place names
Sanskrit texts and genres
Aryā varta • 145
Atharvavedaprātiśākhya • 129
Nālandā • 7, 8
Ayurveda • 18, 105
Pāṭaliputra • 28, 56
Bodhisattvabhūmi • 83
Telāḍhaka • 8
Brāhmaṇa • 130

Bṛhaṭṭīkā • 119, 120
Dharmaśāstra • 127, 141, 142
INDICES 175
Sanskrit and other names 131, 132, 133, 134, 135, 136, 138,
139, 140, 141, 143, 145, 146, 147,
Acā rya Dharmakı̄rti • 12 148
Aryadeva • 87, 88 Sabarasvā min • 15, 19, 33, 40, 45
Bhartṛhari • 35, 142, 146, 147 Sā kya mchog ldan • 111
Bhavadāsa • 19 Śākyabuddhi • 2, 18, 38, 42, 47, 50, 67,
Bhāviveka • 118, 119, 149 68, 72, 74, 77, 78, 95, 96, 98, 99, 110,
Dharmapāla • 88 112, 113, 114, 115, 116
Dharmottara • 3, 44, 53, 55, 58 Sā ntarakṣ ita • 66, 67, 86, 103, 149
Dignāga • 7, 8, 20, 29, 63, 83, 84, 85, Socrates • 149
88, 89, 90, 91, 92, 93, 94, 95, 96, 97, Subhagupta • 149
98, 99, 100, 101, 102, 104, 105, 106, Sucaritamiśra • 19, 115
107, 109, 110, 114, 115, 116, 118 Tāranātha • 121
Acā rya Dignā ga • 62, 96, 98, 99 Tsoṅ kha pa • 111
Jaimini • 15, 17, 23, 24, 27, 28, 33, 34, Udayana • 36
38, 39, 45, 47, 131 Uddyotakara • 36, 68, 87
Jayantabhaṭṭa • 19 Umbeka • 115
Jinendrabuddhi • 92, 94, 99 Vācaspati • 36
Jñānacandra • 8 Vasubandhu • 11, 68, 70, 83, 86
Kālidāsa • 50 Acā rya Vasubandhu • 77
Kamalaśīla • 86, 103, 113, 149 Vasudhararakṣita • 92, 93
Kanakavarman • 92 Vibhūticandra • 19
Karṇakagomin • 2, 18, 38, 42, 47, 50, Wuxing • 8
66, 67, 68, 77, 78, 96, 98, 99, 110, Yājñavalkya • 47
112, 115, 116 Yāska • 50, 129, 132, 135, 142
Kautsa • 129 Yaśomitra • 45
Kullūkabhaṭṭa • 133 Yijing • 8
Kumārila • 7, 10, 19, 60, 62, 114, 115,
117, 118, 119, 120, 122, 125, 131,
133, 134, 136, 138, 140, 141, 142,
Gods and saints
143, 144, 145, 146, 148
Mādhava • 46 Aditi • 132, 133, 145
Sāṅkhyanāśaka • 46 Agni • 50, 128, 129, 135, 136, 137, 138
Manorathanandin • 2, 18, 56 Apsaras • 49, 52
Pakṣilasvā min • 36, 86, 105, 106, 110 Arjuna • 50
Vātsyāyana • 105, 106 Asura • 128
Pāṇini • 45, 60, 141 Ayus • 50
Pārthasārathimiśra • 19, 115 Bodhisattva • 13, 73
Patañjali • 141 Buddha • 83, 105, 106, 107, 109, 110,
Piṅgala • 45 114, 115, 117, 119, 120
Plato • 149 Cakravāka • 133
Prajñākaramati • 66, 67 Dyaus • 133
Praśastapāda • 66 Indra • 49, 50, 132, 136
Protagoras • 149 Jina • 40
Ratnakīrti • 120 Kapila • 113
Sabara • 19, 39, 50, 125, 128, 129, 130, Mahāvīra • 83
176 INDICES
Manu • 109 Saugata • 42

Prajāpati • 128, 130, 135, 138 Sautrāntika • 68
Purūravas • 27, 49, 50, 126 Vaibhāṣika • 60
Ṛṣyaśṛṅga • 50 Vaiśeṣika • 7, 20, 65, 66, 67, 69, 74
Rudra • 128, 130 Vaiyākaraṇa • 10, 60, 74, 144
Soma • 27, 50, 126, 132 Vyākaraṇa • 60, 133, 141, 142
Sūrya • 129 Yājñika • 142, 145
Tathāgata • 118 Yogācāra • 68, 83, 118
Urvaśī • 27, 49, 50, 126
Vāyu • 130
Viśve Devāḥ • 133
Vṛtra • 135, 136
Sanskrit words

akṣaṇika • 68
akṣi • 44
Schools and religious groups agati • 87, 88, 89, 91, 101, 104, 106,
Ahū raka • 47 107, 108, 109, 110, 116
Ahvaraka • 47 agamaka • 14
Bauddha • 17, 34, 42 agamya • 36, 89, 116
Bauddhasiddhānta • 64 aguṇavacana • 134
Jaiminīya • 16, 17, 23, 33 agṛhītagrāhijñāna • 120
dPyod pa pa • 18
agṛhītaśabdārtha • 141
Jarajjaiminīya • 19
Mahāyāna • 46 agṛhītārthagrahaṇa • 138
Mīmāṃsā • 1, 3, 7, 9, 10, 11, 15, 23, 45, agni • 50, 55, 61, 62, 103, 115, 129,
50, 57, 119, 120, 121, 122, 123, 124, 138
127, 128, 129, 131, 134, 135, 137, agnimanthana • 50
138, 139, 146, 147, 148, 149
agnihotra • 15, 16, 24, 27, 40, 44, 50,
Mīmāṃsaka • 9, 10, 11, 13, 14, 15, 16,
17, 18, 19, 23, 24, 25, 26, 27, 33, 34, 51, 52, 53, 54, 62, 69, 112, 113, 126,
38, 40, 42, 43, 44, 48, 49, 50, 51, 53, 137, 138, 148
54, 57, 60, 74, 102, 109, 115, 119, aṅga • 121, 131
120, 122, 123, 124, 125, 126, 127, aṅgabhūta • 70
131, 134, 135, 137, 139, 142, 145, aṅgāṅgībhāva • 70
146, 147, 148, 149
aṅgīkṛtaveda • 40
Jaranmīmāṃsaka • 18, 19
Vṛddhamīmāṃsaka • 18, 19 acidrūpa • 66
Naiyāyika • 10, 62, 63, 65, 66, 67, 77, acetana • 67, 132
102, 103, 116 ajñātārthaprakāśa • 105, 120
Nyāya • 7, 18, 19 ajñāna • 14, 77
Pāñcarātra • 9
atidurlabha • 35
Pāśupata • 9
Sā bara • 19 atiduṣkara • 72
Saṃsāramocaka • 142 atiśaya • 14
Sāṅkhya • 7, 9, 20, 46, 69, 112, 113 atiśayavat • 14, 17
Sāṅkhyasiddhānta • 46 atiśayavatpuruṣapratikṣepa • 17
INDICES 177
atīndriya • 14, 17, 20, 38 anāgamāpekṣānumāna • 86

atīndriyadarśana • 13, 14, 34 anādara • 144
atīndriyadarśipuruṣa • 17 anāditā • 10, 144
atīndriyārtha • 34 anāntarīyaka • 101
atīndriyārthajña • 17 anitya • 128, 129
atīndriyārthadarśin • 17, 34 aniścaya • 76
atīndriyārthadarśipuruṣa • 17 aniścita • 36, 63
atyakṣa • 20, 55 aniṣṭa • 48
atyakṣaphala • 85 aniṣṭaphaladāna • 42
atyantaparokṣa • 36, 72, 87, 98, 99, anuguṇopāya • 70, 86, 96
107, 110, 112 anugṛhīta • 37
atyantaprasiddha • 20, 64 anugrahin • 138
atyantāsambhava • 72 anupadeśa • 112
adarśana • 14, 85, 140 anuparodha • 100, 101
adarśanamātra • 63 anupalabdhi • 14, 63, 75
aduṣṭakāraṇārabdha • 120 anupalabdhimātra • 75
adṛśya • 86 anupalambha • 51
adṛṣṭa • 20, 35 anupālana • 41, 42
adṛṣṭataṇḍulapāka • 103 anupālayat • 42
adṛṣṭavyabhicāra • 64, 89, 96 anumātavya • 18
adṛṣṭārtha • 18, 93 anumātṛ • 35
adharma • 65, 66, 69, 112, 113 anumāna • 13, 14, 18, 35, 36, 38, 55,
adhas • 42 62, 63, 64, 69, 70, 71, 75, 83, 86, 89,
adhikṛta • 70 91, 95, 96, 97, 100, 101, 102, 103,
adhigataparamārtha • 75 105, 107, 108, 110, 111, 112, 113,
adhigamahetu • 94, 99 116
adhigamopāya • 133 anumānatā • 84, 88, 89, 98, 99, 107,
adhi√ṣṭhā • 13 116
adhiṣṭhātṛ • 66 anumānatva • 84, 89, 93, 94, 95, 98,
adhiṣṭhāna • 14, 65, 67, 68 99, 101, 108, 116
adhyayana • 10 anumānabhāva • 96
adhvaryu • 50 anumānamātra • 36
anadhigatārthaviṣaya • 120 anumānāgamya • 87, 89, 96, 116
ananumāna • 116 anumita • 35
anapāya • 101, 102 anumīyate • 66
anabhiprāya • 61 anumeya • 112
anabhiyukta • 142 anuśaṃsāpāya • 85
anavabudhyamāna • 110 anusandhi • 70
anaśana • 78 anusāra • 41
178 INDICES
anṛta • 105, 129 aprasiddha • 51, 53, 54

anṛtavāda • 131 aprākṛtagamya • 36
anṛtavādin • 125 aprāmāṇya • 72
anekārtha • 56 apsaras • 49, 52
anekārthatva • 42 abādhana • 112
anekārthābhidhāna • 145 abādhita • 86
antaraṅgabhūta • 131 abhidhāna • 65, 71, 100
antarikṣa • 133 abhidhāyaka • 70
antargaḍu • 39 abhidheya • 58
antargama • 103 abhinirvartayitum • 132
andhaparamparā • 142 abhiprasanna • 40, 47
anyathāracana • 46 abhiprāya • 15, 57, 75, 84, 89, 116
anyāya • 145 abhiprāyakathana • 45
anvaya • 111 abhibhava • 45
anvākarṣati • 38 abhimata • 16, 17, 34, 48, 58, 64
anvākhyāna • 130 abhiyukta • 134, 141, 142, 145
anvīkṣā • 36 abhiyuktatara • 35, 145
apabhraṃśa • 144 abhiyoga • 100, 105
aparijñātārtha • 39 abhivadanti • 19
apavarga • 70 abhivyakti • 74
apahṛtabuddhi • 36 abhihita • 88, 89, 99, 109, 110, 116
apāṇa • 66 abhyaśnute • 50
apāya • 85 abhyāsa • 34
apuruṣakriyā • 37 abhyudaya • 70
apuruṣaracitatva • 12 abhyupagata • 107
apūrva • 19 abhyupagama • 17, 113, 134
apūrvārthajñāna • 120 abhyupagamya • 12, 112
apūrvārthavijñāna • 120 abhyupāya • 105
apoha • 9, 78, 84, 115, 116 amyak • 132
apauruṣeya • 12, 19, 103 ayuktābhidhāyitva • 77
apauruṣeyatva • 7, 23, 61, 122 ayuktiviṣaya • 104
apauruṣeyatvavādin • 61 araṇi • 49, 50
aprakāśita • 132 aruṇā • 139
apratipatti • 103 arodīt • 128
apratiṣṭhāna • 55 arthakārya • 74
apratyakṣa • 36, 58 arthakriyāvirodha • 68
apratyaya • 104 arthakriyāsamartha • 68
apramāṇīkṛtya • 72 arthagati • 17
aprameya • 77 arthaniyama • 58
INDICES 179
arthaniyamapratīti • 58 aviruddha • 78
arthanirṇaya • 141 avirodha • 14, 51, 136
arthaparīkṣaṇa • 36 avirodhin • 14, 147
arthapratipādana • 141 aviśiṣṭa • 132, 135, 144
arthamātra • 93 aviṣayīkaraṇa • 76
arthavattva • 132, 134 avisaṃvāda • 20, 62, 84, 87, 88, 89, 93,
arthavāda • 128, 129, 130, 131, 133, 94, 96, 97, 99, 100, 104, 107, 110,
134 114, 116, 117
arthavikalpamātra • 93 avisaṃvādana • 18, 20, 43, 61, 63, 64,
arthavipratiṣedha • 132 73
arthavibhāga • 56 avisaṃvādabuddhi • 99
arthasiddhi • 75, 84 avisaṃvādasambhava • 103, 110, 117
arthasmāraka • 132 avisaṃvādādhigamahetu • 94, 99
arthāpabhraṃśa • 144 avisaṃvādijñāna • 105
arthābhidhāna • 12 avisaṃvādin • 43, 63, 87, 98
arthāvisaṃvāda • 89, 93, 94, 99, 110 avaiguṇyena • 49
arthaikatva • 136 avaiparītya • 100, 101
arthotprekṣā • 134 avyabhicārāsiddhi • 63
arvāgdarśana • 16, 36, 58 avyavacchinna • 35
arvāgdarśin • 16, 70, 75 √aś • 50
alobha • 70 aśucisvabhāva • 77
avakāśa • 65 aśubha • 66, 67
avadhāraṇa • 141 aśubhakarman • 67
avadhi • 44, 45, 144 aśvaghāsa • 138
avabodha • 110 aṣṭakā • 142
avalepa • 77, 124 asaṅkhya • 42
avasthādeśakāla • 35 asañjñā • 60
avasthāna • 36 asandeha • 141
avasthita • 141 asamartha • 14
avicchinna • 141 asambhava • 14, 17, 68, 72, 87, 110,
aviccheda • 59 117
avijñāna • 133 asammūḍha • 36
avijñeya • 132 asarvajña • 115
avitatha • 103 asākṣātkṛtadharman • 142
avitathābhidhāyin • 75 asādhana • 103
aviditavedana • 138 asādhāraṇaśaktiyukta • 17
avidyā • 70 asādhukarṣāpaṇa • 144
aviparīta • 75 asādhutvalakṣaṇa • 142
avibhāga • 136, 137 asiddhatā • 115
180 INDICES
asiddhārtha • 53 ānarthakya • 128

asiddhi • 14, 63, 104, 117 ānupūrvī • 122
askhalita • 104 āntaravikāra • 66
ahita • 66 āpta • 10, 18, 29, 94, 105
āptatva • 9
ākṣepa • 129 āptaprayukta • 93
ākhyātṛ • 15 āptaprāmāṇya • 18
āgama • 3, 9, 29, 35, 36, 42, 44, 75, 83, āptavacana • 89, 93, 94, 95, 98, 99
84, 85, 86, 89, 91, 92, 96, 97, 98, 100, āptavāda • 10, 84, 87, 88, 89, 91, 93,
101, 102, 104, 105, 106, 107, 108, 96, 98, 99, 105
109, 110, 113, 114, 115, 116, 117, āptavādāvisaṃvāda • 84, 89, 94, 107
118 āptasambhāvanā • 47
āgamagamya • 35 āptāgama • 83, 86
āgamacakṣus • 36 āptopadeśa • 18
āgamatva • 75 āptopadeśaprāmāṇya • 118
āgamapūrvaka • 35 āmnāya • 128
āgamaprāmāṇya • 9, 64, 85 āmnāyāviccheda • 59
āgamalakṣaṇa • 19, 43, 61, 63, 103, āyurveda • 18
104, 109, 116 āyurvedaprāmāṇya • 18
āgamaviṣaya • 112 āyus • 50
āgamasañjñita • 51 ārambha • 105
āgamasahita • 35 ārya • 144
āgamasiddha • 55 āryagamya • 144
āgamānapekṣānumāna • 64 āryasatya • 100
āgamāpekṣa • 111, 112 āryāvartanivāsin • 145
āgamāpekṣānumāna • 55, 64, 70, 86, ārṣa • 133
111, 112, 113 ālabheta • 130
āgamārtha • 42 ālambana • 38
āgamārthaniścaya • 42 āvaraṇa • 72
āgamārthāśraya • 55, 111 āśaṅkākāraṇa • 47
āgamāśraya • 55, 86, 111, 113 āśaṅkyamāna • 44
āgamāśrita • 36, 86 āśaṅkyeta • 63
āgamāśritānumāna • 107, 111, 113 āsitum • 64, 85
āgnīdhra • 50
ācārya • 12, 62, 85, 103, 116 icchā • 12, 15, 16, 54, 65, 66, 124
ātman • 20, 39, 65, 66, 67, 77, 112, 117 icchāmātra • 73
ātmaliṅga • 66 icchāmātrapratibaddha • 73
ātmasamavāya • 65 iḍā • 50
ādhāra • 67 indriya • 14, 34, 66
INDICES 181
indriyagrāhya • 142 uru • 50

indriyādisambandha • 114, 115 urvaśī • 26, 27, 49, 50, 51, 52, 126, 148
indriyārthasambandha • 19, 114 uṣṇa • 115
iṣṭa • 48 uśanti • 42
iṣṭāniṣṭaphaladāna • 42
iṣṭārtha • 58 ūnatva • 131
ūru • 50
īkṣitāraka • 40
īśvara • 20, 77, 112, 117, 120 ṛṣi • 35, 133, 142
ukti • 14, 105 ekadeśa • 18, 103, 129

uttāna • 137 ekadeśasaṃvāda • 114
udāhṛta • 63 ekadeśāvisaṃvāda • 11, 62
udbhāvayatā • 103 ekadeśāvisaṃvādana • 18, 20, 43, 61,
udbhid • 139 63, 64, 73
udbhidvat • 139 ekadeśāvisaṃvādirūpa • 63
unmeṣa • 66 ekavākya • 130
upakārin • 12 ekavākyatā • 131
upakṣepa • 56 ekavṛttiniyamakāraṇa • 56
upagama • 100 ekasthālyantargama • 103
upa√diś • 47 ekārtha • 33, 48, 74
upadiṣṭa • 100 ekārthaniyama • 57
upadeśa • 12, 18, 39, 42, 45, 86, 110, ekārthapratiniyama • 44
142 ekārthopasaṃhāra • 70
upadeśanirapekṣa • 39, 58 aiśvarya • 71
upadeṣṭṛ • 45, 147 odana • 145
upadhārayet • 45 oṣadhi • 143
upapatti • 134 kathana • 45, 57, 133
upapadyate • 35 karaṇa • 59
upabhoktṛ • 67 karambhapātra • 143
upayogin • 42, 100 karuṇāsātmībhūta • 73
uparodha • 61 karṇaprāvaraṇa • 136
upalakṣaṇa • 77 kartṛ • 11, 66, 67, 68, 77
upalabhmahe • 136 kartṛtva • 66, 67, 68, 71
upavarṇita • 101, 108, 110, 116 kartṛvāda • 77, 124
upasaṃhāra • 70, 74 karmakartṛ • 67
upahata • 19, 61 karmakleśa • 41
upādeya • 41, 100, 105 karman • 19, 66, 67, 68, 69, 77, 112,
upāya • 70, 96, 100, 133 113, 116
182 INDICES
karmaphala • 67, 68 gaḍu • 39, 40

karṣāpaṇa • 144 gaḍupraveśa • 40
kalpanāyāsamātra • 39 gati • 14, 17
kalpayitavya • 133 gamaka • 73, 74
kalpita • 33 gamika • 63
kalpya • 123 gamya • 35, 36
kākatālīya • 73 garīyāṃs • 71
kāṇa • 71 garbha • 77
kāṇukā • 132 gava • 136
kāmyeṣṭi • 128 guṇa • 45, 66, 71, 77, 134, 137, 139
kāyavākkarman • 68, 77 guṇamātra • 136
kāyavāgvyavahāra • 76 guṇavāda • 134
kāraṇaśuddhatva • 87 guṇavidhi • 40
kāruṇika • 40 guṇavaikalya • 71
kāryakāraṇabhāvāsambhava • 68 guṇādhāra • 67
kāryahetu • 98 guṇin • 66
kāla • 35, 42 go • 44, 45, 137
kālabheda • 36 gocara • 65
kāṣṭhahāraka • 71 gauṇa • 144
kila • 61 grantha • 133
kubjapṛṣṭha • 39 grāmya • 71
kuśala • 35 grīṣma • 36
kusumavinda auddālaki • 50
kūṭasthanitya • 10 ghāṭāmastaka • 39
kūpa • 36 ghṛta • 50
kṛtāvadhi • 44, 45 ghṛtādiprakṣepa • 44
kṛtrima • 59
kṛtsna • 129 cakṣus • 36, 60, 71, 147
kramayaugapadya • 69 catuḥsatyaprakāśana • 105
kriyā • 66 caru • 145
kriyārtha • 128 cikīrṣā • 68
kleśa • 41 citrā • 138, 139
kṣaṇa • 75 cintā • 70, 87, 112
kṣaṇika • 98, 115 cetana • 66, 67
kṣaṇikatva • 115 cetanā • 42, 57
kṣaṇikādivākya • 115 caitasika • 68
caitta • 68
khādet • 15 codanā • 52, 136
khyāpana • 37 codyaparihāra • 70
INDICES 183
jaḍarūpa • 67 tīrthika • 77, 121

jana • 53 turpharītū • 132, 134
janítra • 50 tulyarasa • 103
jambu • 140 tulyarūpa • 103
jarat • 71 tṛtīyasthāna • 43
jarbharī • 132, 134 tailaghaṭa • 71
jala • 36 trairūpya • 55, 111
jalasaṃśleṣa • 77
jalpa • 149 dakṣinā • 129, 136
jāḍya • 77, 124 dayālu • 105
jāti • 134, 135 darśana • 14, 46, 58, 74
jātivādāvalepa • 77, 124 darśapūrṇamāsa • 129, 135, 139
jīvana • 66 dahanadravya • 54
juhuyāt • 15, 24, 27, 40, 44, 53, 126, dānahiṃsādicetanā • 42
137, 138, 148 dāraka • 71
°jña • 17 dāridrya • 71
jñāna • 34, 35, 41, 45, 63, 65, 66, 67, dāhādisamartha • 44
68, 87, 105, 120 diṅmātra • 77
jñānaprāmāṇyakāraṇa • 87 diś • 44
jñānamātra • 115 dīrghaśūka • 140
jñeyatva • 120 duḥkha • 41, 65, 66, 68
jyotis • 129 duḥkhasaṃvitti • 67
duḥkhasaṃvid • 68
tajjanya • 73 duḥkhasatya • 64
taṇḍulapāka • 103 durjñāna • 36
tattva • 16, 34, 77, 100, 105, 142 duṣṭa • 119
tattvajñāna • 41 dṛśyādṛśya • 86
tattvatas • 145 dṛṣṭa • 72, 103
tatpuruṣa • 138 dṛṣṭavyabhicāra • 71, 72
tatsvabhāvatā • 73 dṛṣṭānta • 62
tadutpatti • 59, 73 dṛṣṭārtha • 18
tarka • 35, 141, 147 devatā • 133
tāthāgata • 118 deśa • 35, 140
tādātmya • 59, 73 deśakālasvabhāvaviprakṛṣṭa • 42
tādrūpya • 73 deśanā • 52, 119
tāya • 105 daiśika • 44
tīrthakara • 56 doṣa • 119
tīrthasnāna • 69 doṣopahata • 19, 61
tīrthasnānavāda • 77 dyotaka • 61
184 INDICES
dravya • 54, 140 niyāmaka • 56

dravyatas • 83 ni√yuj • 60
dravyalakṣaṇa • 66 nirākṛta • 12, 14, 77
draṣṭṛ • 18 nirāśaṃsa • 69
draṣṭṛpravaktṛsāmānya • 18 nirūpaṇa • 142
dvādaśa • 62 nirṇaya • 33, 141
dveṣa • 65, 66, 77, 78 nirdiṣṭa • 96
nirdhārya • 142
dharma • 19, 35, 68, 70, 71, 75, 77, 83, nirvartya • 19
128, 141, 142, 144, 145 nirvyāja • 56
dharmapatnī • 71 nirhetuka • 69
dharmavṛddhi • 70 nivartante • 141
dharmasiddhi • 142 nivedayati • 89
dharmāṅga • 145 nivṛtti • 9, 66, 75, 85
dharmādharma • 65, 66 niś√ci • 43
dharmecchā • 124 niścaya • 35, 41, 42, 44, 45, 104, 141
dhātu • 133 niścāyakapramāṇa • 46
dhvani • 73 niścita • 63, 120
dhvastaprajñāna • 124 niścinvanti • 42
niścetum • 12
nandanavana • 49, 50 niṣiddha • 12
naya • 118 niṣedha • 12, 77
nava • 45 niṣparyāyam • 75
navārtha • 44 niṣpramāṇaka • 54, 56
nānārtha • 48 niṣprayojana • 100
nāntarīyakatā • 75, 84 niṣphala • 39
nāntarīyakatva • 103 nisargatas • 60
niḥśreyasa • 70 nisargasiddha • 45
nitya • 10, 12, 60, 67, 71, 77, 129 nīti • 83, 100
nityatva • 12 nṛ • 53
nityaprakāra • 75 netra • 71
nidāna • 70, 78 nairātmyabhāvanā • 70
nibandhana • 132 nyāya • 16, 25, 36, 40, 42, 54, 103, 116,
nimitta • 73, 141, 144 141
nimeṣa • 66 nyāyavid • 116
niyata • 15, 60 nyāyānupālana • 42
niyama • 57, 58 nyāyānusāra • 41
niyamakāraṇa • 56 nyāyānusārijñāna • 78
niyamyamāna • 145
INDICES 185
pakṣa • 77 pāpa • 78
pakṣapāta • 47 pāpaśuddhi • 78
paksadharmatā • 111 pāpaśodhanasāmarthya • 113
patnī • 71, 72 pāpahāna • 124
pathin • 35 pāratantrya • 114
pada • 12, 135 pāṣaṇḍa • 10
padabhūta • 12 piṅgākṣī • 139
padarūpa • 12 piṇḍa • 39
padavākyārthanirṇaya • 141 piṇḍārtha • 70
padārtha • 70, 75 pitṛ • 77, 133
padārthasandeha • 141 pilu • 144
payas • 40, 137 pīḍana • 78
paramārtha • 75 puṃs • 14, 65, 71, 116
paramparā • 141, 142 puṃstva • 14, 17
paralokin • 67 puṃstvādisāmya • 14
parārtha • 105 putra • 133
parikalpanīya • 54 punarādheya • 128
parigraha • 52, 54, 60, 112 puruṣakṛta • 14
pariccheda • 87, 99 puruṣatva • 17, 34
parityajya • 19, 61, 103 puruṣatvādisāmya • 17
paripālana • 133 puruṣapravṛtti • 116
paribhāṣā • 60 puruṣaprāmāṇya • 48
parihāra • 66, 70 puruṣaviśeṣa • 13, 49
parīkṣaṇa • 36 puruṣātiśaya • 13, 14, 17, 34
parīkṣā • 37, 85, 142 puruṣātiśayapratikṣepa • 14, 17
parīkṣādhikṛta • 70 puruṣātiśayābhyupagamavāda • 17
parīkṣin • 144 puruṣārtha • 86, 96
parīkṣya • 85 puruṣārthābhidhāyaka • 70
parokṣa • 36, 38, 41, 54, 87, 88, 89, 92 puruṣārthāsambaddha • 76
paropadeśa • 45 puruṣārthopayogin • 42, 100
paropadeśamātra • 42 puruṣāśraya • 147
paryanuyoga • 53 puruṣopadeśa • 12
paryāyakathana • 133 pūrvaka • 35, 38, 57
paśu • 44 pūrvakarman • 67
paśukāma • 138 pūrvapakṣa • 91, 126, 140, 141, 142
paśupuroḍāśa • 50 pūrvapakṣin • 128, 129, 132, 134, 135,
pāka • 103 138, 139, 140, 141, 144
pātra • 143 pūrvāparavirodha • 20
pātrī • 49, 50 pūrvāparavirodhacintā • 70
186 INDICES
pṛṣṭha • 39 pratyakṣānumāna • 71, 113

pauruṣeya • 12, 41 pratyakṣānumānāgamya • 87, 89, 96,
pauruṣeyatva • 12 116
paurvāparyavirodha • 112 pratyāyayati • 45
prakaraṇa • 133 pratyāsattiviprakarṣa • 49
prakāśa • 105, 120 pradarśyate • 63
prakāśana • 60, 105, 132 pradeśa • 55
prakṛta • 112 pradhāna • 112
prakṛti • 117 pradhāna • 112, 117, 139
prakṣepa • 44 pradhānārtha • 20
prajñāna • 124 pradhānārthāvisaṃvāda • 100
praṇīta • 119 prapadyate • 36
pratikṣepa • 14, 17 prabhava • 69, 112, 113
pratikṣepasādhana • 14 pramā • 145
pratijñā • 13 pramāṇa • 9, 18, 23, 24, 25, 27, 29, 33,
pratiniyata • 15, 56 36, 37, 43, 44, 46, 51, 52, 53, 54, 55,
pratiniyama • 33, 44 56, 63, 70, 72, 78, 83, 84, 87, 91, 103,
pratipatti • 48, 89, 94, 96, 118 104, 105, 106, 107, 110, 114, 115,
pratipattimāndya • 57 117, 118, 120, 123, 138, 147
pratipadyate • 104, 118 pramāṇatā • 114
pratipadyemahi • 72 pramāṇatrayanivṛtti • 75
pratipanna • 46 pramāṇabala • 78
pratipādaka • 115 pramāṇabhūta • 105
pratipādana • 141 pramāṇabhūtapuruṣa • 15, 48
pratipādayitavya • 12 pramāṇasaṃvādin • 35
pratipādayiṣyati • 112 pramāṇānugṛhīta • 37
pratipādita • 14 pramāṇābha • 13, 14
pratiprāṇibhinna • 67 pramīyate • 140, 141
pratibaddha • 73, 145 prameyatva • 120
pratibandha • 59 prayatate • 36
pratirūpaka • 46 prayatna • 57, 65, 66, 67, 68
prativyūḍha • 14 prayatnavat • 66
pratiṣedhasiddhi • 64 prayuñjiran • 145
pratisamādhāna • 70 prayujyamāna • 135
pratīta • 44 prayujyate • 140
pratīti • 48, 49, 55, 58, 60, 93 prayoga • 139, 142
pratyakṣa • 38, 58, 63, 64, 75, 83, 86, prayogacodanā • 136
115 prayoganimitta • 144
pratyakṣāgamāśrita • 36, 86 prayogasamaveta • 132
INDICES 187
prayogāvadhi • 144 prāvaraṇa • 136

prayojana • 70, 132, 136, 137, 141 prāsādikatva • 71
pravaktṛ • 18 priyaṅgu • 140
pravartate • 55, prīti • 140
pravartante • 42 prītimat • 140
pravartana • 42 prekṣāpūrvakārin • 42
pravartamāna • 85
pravartayati • 112 phala • 39, 68, 85, 103
pravartitavya • 85, 110 phalada • 14
pravāda • 53, 74 phaladāna • 42
pravādamātra • 42 phalopabhoktṛ • 67
pravṛtta • 85, 110
babara prāvāhaṇi • 50
pravṛtti • 44, 51, 68, 101, 104, 108,
barhis • 130, 135
110, 116, 117
bahuvrīhi • 137, 138
pravṛttikāma • 43, 116
bahuvrīhigarbha • 53
pravṛttinivṛtti • 66, 85
bahuśāstrārthavedin • 145
pravṛttyaṅgabhūta • 70
bādhaka • 43, 51, 72
pravṛttyasambhava • 110
bādhate • 35
praśaṃsā • 134
bādhante • 17
prasādhana • 53
bādhavarjita • 120
prasiddha • 20, 35, 51, 55, 64
bādhā • 55
prasiddhatara • 134
bādhita • 55, 113, 114
prasiddhaviparyaya • 113
bāhulya • 53
prasiddhi • 16, 25, 35, 53, 100, 123,
bāhya • 93, 98
124, 138, 139, 145, 147
buddhaputra • 83
prastara • 135
buddhavacana • 84, 118
prahāṇa • 112
buddhavākya • 115
prahīṇasarvāvaraṇa • 72
buddhi • 14, 34, 36, 64, 65, 66, 68, 88,
prākṛta • 16, 36
89, 94, 99
prākṛtapuruṣa • 13
buddhīndriyavacana • 14
prāṇa • 66
buddhīndriyoktipuṃstva • 14
prāṇabhṛt • 110
bodhaka • 133
prātiśākhya • 129
brahman • 50
prādhānya • 36
brāhmaṇa • 8, 11, 30, 33, 41, 50, 77,
prāptiparihāra • 66
121, 128, 135, 142, 146
prāmāṇya • 9, 12, 18, 19, 47, 48, 77,
brāhmaṇatvasāmānya • 77
85, 87, 98, 103, 104, 116, 118, 122
brāhmaṇī • 77
prāya • 131
pra√arth • 46
bhakṣaṇa • 44, 54
188 INDICES
bhaṇyate • 66, 67 mukhya • 71

bhadra • 118 mṛgya • 118
bhāga • 18 mṛṣā • 119
°bhāj • 85 medhāvin • 44
bhājanaloka • 42 mokṣa • 41
bhāvanā • 70, 87 modamāna • 143
bhāvika • 116 moha • 77, 78
bhāsvara • 115 mlāyante • 143
bhinnakrama • 96 mleccha • 144, 145
bhū • 44 mlecchagamya • 144
bheṣaja • 61, 62, 103
bhoktṛ • 66, 67, 68 yajeta • 138, 139
bhoktṛtā • 66, 67 yajña • 132
bhoktṛtva • 67, 68 yajñāṅga • 132
bhraṃśya • 144 yatkiñcanagrahaṇa • 54
bhrānti • 144 yatna • 35, 66, 67
yathāśruta • 136
maṇḍala • 42 yathāsthita • 48
mati • 118, 144, 145 yatheṣṭa • 60
madīya • 115 yava • 140, 142, 143, 145
madukta • 115 yaśas • 50
madhyamapadalopin • 71 yāga • 132, 139
manogati • 66 yāgādikarman • 19
mantra • 11, 13, 14, 50, 105, 122, 129, yājñika • 142, 145
131, 132, 133, 134, 135, 136, 142 yukti • 16, 55, 111
mantrimukhya • 71 yuktibādhā • 55
malina • 46 yuktimannaya • 118
mastaka • 39 yuktiyukta • 42
mahākāruṇika • 40 yogya • 57
mahāyānapratirūpaka • 46 yogyatā • 57
mahāyānavidviṣṭa • 46 yaugapadya • 69
māṃsa • 15, 44, 54
māṃsapiṇḍa • 39 ratnaparīkṣā • 142
mātṛ • 71, 133 rathakarman • 66
māndya • 57 rathyāpuruṣa • 12
mārga • 105 raśmi • 44
māsa • 62 rasa • 63, 103, 142
mīlana • 70 rāga • 69
mukhavarṇa • 46 rāgādidoṣaduṣṭa • 119
INDICES 189
rāgādimalina • 46 varāha • 140, 143

rāgādirūpa • 112 varṇa • 12
rāgādisaṃyuta • 123 varṇavāda • 10
rāgādisamutthāpita • 113 varṇānupūrvī • 122
rāgādyabhibhava • 45 varṇyate • 14
rāgitā • 119 √vaś • 42
√rud • 128 vastu • 68, 70, 73, 75, 84, 98
rūḍhārtha • 133 vastutas • 116
rūpa • 59, 60, 69, 71, 75, 83, 103, 131 vastutva • 120
rūpaprakāśana • 60 vastubalapravṛttapramāṇa • 86
rūpasāmānya • 36 vastubalāyātānumāna • 64
rūpasāmyamātra • 63 vastubhūta • 77
rūpasthāna • 71 vastusvabhāva • 60
vahanti • 137
lakṣa • 42 vākkarman • 68, 77
liṅga • 66, 98, 99, 101, 112, 124 vākya • 12, 53, 62, 70, 103, 115, 129,
liṅgasamavāya • 134 132, 135
loka • 12, 52, 135 vākyabheda • 138
lokaprasiddha • 35 vākyaśeṣa • 143, 147
lokaprasiddhi • 135 vākyārthanirṇaya • 141
lokavāda • 16, 25, 49, 123, 124, 145 vākyaikadeśa • 129
lokavyavahāra • 59 vāgvyavahāra • 76
lokasanniveśa • 104 vācaka • 51, 74
lokasammata • 120 vācikākhya • 140
lopin • 71 vāc • 44, 68, 125
lobha • 78 vācya • 44, 58, 73, 74, 136
laukika • 45, 48, 55, 139 vātaputrīya • 55
vāda • 10, 16, 17, 25, 49, 77, 124, 134
vaktṛ • 15, 57, 73, 84, 100 vādin • 70
vaktṛtva • 120 vāyumaṇḍala • 42
vaktrabhiprāya • 75, 84 vāri • 44
vacana • 14, 19, 34, 37, 71, 72, 74, 119 vāstava • 12, 41
vacas • 115, 118, 119 vāstavasambandhavāda • 12
vajra • 44 vāstavārthasambandha • 12
vañjula • 140, 143 vikalpa • 92, 93
vati • 63, 103 vikāra • 66
vayas • 71 vigraha • 66
vayovaikalya • 71 vicāra • 37, 103
varam • 85, 101, 103, 108, 110 vicārita • 96
190 INDICES
vijñāna • 78, 120 vihita • 143

vitathābhidhāna • 100 vīrya • 142
vidyā • 7 vṛttāntānvākhyāna • 130
vidvas • 40, 137 vṛtti • 13, 14, 51, 56, 63, 76, 103
vidviṣṭa • 46 vṛtra • 135, 136
vidhi • 40, 130, 138, 148 vṛddha • 19
vidhisiddhi • 64 vṛddhavyavahāra • 141
vidhurayanti • 17 vṛddhi • 45, 70
vināśahetu • 69 vṛṣaṇa • 50
vināśitvānumāna • 11, 69, 122 vṛṣabha • 132
vinirgama • 40 vetasa • 140, 143
vipakṣa • 13, 14, 63, 119 vettṛ • 17, 33, 34
viparīta • 48 vedaka • 105
viparyaya • 113 vedaprāmāṇya • 12, 18, 77, 124
viparyastamati • 118 vedabhāga • 18
vipāka • 142 vedamūla • 9
viprakarṣa • 49 vedavākya • 103
viprakṛṣṭa • 42 vedaśāstrāvirodhin • 147
vipratipatti • 71, 72, 140 vedāpauruṣeyatva • 12
vipratiṣedha • 132 vedārtha • 15, 18, 33, 141
vipralambha • 100 vedārthagati • 17
vibhāga • 56, 136 vedārthajñāna • 34
vibhāgakriyā • 75 vedārthatattva • 34
viruddhadharma • 75 vedārthabodhaka • 133
virudhyate • 86 vedārthavitparamparā • 141
virodha • 20, 51, 68, 70, 112, 140 vedārthopadeṣṭṛ • 45
virodhādarśana • 85 vedin • 145
virodhāsiddhi • 14 vedaikadeśa • 62, 103
vivakṣā • 15, 58, 74 vedoktārtha • 72
vivakṣita • 58 vedopakṣepa • 56
viveka • 16 vaikalya • 71, 77
vivektum • 144 vaidagdhya • 71
vivekṣyante • 142 vaidika • 12, 17
vivecana • 16, 34, 45 vaidya • 142
viśuddha • 112 vaiparītya • 71
viśuddhi • 64, 87, 103 vaiphalya • 100, 105
viṣaya • 19, 36, 56, 63, 64, 68 vairūpya • 71
visaṃvādabhāj • 85 vyatireka • 14, 63, 111
vistāryate • 12 vyatirekāsādhana • 63
INDICES 191
vyapadiśyatām • 137 śāstra • 55, 64, 68, 70, 86, 144, 145,
vyabhicāra • 62, 63, 64, 71, 72, 89, 96, 147
103, 104 śāstrakāra • 68
vyavasthāpanīya • 55 śāstrastha • 141
vyavasthita • 35, 55 śāstrārtha • 55, 70
vyavahāra • 45, 59, 76, 141 śiṣṭa • 141
vy√ākhyā • 141 śiṣya • 59
vyākhyā • 141 śītanodana • 62
vyākhyātṛ • 37 śīlādiguṇa • 77
vyākhyātṛparamparā • 141 śuddhatva • 87
vyākhyābhūta • 53 śuddhi • 78, 87
vyākhyeya • 94 √ś ubh • 85
vyāja • 56 śubhāśubha • 66, 67
vy√āp • 75 śūdra • 11
vyāpti • 111, 117 śṛṅgagrāhikayā • 46
vyāptyasiddhi • 104, 117 śeṣa • 63, 141, 143, 147
vyutpanna • 58 śeṣavat • 13, 14, 20, 29, 62, 63, 103
śeṣavadanumāna • 13, 62, 63, 95, 102,
śakti • 35, 140, 141, 144, 145, 147 105, 110, 113, 114, 115, 117, 119
śaktiyukta • 17 śeṣavadanumānavyabhicāra • 103
śaktisādhana • 14 śodhana • 113
śakyapariccheda • 87, 99 śobheta • 85
śakyavicāra • 103 śraddheya • 19, 114
śapatha • 46 śravaṇa • 85
śabda • 12, 17, 42, 44, 45, 60, 75, 84, śravaṇamātra • 39
93, 98, 101, 103, 116, 123, 140, 141 śruti • 87, 136, 141
śabdagaḍumātra • 39 śrutisāmānyamātra • 50
śabdaprabhava • 89 śrutismṛtyavadhāraṇa • 141
śabdasāmānya • 135 śvamāṃsa • 15, 44
śabdāpabhraṃśa • 144 śvamāṃsabhakṣaṇa • 54
śabdārthasambandha • 10
śarīra • 77 saṃvatsara • 62
śarīrapīḍana • 78 saṃvāda • 43, 149
śarīrasamavāyin • 66 saṃvādin • 35
śākhā • 16, 47, 143 saṃvitti • 66, 67
śākhābhiyukta • 145 saṃvid • 68
śābda • 89 saṃśaya • 40
śābdavyavahāra • 45 saṃśayahetu • 63
saṃśayita • 103
192 INDICES
saṃśleṣa • 77 samāhāra • 12
saṃsāra • 85 sampratyayin • 86
saṃskāra • 37, 65, 98, 115 sampra√dā • 142
saṃskṛta • 37, 115 sampradāya • 16, 41
saṃsyandayati • 37 sambaddha • 86, 96
saṃhitā • 128 sambaddhatā • 70
sambaddhānuguṇopāya • 96
saṅkara • 142 sambandha • 10, 12, 17, 19, 66, 67, 70,
saṅketa • 15, 28, 60 114, 115, 116
saṅketayati • 58 sambandhānāditāmati • 144
saṅketasmṛti • 60 sambhāvanā • 47
saṅketānupālana • 41 sambhāvanīyapuruṣavacana • 104
saṅketāpekṣa • 45 sambhāvyate • 43
saṅgatārthatā • 70 sammata • 120
sañjñā • 60 samyagjñāna • 45
sañjñita • 51 sarvagata • 66, 67
satya • 62 sarvagāmitva • 141
satyatā • 64 sarvajña • 40, 75
satyavāditā • 19, 114 sarvanirāśaṃsa • 69
satyārtha • 43 savana • 133
saddharmanīti • 83 sāṃvyavahārikanitya • 10, 60
santāna • 76 sākṣātkṛtadharman • 142
santāpārambha • 78, 124 sātmībhūta • 73
sandigdha • 143 sādhaka • 51, 56
sandigdhaviparyastamati • 118 sādhana • 14, 17, 34, 63, 68, 103
sandigdhāsiddhatā • 115 sādhanaphala • 62
sandeha • 13, 14, 141 sādhanabhūta • 71
sanniveśa • 42, 104 sādhayitukāma • 19, 103
saptamī • 73 sādhayitum • 36
samaya • 13, 15, 46 sādhāraṇa • 93
samartha • 44, 64, 68, 85 sādhita • 114
samavāya • 65, 66, 67, 68, 134 sādhutva • 142
samavāyikāraṇa • 65, 66, 67, 68 sādhuśabdanirūpaṇa • 142
samavāyin • 68 sādhya • 62, 115
samaveta • 132 sādhvasādhukarṣāpaṇa • 144
samāmananti • 143 sādhvasādhutvalakṣaṇa • 142
samāropa • 74 sāmayika • 12
samāśraya • 66 sāmayikaśābdavyavahāra • 45
samāsa • 71 sāmarthya • 113
INDICES 193
sāmānya • 14, 18, 36, 50, 77, 84, 87, svadarśana • 72

88, 89, 90, 93, 94, 95, 96, 97, 99, 107, svadṛṣṭamārgokti • 105
108, 135 svapratīti • 60
sāmānyalakṣaṇa • 96 svabhāvatas • 58, 60
sāmya • 17, 63 svabhāvaniyata • 60
sāmyāsiddhi • 14 svabhāvahetu • 97, 114
sārathi • 66 svarūpatas • 66, 67
sārūpya • 134, 135 svarga • 26, 27, 44, 49, 51, 52, 70, 93,
siddhānta • 46, 64, 126, 129, 132, 141, 94, 96, 126, 139, 140, 148
142 svargakāma • 15, 24, 27, 40, 126, 137,
siddhāntin • 132, 136, 138, 140, 144 138, 139, 148
siddhi • 14, 64, 75, 84 svalakṣaṇa • 92
sidhyati • 12 svavacana • 72, 86
sukha • 65, 66, 67, 68 svavāc • 17
sukhaduḥkha • 65 svaviṣaya • 37
sukhasaṃvitti • 67 svaśāstra • 68
sukhasaṃvid • 68 svābhāvika • 33
sudhī • 42 svābhyupagama • 46
sūkta • 133 svārthatattva • 142
sūkṣma • 36 svārthānumāna • 9
sūcaka • 75, 84 svecchānurūpa • 16
sūtrārthavādin • 70
sūrya • 129 havana • 44
soma • 27, 50, 132, 139 hasta • 129
stutyartha • 130 hastasañjñā • 60
steya • 131 hāna • 124
sthāṇu • 57 hāyanī • 139
sthāna • 52, 59, 69, 71 hiṃsā • 42
sthānīya • 71, 72 hitāhita • 66
sthūla • 141 hima • 61, 62, 103
snāna • 69, 77, 112, 113, 124 hiraṇya • 129, 131
sneha • 77 √hu • s. juhuyāt
sparśa • 36 hetu • 8, 18, 56, 62, 94, 97, 98, 99, 101,
sphoṭa • 10, 122 111, 114, 115, 119, 137
smāraka • 132 hetutvapratikṣepa • 14
smṛti • 141 hetuvidyā • 7
smṛtyavadhāraṇa • 141 hemanta • 36
svakriyā • 68 heya • 41, 100
svataḥprāmāṇya • 122 heyopādeya • 100
194 INDICES
heyopādeyatattva • 100, 105 don du rnam par rtog pa tsam • 93

√hu • s. juhuyāt don tsam • 93
don la mi slu ba • 94, 99
hotṛ • 133 draṅ po • 56
hotra • 44, 137, 138 gdon mi za bar • 103
homa • 138 bdag • 67, 68
bde ba • 68
ʾdu ba daṅ ldan pa • 68
Tibetan words ʾdus byas • 98, 115
sdug bsṅal • 68
kun la rag ma las • 69
bkaʾ • 86 rnam par rtog pa • 92, 93
skad cig ma • 98, 115 dpe • 103
skyes bu tshad mar gyur pa • 15 dpyad pa gsum • 86, 88
skyes buʾi ʾjug pa • 116 spyi • 14, 93, 98
mkhas pa • 86 phyi rol gyi dṅos po • 98
ʾkhrul pa • 103 phyi rol gyi don • 93, 98
go skabs med • 110 ʾphrod pa ʾdu ba can gyi rgyu • 67
dgaʾ baʾi tshal • 49 ʾphrod pa ʾdu baʾi rgyu • 68
dge bar ṅes pa • 86 dbaṅ po las ʾdas paʾi don • 38
sgra • 93, 98 bar du • 86
ṅag • 67 byin gyis brlabs pa • 67, 68
ṅag gi las • 67 byed pa po • 67
ṅan soṅ • 85 bram ze ñid kyi spyi • 77
dṅos su • 116 blo • 99
mṅon par ʾdod pa • 68 ʾbru maṅ poʾi sñiṅ po can gyi ʾbru maṅ
mṅon par ʾdod pa ma yin pa • 68 po pa • 53
mṅon sum • 36, 38, 99 ma mthoṅ baʾi don • 93
mṅon sum daṅ mtshuṅs pa • 103 mi ʾdra ba • 93
ʾjig rten paʾi sgra daṅ thun moṅ ma mi slu ba • 86, 93, 98, 99
yin pa • 93 mi slu bar rtogs paʾi rgyu • 94
ʾjug pa • 110, 116 tshad ma ñid • 98, 116
rjes su ʾjug pas • 75 tshad mar gyur pa • 15
rjes su dpag pa • 38, 55, 99 tshul gsum gyi brtag pa • 86
rjes su dpag pa ñid • 93, 94, 98, 99, mtshuṅs • 93, 99
116 za ba po • 68
ñes pa zad paʾi tshig • 96, 98, 99 gzugs • 103
sñiṅ rje goms par gyur pa can • 73 yid ches paʾi rjes dpag • 113
rtags • 99 yid ches paʾi tshig • 93, 94
brtag pa • 86 yid ches pas rab tu sbyar ba • 93
brten par bya ba • 86 yid ches tshig • 93, 98
bstan bcos • 55 yoṅs su gcod pa • 99
yoṅs su dag pa • 86
tha ma • 86
thun moṅ ma yin pa • 93 raṅ gi bya ba • 67
thog ma • 86 ro • 103
mtho ris • 93 las • 67
INDICES 195
las kyi ʾbras bu • 68 Beardsley • 149

luṅ • 98, 116
luṅ gi don la brten pa • 55 Derrida • 149
lus • 67 Dunne • 89, 90, 91, 95, 96, 98, 100,
lus daṅ ṅag gi las • 67 106, 112
log pa • 56 Eltschinger • 89, 92, 97
ś in tu lkog tu gyur pa • 99, 110 Franco • 105
ś in tu lkog tu gyur paʾi don • 36, 98 Hayes • 92, 95, 97
sems las byuṅ ba • 67 Kataoka • 109, 114
bsod nams • 67 Lasic • 95
bsod nams ma yin pa • 67 Ricœur • 149
slob dpon • 98, 99, 116 Tillemans • 88, 89, 90, 92, 93, 95, 97,
100, 106, 107, 108, 110, 111, 113,
lhag ldan • 103 114
lhag ma daṅ ʾdra ba • 103 van Bijlert • 91, 95
lhag ma daṅ ldan pa • 103 Wimsatt • 149
Witzel • 146
Yaita • 89, 91, 95, 100

Modern authors
Barthes • 149

Can The Veda Speak - Dharmakirti

Uploaded by

Copyright:

Available Formats

Can The Veda Speak - Dharmakirti

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Can The Veda Speak - Dharmakirti

Uploaded by

Copyright:

Available Formats

Contents

Abbreviations and Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . 151

Sanskrit texts and genres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174

Place names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174

Sanskrit and other names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

Gods and saints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

Schools and religious groups . . . . . . . . . . . . . . . . . . . . . . . . . . . 176

Sanskrit words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176

Tibetan words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194

Modern authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

CAN THE VEDA SPEAK?

BEITRÄGE ZUR KULTUR- UND GEISTESGESCHICHTE ASIENS

Herausgegeben von Helmut Krasser

Can the Veda speak?

Vincent Eltschinger, Helmut Krasser, John Taber

British Library Cataloguing in Publication data

Die verwendete Papiersorte ist aus chlorfrei gebleichtem Zellstoff hergestellt,

Alle Rechte vorbehalten

1 See Wakahara 1990.

The Place of PVSV 164,24–176,16 in the Work

1 On Dignāga, see Frauwallner 1959 and Hattori 1968.

2 General introductions to Dharmakīrti’s thought include Steinkellner 1971,

1.2. We know next to nothing of Dharmakīrti’s life, though he is

2007a: 25–29. I accept Krasser’s new chronology.

hence with private inferential judgment (svārthānumāna), this topic

sion on scriptural authority occurs towards the end of Dharmakīrti’s treatment

11 Properly speaking, Dharmakīrti’s own doctrine is expounded in PVSV 107,14–

by ephemeral articulated sounds; PVSV 134,26–141,14).12 In all this,

12 On all this, see Eltschinger 2007a: 239–385.

13 See Sakai 2011.

Dharmakīrti’s own doctrine concerning phonetic series (PVSV 157,29–162,11).

ing but arbitrary conventions, which Dharmakīrti duly demonstrates in PVSV

21), Steinkellner 1979: 73–78 is to be considered as an important contribution

In PV 1.312–329/PVSV 164,24–173,13, Dharmakīrti turns to the ab‐

19 PVSV 164,9–24 (together with PV 1.310cd–311): aśaktisādhanaṃ puṃsām ane‐

naiva nirākṛtam // pratipāditā hi puruṣakṛtās tadadhiṣṭhānāc ca phaladā man‐

hence intentionality, the Mīmāṃsaka deprives it of any meaning, for

126–127 and 135–140.

sented as vedārtham ākhyātāraḥ. In PVṬ Ñe D47a6/P54a1–2, these are termed

is beyond the reach of ordinary experience.25 In other words, the

25 See PVṬ Ñe D46a7/P52b8 = PVSVṬ 585,27–28 and PVṬ Ñe D50a1–2/P57b3–4

30 (to be compared with PVṬ Ñe D46b1/P52b8–53a1).

= PVSVṬ 593,27, PVṬ Ñe D46b7–47a1/P53b2–3 ≈ PVSVṬ 586,18–19.

31 PVṬ Ñe D51a6/P59a8–b1 = PVSVṬ 592,7, PVṬ Ñe D51a6–b1/P59b1–4 ≈ PVSVṬ

their own assertion.”33 The Mīmāṃsakas gain access to the invisible

iti kathaṃcid atiśayavato jaiminyādeḥ sakāśād vedārthagatir bhavatīti puruṣāti‐

bādhante ’tiśayavatpuruṣapratikṣepeṇa vedārthagater asambhavāt /.

ndriyārthadarśī puruṣa ity evaṃvādino jaiminīyāḥ, and PVṬ Ñe D45b2/P51b7–8

tīndriyārthadarśipuruṣapratikṣepasādhanas is puruṣatva or puṃstva. See PVṬ Ñe

ptopadeśenāyurvedenādṛṣṭārtho vedabhāgo ’numātavyaḥ pramāṇam iti, āptaprā‐

rather unorthodox Mīmāṃsaka be? Should we think of Bhavadāsa,

contrasts the vṛddhamīmāṃsakas (jarajjaiminīyas in NM I.664,16) and the fol‐

ṃsaka(s), ŚV codanā 121 testifies to Kumārila’s familiarity with such a strategy:

garding”) in the following statement (PVSV 173,16–17): anyas tv apauruṣeyam

kādiḥ) might testify to. See also Steinkellner 1979: 78 n. 252.

on this passage. The second problem is of a philosophical nature.

142, p. 77 n. 172 and p. 78 n. 174.

withstands critical analysis. In other words, the first strategy (PV

A Synopsis of PV(SV) 1.312–340

1. The validity of the Veda cannot be based on “authorlessness”