Professional Documents
Culture Documents
Early Saivism and The Skandapuraa Intro
Early Saivism and The Skandapuraa Intro
VOLUME XXI
Editor
Editorial Board
Advisory Board
J. WiIIiams, Berkeley
Peter C. Bisschop
2006
This book was printed with financial support from the
Netherlands Organisation for Scientific Research (NWO)
All rights reserved. No part ofthis publication may be reproduced, stored in a retrieval
system, or transmitted, in allY form or by allY means, electronic, mechanical, photocopying,
recording, or otherwise, without the prior written permission a/the publisher.
Preface
v
VI Preface
following sites:
<http://www.uc1.ac.uk/-ucgadkw/indo1ogy>
<ftp://ccftp.kyoto-su.ac.jp/pub/doc/sanskrit>
<http://www.sub.unigoettingen.de/ebene_l/fiindo1o/greti1>
<http://www.sansknet.org>
In addition I had access to a number of electronic texts of Pura'.las
at the Institute of Groningen, La. the entire text of the editio princeps of
the SkandapuralJa, entered by Rob Adriaensen and Harunaga Isaacson.
The importance of searchable electronic texts for studies such as the one
undertaken here needs no explanation, and I am indebted to all those who
have contributed to this revolution in Indology.
The layout of this work follows the example set by SP Vol. I and IIA:
the edition has been prepared in EDMAC and the English parts have been
typeset in Jg.'I'gX. To have had the electronic files of SP Vol. I and IIA as
an example before me was a major advantage, which I duly acknowledge.
I am grateful to Prof. Hans Bakker and the editorial board of the
Groningen Oriental Series for accepting this book in their splendid series
and to Egbert Forsten for making its publication possible.
On a more personal note I thank Jannet, Rasmus and Elva, for being
with me.
Peter Bisschop
Edinburgh, April 2006
Contents
PREFACE v
CONTENTS vii
INTRODUCTION 1
SYNOPSIS 63
65
SPRA 1 74
SPRA 2 76
SPRA 3 78
Vll
viii Contents
81
82
EDITION 87
Symbols and Abbreviations in the Critical Apparatns 89
SPs 91
SP RA 1 112
SP RA 2 122
SPRA 3 131
SP RA 4 146
SP RA 5 154
ANNOTATION 171
SPs 173
SP RA 1 224
SP RA 2 235
SP RA 3 248
SP RA 4 270
SP RA 5 278
ApPENDIX 299
Concordance of SPs and SP RA 301
BIBLIOGRAPHY 303
Abbreviations 305
Manuscripts 307
Printed Sources 308
INDEXES 333
Index of Names and Subjects 335
Index of Text Passages . 349
Introduction
Saiva Topography and the
The study of the sacred topography of early Saivism, important for our
understanding of the history of Saiva religion, is still in its infancy. A
systematic exploration of the many lists of Saiva places in Sanskrit litera-
ture has never been undertaken. 1 The present study does not remove this
lacuna in scholarship but concentrates on one such list, contained in the
original Skandapura1)a. The importance of this text which has been as
good as forgotten since the fourteenth century' for the study of early
Saivism, including its topography, is undeniable.
THE
1 Sanderson (2003-04, pp. 403-409) refers to four such lists: 1) the five lingas
A vimukta, Kedara, Orp.kara, Amara and Mahakala - mentioned in an inscription
of 1063 AD on the Ardhamal).9.apa of the Amaresvara temple at Mandhata (El 25,
p. 185); 2) the forty Saiva sites, in five sets of eight (paiica§taka) , taught in the
Sivadharma and the literature of the Saiva Mantramarga; 3) the forty-eight Siva
sites taught in Skandapurii.{la (SP Bh ) 167, the subject of the present work; 4) the
sixty-eight Siva sites taught in the Tfrthamlihatmya of the Nagarakhal.lr;la of the
Skandapural.la (SkP Nagarakhal.lc;la 108-109). Regarding item 3, I wonder whether
the number forty-eight is significant; it is not stated in the text itself, as Sanderson
also observes. To arrive at the number forty-eight one has to count the toponyms
mentioned and not the sanctuaries. Thus Sanderson counts Kumbhakaresvara,
Utkutukesvara and ChagalaI).Q.esvara as separate items, three toponyms which are
in fact all subsumed under Mahabhairava (SPs 69). On the other hand, eight
places in Magadha connected with LakulIsa (SPs 169) are not counted among the
forty-eight but included together with Prahasitesvara in Pataliputra in Magadha
(Sanderson 2003-04, p. 406, n. 204).
2 The last author who quotes from our text appears to be CaI).Q.esvara, the minister of
Hari!?el).a, the Karl).ata king of Tirhut, and writer of several Dharmanibandha Ra-
tnakal'as. Cf. SP Vol. I, pp. 11-13, 18. For detailed discussions of the version of the
SkandapuraI}a known to Cal).Q.esvara see Torzsok 2004, pp. 31-39, and Harimoto
2004a, pp. 48-55.
3
4 Introduction
by a second volume in 2004 (SP IIA), which contains the Varalfasl cycle
told in chapters 26-31.14. A volume of individual studies of the Skanda-
pura1).a was published in 2004 as well (Bakker 2004).
The text is to be distinguished from the Skandapura1).a (SkP) pub-
lished by the Veilkatesvara Press in 1910. An examination of the quo-
tations from the Skandapura1).a in the Dharmanibandha literature estab-
lishes that we are concerned here with the original Skandapura1).a. In
the course of history the name Skandapura1).a became a locus of attribu-
tion which could accomodate various texts seeking to authenticate their
authority, by attaching to them the label skandapuriif!asya x-kharpjalJ.3
The text has come down to us in three different recensions: a Nepalese (S)
recension, a Revakha1).Qa (R) recension and an Ambikal<ha1).Qa (A) recen-
sion. The latter two, although bearing a khaw!a name, in fact transmit a
version of the original Skandapura1).a and are thns not on a par with other
khaf!ljas such as e.g. the Kasfkha1).Qa or the Kedarakha1).Qa. The S recen-
sion is transmitted in four Nepalese palm-leaf manuscripts, 4 the earliest of
which (SI) is dated 810 AD. This recension is in principle the text that is
edited in the critical edition of the Skandapuriil).a (SP Vol. I, p. 41). The
recension which styles itself Revakha1).Qa of the Skandapuriil).a is trans-
mitted in a single Bengali paper manuscript dated 1682 AD. The Ambi-
kakha1).Qa recension has been transmitted in a number of recent paper
manuscripts which all descend from a single hyparchetype. Four manu-
scripts of this recension had been used in SP Vol. I; in SP Vol. IIA and
the present work an additional Ambikakha1).Qa manuscript from Dhaka
(A,), of which a copy has been acquired by Dr. Kengo Harimoto, has been
used as well. 5 The additional manuscript from Dhaka (Dhaka University
Library No. 3376), is a paper manuscript written in Bengali script and is
clearly the oldest of the A manuscripts, although no date is given in the
manuscript itself. Its script at times shows some similarity with that of R.
SP RA A, A2 A3 A, A, R
1 170 170 179
2 171 171 180
3 172 172 181
4
5 173 173 182
In order to give a general idea about the amount of change, I have listed
the sanctuaries mentioned in the ayatana account of 8Ps and 8PRA in
table 2 below. The fourth column enumerates the number of verses in
8P RA in comparison to 8Ps:
6 See below, n.151 on p.53, for some preliminary observations on the value of this
manuscript.
7 Adriaensen, Bakker, Isaacson 1994, p. 326.
8 BhaHaraI only used the S manuscripts for the constitution of this chapter in the
editio princeps. The siglum SPS is used throughout the present work to refer to
the critical edition of the Nepalese recension of the chapter corresponding to SFBh
167.
9 The text of the R and A recensions has not been edited before. For references to
this text I use the siglum SPRA; the number immediately following this siglum
refers to one of the five sub-chapters, while the subsequent number refers to the
relevant verse in that sub-chapter. For a discussion of the editorial principles
adopted in establishing the editions of SFS and SFRA see below, pp. 51 ff.
10 Cf. Harimoto 2004a, pp. for a discussion of the different adhyaya numbers
in the A manuscripts.
6 Introduction
13 The following section partly reproduces some remarks made by me in a paper pre-
sented at the XIIth World Sanskrit Conference in Helsinki 2003 (Bisschop 2004).
14 Cf. T6rzs6k 2004, p. 18, and Harimoto 2004a, p. 45.
15 A similar frame is used only sporadically in the S recension. A fine example is the
Viira1;lasfmaha-tmya (SP 26-31.14), in which Siva shows the Goddess the sacred
lingas of Varar:lasI.
8 Introduction
16 MBh 1.57 and SP 19.1-13. Cf. SP Vol. I, p. 95, n. 107. For the three accounts of
Vyasa's birth in the l\1ahi'ibharata see Sullivan 1990, pp. 27-29.
17 Kalinandana: SPRA 1.4a, SPRA 4.52d; Kalija: SPRA 1.8d, SPRA 2.21b, SPRA
5.23b, SP RA 5.26d; Kiileya: SPRA 2.1d, SPRA 3.lb, SP RA 5.28a. A testimonium
for one of these epithets (Kaleya) is Cal).<;iesvara's Krtyaratnakara, where a long
passage is quoted from the Bhairavotsava (on which see Torzsok 2004, pp. 31-39)
which starts as follows: srutva kiileyavacanar[L kiilavedf tapodhanaf:t I sanatkumaraiy,
provaca vyiisam ambudaniJ:tsvanaJ:t II nflotpaladalasyiimaJ;, syiimiimbujasamanvital} I
811'U ki"ileya ki"ili"ibha ki"ilaka1.'!havicee!itam II (KR p. 386, ll. 19-22). These two verses
are absent in the Krtyakalpataru's quotation of the Bhairavotsava (NK pp. 413-
421).
18 SP 19.22 4*, 1.19. In the same passage, SP 19,224*,1. 37, R reads kiileccha which
is probably a corruption of kiileya.
19 Cf. also the additional passage quoted as occuring "gha-pustake" (Le. A 3 ) by
Bhattarru after SPBh 112.120, in which the first verse has Kaleya and the last
Kalija.
Saiva Topography and the SkandapuraI.Ja 9
redactor(s) of SPRA seem to have tried to adapt the verb cyavanti, orig-
inally belonging to the word-play, to the new context (cf. annotation ad
SP RA 1.18-19). In SPRA 3.1 an apparent contradiction has been obliterat-
ed by clever rewriting: in SPs 62 an unspecified twice-born (dvija) is said
to have performed japa at J apyesvara, which seems to contradict a story
told in SP 21-22, according to which Nandin did japa there (cf. ad SPs
62d). In the parallel SPRA 3.1 the nominative dvija1;t has been replaced
by a vocative and a nominative Nandi(n) has been introduced, making
him the subject of the japa (cf. ad SPRA 3.1). A final example is the
metrical problem in SPRA 4.20f (cf. ad lac.), which is very probably the
result of the transformation of SPs 8lf into a popular emology (ni7'ukti).
In the annotation the reader will find many more instances pointing to
the secondary nature of SPRA in comparison to SPs.
In general, we can observe that the order of the ayatanas listed in
SPRA is the same as in SPs. The most significant difference in this respect
is Varal).asT, which is mentioned earlier, somewhere in the middle of the
ayatana account in SPRA (4.1-9), and thus occupies a less prominent
position (cf. annotation ad SPRA 4.1a). This is also one of the few places
where SPRA has fewer verses than SPs. Only two ayatanas mentioned in
SPRA have no parallel in SPs: the Himalayan peak Naubandhana (SPRA
1.11-18)20 and Audaka, the oceanic city ofVarul).a (SPRA 1.62-65). Two
toponyms in SPs on the other hand, Kedara and Madhyamesvara, find no
mention in SPRA. The last toponym in SPRA is Karohal).a, to which an
elaborate account of Siva's descent there is added (SPRA 5.39-108). In
the concluding verses (SPRA 5.109-116), the frame structure is referred to
again, in that the Goddess is addressed in the vocative (ambike, anaghe).
The ayatanas are divided over the five sub-chapters in a more or
less logical way, while each sub-chapter is concluded with a few verses
in non-sloka metre. The text distinguishes between supramundane sanc-
tuaries dealt with in SPRA 1 and Himalayan mountains in SPRA 2, but
no distinction is made in the text itself between the sanctuaries listed
in SPRA 3-5. The last four sub-chapters all end with a sanctuary to
which a rather elaborate story is added (Bhastresvara, Devadaruvana,
Amratakesvara and Karohal).a). That this underlies the division is made
explicit by the colophons of SPRA 2-4: devatayatanoddeSa1;t (SPRA 1); bha-
streSvarakhyanam (SPRA 2); devadaruvane mahadevasthana1!' mahatmya-
va'T"1,1ane (SPRA 3); amratakesvaranusa1!'sanam (SPRA 4); ayatanava'T"1,1ane
(SPRA 5).
Although the distinction between two kinds of sanctuaries, mundane
(gamya) and supramundane (agamya) , is also made in SPs 27, the con-
trast is made more explicit in SPRA by listing all the supramundane aya-
tanas in a separate, initial chapter. In addition, SPRA has made the text
more consistent in this respect, as can be observed from the following cas-
es in the part about the supramundane sanctuaries: SPRA 1.lOa has devo
va purvadevo where SPs 7c has manujo; SPRA 1.1Sa has tad ayatanarf'
divyam instead of tad ayatanarf' sambhor (SPs 9a); manava/:t (SPs 9b)
is omitted in SPRA 1.1Sb, which instead reads vai dvija; according to
SPRA 1.21d (no parallel) gods obtain the fruit of a thousand asvamedhas
when they go to Cyavana's sanctuary; SPRA 1.30a has tasminn ayatane
divye instead of tasminn ayatane rvdmrf' (SPs 15a); manuja (SPs 20c)
is omitted in SPRA 1.41; SPRA 1.43c mahad ayatanarf' divyarf' has no
parallel in SPs. In addition one may note the clarification of the expres-
sion yani gamyani manu§ai/:t (SPs 27b), referring to mundane sanctuar-
ies, in SPRA 1.6Sab manu§yagamyani ca yani yani kale§u pU'f}yayatanany
avanyam, and the first verse of the conclusion of SPRA, where the distinc-
tion between sanctuaries located on earth and those in heaven is once more
recalled (SPRA 5.109ab [no parallel]). We can thus observe a tendency in
SPRA to elaborate upon a point made only in passing in SPs. 21
The descriptions of the various ayatanas generally have one or more
verses in common with SPs, but the accounts given in SPRA are much
more elaborate. It is conspicuous that the longer parallels (eight padas or
more) are found relatively fi'equently at the beginning or end of an ayatana
description. 22 Thus it seems that the redactor(s) of SPRA used the original
text as the starting point and end for each toponym. In a number of
cases the additional material is placed in between these two points. The
additions are mainly mythological and can be classified broadly into two
categories.
The first category consists of what I would call text-internal cross-
references. These are re-tellings of or allusions to myths related earlier
in the text. A few examples should suffice (more detailed discussions of
these and other additions can be found in the annotation on the relevant
verses):
Gaurlsikhara: SPRA 2.13 adds a reference to the story of Somanandin, a
tiger who became a GaI,lesvara thanks to Dev!'s tapas. This story is
narrated in full in SPy 55 (ef. also SPy 69).23
21 For a similar idea in the Nlahabharata cf. MBh 3.83.88-89, which distinguish-
es between accessible (gamya) and inaccessible (agamya) tfrthas, which can be
reached by thought (manasa) alone and are occupied by Vasus, Sadhyas, Adityas
etc.
22 Cf. the concordance of SPs and SPRA in the appendix, where the number of padas
in common between the two recensions is given in the last column.
23 The critical editions of these chapters have been prepared by Prof. Yuko Yokochi
Saiva Topography and the SkandapuriiJJa 11
in her thesis on the worship of the Goddess in the Vindhya mountains (Yokochi
*2004).
24 There is some similarity with the tfrtha descriptions in the tfrthayiitrii accounts of
the Mallabharata. Cf. Griinendahl 2002, pp. More extensive descriptions
are given in the case of MahaIaya (SPs 28-35), Kedal'a (SPs 36-42), Mahabhairava
(SPs 64-69), Devadaruvana (SP s 72-80), Karohal,la (SPs 110-138) and Avimu-
kta (SPs It is noteworthy that the myths told in connection with these
sanctuaries have no parallel in the rest of the SkandapuraI}a, with the exception
of A vimukta, whose description is reminiscent of the Vara-QasYmahatmya in SP
26-31.14.
12 Introduction
mendicant and his ling a is struck down by the angry sages. This is
the only passage in the entire Skandapural)a in which this myth is
narrated and it seems to be the earliest known version. The parallel
in SPRA 3.36-109 contains a much more elaborate version of the
same myth. It had probably attained major importance among the
Saivas by the time of the redaction of SPRA, and the need was felt
to add more details.
Amrat(ak)esvara: SPRA 4.27-52 adds a unique myth in which the God-
dess turns Nandi's face into that of a monkey. From the triple laugh-
ter produced by Siva, Parvatl and Nandi at the sight of his monkey
face a white goddess is born whose moonlight lustre is said always
to be present at the sanctuary of Amratakesvara. No such myth is
told in the SPBl, nor do I know it from any other source. 25
Karohal).a: In addition to the elaborate account of Karohal).a and Siva's
descent there in SPs 110-138, many extra elements are found in
SPRA 5.39-107, such as the creation of the universe by Brahma, a
description of the evils of the Kaliyuga and a praise of dharma.
What unifies the additional material in SPRA is its mythological char-
acter, with a tendency to add hyperbolic imagery, suggesting that the
sanctuaries belong to a mythical sphere. By contrast, the descriptions in
the Nepalese recension appear much more 'down-to-earth' and exhibit real
topographical knowledge.
25 The story is, however, nicely integrated into the cycle of myths of the Skandapura1).a
by a cryptic reference in SPRA 4.47, in which it is said that 'the powerful one
who restrained the sound and agitation of will become the wife of this white
goddess. This saying refers to a story narrated in SP Bh 33.104-117: cf. annotation
ad SP RA 4.47b.
Saiva Topography and the SkandapuriiI.Ja 13
26 Although the text does not explicitly say so, Uslrablja could represent the nadir,
in that the word bfja ('seed') suggests a downward direction. Alternatively, Meru
may be considered to be the axis mundi and as such represent the centre, but in
that case the presence of USIrablja cannot be explained. As Prof. Henk Bodewitz
has pointed out to me, the order south, north, east, nadir, zenit, west, suggested
here, is awkward, and there seems to be indeed no parallel for this (cf. Bodewitz
2000). Although the classification need not be as rigid as proposed here, there can
be no doubt that some sense of different directions of cosmic space underlies the
presentation of the initial toponyms.
14 Introduction
27 It is possible, however, that some toponyms are in fact to be located in the north-
east. See the annotation on SPS 62 (Japyesvara) and SPS 69 (Kumbhakaresvara,
Utkutukesvara, Chagala1).<;I.esvara).
28 Cf. e.g. Arbman 1922, pp. 35-47.
29 See the annotation on individual places for details.
30 See annotation on SPS 88a for details.
31 The same dating is suggested as the origin of the SkandapuraIJa by Yokochi (1999,
p. 68) on the basis of a comparison of the mythology of Mahi::;asuramardinI and
iconographical evidence. Bakker and Isaacson date the composition of the original
SkandapuraI)a to "the 6th or, maybe, first half of the 7th century" (SP HA, p. 52,
with n. 174). Cf. also below, p.33, p. 37 and n. 266 on p. 205.
Saiva Topography and the Skandapura1)a 15
AN ADDITIONAL LIST IN SP s
After the praise of Avimukta as the most holy sanctuary on earth, where
one attains liberation at death, one would expect that the chapter on Si-
va's sanctuaries is finished. However, SPs continues with an additional list
(SPs 163-187) in which a number of sanctuaries in northern, western and
eastern India are mentioned. The character of this list differs significantly
from the preceding in that it just introduces the name of a sanctuary
and then continues with another in the next verse. 32 The list is free
from mythological elements and in many cases a topographical location
of the sanctuary at issue is given, e.g. 'in Pataliputra,' 'in Magadha,' 'in
Angadesa,' etc. Another peculiarity of this additional list is the apparent
absence of a topographical pattern: at times the text seems to switch from
the extreme east to the west and back again. There are indications that
this additional list did not belong to the text originally but was added
at a later stage, probably in the transmission of the Skandapura1)a in the
Nepalese recension.
Most significant in this respect is the absence of this list in the parallel
text of SPRA, which concludes with a long praise of That
this recension does not finish with Avimukta is due to the fact that this
sanctuary has been transposed to an earlier part of the text (SPRA 4.1-
9). The additional list in SPs only mentions sanctuaries located in the
north; South India is conspicuously absent. The last two toponyms are
both in the Himalaya: Nepala, where Siva is known as PaSupati, and
an unidentified sanctuary with the illustrious name Naikatungadhipesva-
ra ('Overlord of many peaks'). The mentioning of Nepala at the end
of the list indicates that it may have been added in the transmission of
the Nepalese recension. 33 Admittedly, PasupatTsvara, the name of Siva's
sanctuary in Nepala, is already attested in the Viira1)asfmahatmya (SP
29.82d) in a list of ling as which are said to enter KasI on parvan days.
Six of the lingas mentioned in this passage (SP 29.81-85) occur also in
the additional list found at the end of SPS. 34 The Viira1)asfmahatmya
32 For the names of sanctuaries listed in this part of the text see ll. 12 on p. 6.
33 The list also has a significant number of places located in the north-east, espe-
cially Magadha: cf. annotation ad SPS 169. The latter verse contains a curious
reference to eight sites in Magadha visited by LakulTsa with his pupils. Cf. also SP
Vol. IIA, p. 54, n. 175, for a proposed connection between this additional part and
the possible bringing of the Skandapura1).a to Nepal under the Maukhari princess
VatsadevI, or someone of her entourage.
34 PaSupatIsvara, SankukarI).e.svara, DrimicaI).Qesvara, Bhadresvara, Ekamra, {Ma-
ha- )Kalesvara. The other names mentioned, Kedara, Mahalaya, Madhyamesvara,
the two GokarI,).as, Sthanesvara, Ajesvara (= Bhastresvara?), Bhairavesvara (=
Mahabhairava) and KarohaI,).a, are all attested in the first part of SPs. This
testifies to the unitary character of the SkandapuraI)a. Cf. also the list of lingas in
16 Introduction
purification of sins, has been taught to you, son of the best of sages.
Tell [me] Vyasa, what more [shall] I tell you?
A possible candidate for this place could be NaikatUl1gadhipesvara, the
last item on the list, which shares the qualification sarvakamaprada (SPs
187f). However, in that case one would expect the verse to be immediately
after SPs 187, whereas now it follows upon a general statement concerning
the entire chapter. Considering the fact that there are indications that
the portion of the text after VaraJ.1asI has been added later, another, more
likely candidate is VaraJ.1asL The verse indeed fits with the conclusion of
the large VaraJ.1asI eulogy, which in the present state of the text also ends
with a non-sloka verse. Thus its presence here can be explained by the
fact that initially it followed immediately after SPs 162. For these reasons
my hypothesis is that the chapter originally ended with the mahatmya of
VaraJ.1asI and that the additional list in SPs in between SPs 162 and SPs
191 is an early interpolation. 37
Besides the ayatana account, which forms the subject of the present study,
the Skandapural}a contains many more chapters in which sacred topogra-
phy plays a role. A brief glance at indexes 3 (sthiinavise$e$u prati$thitani
lingani) and 4 (pu7.'yak$etra7.'i) in BhaHaral's edition reveals the large
number of lingas and holy places mentioned in the text. 38 In this para-
graph I present some of the material relevant to the study of its sacred
topography.
One who reads the text carefully is struck by the way the topograph-
ical references are intrinsically linked with the mythology of the PuraJ.1a.
As has been noted above, this tendency is even stronger in the case of
the ayatana account transmitted in the RevaklJal}ga and AmbikaklJal}ga
recensions. In general, the sacred topography is found at the end of a
chapter, and consists of a eulogy of the sacred site at which the event
narrated earlier is supposed to have taken place and has left traces. 39
37 Another indication of the secondary nature of this list is the double occurrence of
UjjayanI in SPS: first as the place where Siva's first pupil Kausika was initiated
as a PMupata (SP s 122), a second time as the sanctuary of Mahakalesvara (SPs
176).
38 Of the forty-eight (!) indexes given at the end of Bhattarru's edition the follow-
ing also contain relevant topographical references: 2 (sthiinaviSe§iil} [loka de§iib
purii'(1-i ca]), 5 (nepii1avi§ayiintarvarttfni katipayapu1}yasthaliini), 6 (parvaUily.) , 7
(nadfjaliisayalJ,) and 22 (deiavise§e§u sarrmivesitli devyaly,
39 Examples in SP Vol. I are: Naimisa (SP 4); Mahakapala (SP 7); Mount Mainaka
and SvarQaS,Qga (SP 9); Japyesvara and Paiicanada (SP 22). Cf. also SP 12.25
(Citrakii\a) and SP 25.22 (Srlparvata).
18 Introduction
40 The editors of SP Vol. I (p. 67, n. 23) note that the same etymology occurs in YaP
1.2.8 (= B<;lP 1.1.2.8).
41 Cf. SPRA 4.20.
42 I will deal with this parallel in a separate study and so I will not present here all
the arguments why I consider the Skandapura1)8 to be the borrower. Let me just
draw attention here to the naJ:t in SP Bh 114.4ab (etad akhyahi naT). samyag yathii-
vrtta7[t tapodhana) , a first person plural, naturally referring to the seefS questioning
the suta in YaP 1.54.1, but out of place in the SkandapuraI}8, where it refers to
Sanatkumara who is suddenly questioning and addressing an unintroduced silta.
Cf. also the vocative in SP Bh 114.5c (VaP 1.54.2c) and s'f'T)udhva'f[t
sa'f[tsitavmtiil;t in SP Bh 114.6d (VaP 1.54.3d). For the last verses preceding the Avi-
mukta verse there is no parallel in YaP 1.54, but Bc;lP is parallel
to SP Bh indicating that there is a lacuna in the text of the YaP. A
study of these parallels may throw new light on the text of the original Vayupura{la
(on which see Kirfel1927, pp.
Other Sources on Saiva Topography 19
THE MAHABHARATA
45 To be sure, the dating of the lVIahlibl1arata is a much debated issue and one should
always take into account the specific place in relative chronology of the part of
the text in which a particular passage occurs. It is generally agreed that the
major Tfrthayatraparvan - the book in which the U}rthayiitrii accounts are most
prominent - is a relatively late part of the text. Cf. Brockington 1998, p. 135,
144-145. On the problems involved in dating the 'normative redaction' of the
1vlahablliirata see Bigger 2002.
46 Some initial research into the directions of the pilgrim routes described in the Ma-
habharata has been undertaken by Bhardwaj (1973, pp. but much remains
to be done in this field. For a comprehensive list of the tfrthayatra texts in the lvla-
hiiibharata and the occurrences of the word tfrtha in the great Epic, see Vassilkov
2002. Cf. also Grlinendahl 1993, 2002; Oberlies 1995; Bigger 200l.
47 Cf. MBh l.207.4d, l.207.14b, l.210.1a, 3.81.169b, 3.85.22b, 3.85.23d, 3.86.16b,
3.87.1d, 3.88.1b and 3.92.13d.
Other Sources on Saiva Topography 21
48 The classification of twelve jyotirlingas does not seem to be attested before the
SivapuraJ;:Ja (SiP Satarudrasarphita 42.2-4). In an inscription on the Amaresvara
temple at Mandhata, dated 1063 AD (see n.l on p.3), which also records the
Halayudhastotra and the Sivadvadasanamastotra, occurs a verse enumerating a
set of five lirigas, Avimukta, Kedara, Orp.kara, Amara and Mahakala. The last
four are included in the traditional list of twelve jyotirlingas; in fact Amaresva-
ra is the liriga in Orpkara. Instead of A vimukta the jyotirlinga list includes the
historically later sanctuary Visvesvara in Varal].asT. Cf. also the related list in SkP
KedarakhaQQa 7.30-35.
49 See above) p. 18.
22 Introduction
50 Cf. SP Vol. I, pp. 20-22. See also above, n.42 on p. 18, and Bisschop 2004, p. 75f.
51 See e.g. Kirfel 1927, pp. x-xix, Hazra 1987, pp. 13-19, and Vielle 2005.
The five topics are: sarga, pratisarga, va'Tf1sa, manvantara and
var[l'siinucarita.
Other Sources on Saiva Topography 23
52 The verse from the B<;lP is strikingly similar to KuF 2.36.37, which deals with a
footprint in Gaya: muiijap'(?the padarp, nyastaT['L mahiidevena dhfmaUi I hitaya sa-
rvabhutanti1Jl. ntistiktiniiTfL nidarsanam II. The first half of this verse is parallel again
to 2.3.13.110 (= YaP 2.15.102): mU7J¢apT?!he pada,!, nyasta,!, mahiidevena
dhfmatii I bahudevayugtiTf1,s taptva tapas tfvrar[t suduscaram II. This last verse is
quoted from the VayupuralJ-a at TVK p. 168, II. 13-14. is the hill in
Gaya which has the (Kane IV, p. 782). Jacques 1962, discusses
the TVK quotation, but does not notice the parallel of the B<;lP (and the KuP) for
this passage, which is important for the early history of Gaya, because it pushes
the date of the tradition of the Murp;iapr:'?tha footprint further back in time. The
fact that the passage is shared by the BqP and the YaP indicates that it probably
already existed by the time of the original Vayupura{la, Cf. Jacques 1962, xliv-xlv,
with regard to the quotation: "On y trouve rattestation de quelques empreintes de
pied, notamment de celles de Matanga dans Ie Bharatasrama et celles de Mahadeva
sur Ie MUl,lqapr:'?tha, ce qui confirme notre hypothese qu'a l'origine, ces empreintes
de pied etaient dispersees sur tout Ie territoire de Gayaj eIles devaient provenir tres
vraisemblablement de Bodh-Gaya. On remarque rabsence du Vi.?1;/'upada dans ce
passage comme dans tout Ie chapitre du Kalpataru consacre a Gaya. Le MUl,lQa-
pr:'?tha semble avoil' l'importance qu'il a dans Ie Gayamahatmya du Vayu-pul'a.{Ja,
c'est-a-dire etre en fait Ie point central du Gayak:,?etra. L'objet de culte sur cette
colline n'aurait donc pas ete alors Ie Vi§1,tupada, mais Ie Mahadeva-pada." The
variant name in the KliP is also found in MBh 12.122.2c, MBh 12.122.4c and
MBh 13.26.40a, referring, however, to a peak of the Himavat. Cf. Falk 2003 on the
location of mount Munjavat or MufijaSikhara, which has a cave called the Kashmir
Smast. Possibly Siva was called there (Falk 2003, pp. 15 and 17). In this
connection it is interesting to observe that in MBh 12.122 not discussed by Falk
a story is told in which the Anga king Vasuhoma, who resides at Mufijapr:'?tha,
tells king Mandhata about the origin of da1,tq,a (punishment).
53 The Venkatesvara edition, which I refer to here, divides the Vayupura{la into two
parts, Plirvarddha and Uttararddha.
54 I have noted the following toponyms in common: Mahendra, Saptagodavara, Go-
karl,la, Srlparvata, Pu:,?kara, Prabhasa, Viraja, Kuruk:,?etra, Mahalaya, Devadaru-
vana, Prayaga, Kalafijara, Naimisa, Varal,lasI and Gaya.
24 Introduction
time. The contents of this list is discussed below in the section on the
Pasupatas (pp. 41 If.).
The MatsyapuralJa also contains a list of sriiddha places (MtP 22), but
probably this chapter stems from a later date. 55 The VaralJaslmallatmya
of the MatsyapuralJa (MtP 180-185) seems to be relatively close in time
to SPRA. It has one of the earliest Pural)ic descriptions of the cremation
ground (smasiina), treated as holy ground and identified with Avimu-
kta,56 which is a feature shared with SPRA. While the cremation ground
is conspicuously absent in the eulogy of Varal!asl in SPs 139-162, the
eulogy in SPRA 4.1-9 is entirely devoted to VaraI!asl's cremation ground
Avimuktaka. 57 Interestingly, both texts share a common verse in praise
of VaraI!asl (SPRA 4.9 MtP 185.38).58
The DevlpuralJa (DeP 63) and the NagarakllalJ<ja of the SkandapuralJa
(SkP NagarakllalJ<ja 109) contain a list of sixty-eight Siva sites together
with the particular name of Siva at each site. Although there is little
evidence to date this material, some of its features look relatively archa-
ic. 59 Sanderson (2003-04, p. 407f), who only refers to the
version) observes:
55 Hazra (1987, p. 73) dates the section on sraddha (MtP 16-22) to ca. 850 AD.
56 MtP 184.19: smaSanam iti vikhylitam avimuktarp, iivalayam I tad guhyaT(t
devadevasya tat tfrtha7[t tat tapovanam II.
57 Cf. also Bisschop 2002 and SP Vol. IIA, pp. 288-292.
58 The Vara{lasfmalultmya of the MtP also has a few verses (MtP which
show familiarity with the paiicii#aka (on which see below, pp.27ff.).
59 On rather weak grounds, Hazra (1963, p. 94f) dates the present to
the sixth century AD, which seems much too early. None of the Dharmanibandha
authors listed by Hazra (pp. 92-94) quote from the sixty-third chapter. Parts of
the Nagaraklla1).Qa must have been in existence at least by the twelfth century AD
(Hazra 1987, p. 165).
60 The relevant passage from the Picumata (f. is quoted by Sanderson at
p. 407, n. 208; refers to the list in the Svaccllandoddyota, his commen-
tary on the Svaccllandatantra (Sanderson p. 408, n. 209).
Other Sources on Saiva Topography 25
25 Siva
26 DaIfqaka / Devadaruvana Dal)q.in / Dil).c;li
27 / Urdhvareta(s)
28 Krmijangala / rVladhyamajangala /
29 Ekamra / Ekagra Krttivasa
30 Chagaleya / Sukalanta Kapardin
31 Kalaiijara NIlakal).tha
32 MaIf9-alesvara / Mal).9-akesvara SrfkaIftha
33 KasmIra Vijaya
34 Marukesvara Jayanta
35 Hariscandra Hara / Hari
36 Puraiicandra / Puriscandra Sailkara
37 VameSvara / Ramesvara Jatin
38 Kukkutesvara / Kukkutakesvara Saumya
39 Bhasmagatra / Bhutesvara Bhasmagotra / Bhutesvara
40 Orpkara / Amarakal).thaka AmarakaI).taka / Orpkara
41 Trisandhya / PaScimasandhya Tryambaka / Tamra
42 Viraja Trilocana
43 Arkesvara / Dlptacakresvara DIpta / Veda
44 Nepala PaSupalaka / PaSupati
45 / - Yamalinga / -
46 Karavlraka Kapalin
47 Jagesvara / T!'ptesvara TrisUlin
48 SrTsaila 'n'ipurantaka
49 Ayodhya Roha1).a
50 Pa:tala Hatakesvara
51 KarohaI).a / Kayarohal).a(?) 65 NakulIsa / Laku\i
52 Devika / Vedlka 66 Umapati
53 Bhairava Bhairavakara
54 Purvasagara / Gangasagara Amara / Amaya
55 Saptagodavara BhIma
56 Nirmalesvara / Nakulesvara Svayambhu
57 Karl).ikara
58 Kailasa / - / -
59 Gangadvara / Himasthana / -
60 Jalalinga Jalapriya / Jalesvara
61 Vac)avagni / - Anala /-
62 Badarikasrama / - Bhlma /-
63 / - KotIsvara / -
64 Vindhyaparvata Varaha
65 Hemaku\a /-
66 Gandhamadana BhUrbhuva(s)
67 Lingesvara / - Varada /-
68 Lanka Narantaka
table 3: sixty-eight Siva sites
At least thirty-two places of this list are included among the iiyatanas of
SPs, the majority of them belonging to the first thirty: nos. 1-7, 10-11,
65 DeP 63.14c reads srikii:rii roha'(te lakutim which is hypermetrical and non-sensical.
66 Variant reading devikiiyiim instead of vedfkiiyiim.
Other Sources on Saiva Topography 27
THE PANCA.$TAKA
67 For recent discussions of the ogdoad doctrine see Goodall 2004, pp. n. 620,
and Sanderson 2003-04, pp. 403-406. Cf. also Brunner-Lachaux 1977, pp. 299-303
and plate VII. In translating a$taka with 'ogdoad' I follow Goodall.
68 The other ones rather seem to bear the names of particular Rudras or forms of
Siva. cr. the names on the lists reproduced in plate VII in Brunner-Lachaux 1977.
69 The Sivadharma corpus is a collection of texts intended for the lay Saivas. Various
texts of this kind exist, of which only the Siva- Upani$ad has so far been published.
Two early Nepalese palm-leaf manuscripts consist of a group of such texts, the most
important of which are the Sivadbarma and the Sivadbarmottara: cf. Goodall 1995,
p. 375f, n. 616, and Sanderson forthc. The Sivadl1arma and the Sivadbarmotta-
ra have been summarized by Hazra (19S3 and 19S5). In a recent article Magnone
argues for a southern origin of the Sivadharmottara and argues vehemently against
Hazra's eighth century dating of the text, himself advocating a twelfth or post
twelfth century date (Magnone 2005, pp. 58S-591). This is certainly wrong, as the
existence of a ca. ninth century, fragmented palm-leaf manuscript (NAK MS 5-
S92, NGMPP Reel No. A 12/3) of the Sivadl1armottara written in Licchavi script
28 Introduction
74 Srlparvata - rather than SrIsaila - is the name found in the texts transmitted
in the oldest manuscripts and also the name of the sanctuary in SPs 96.
Other Sources on Saiva Topography 31
75 Although the name Bhastrapada occurs in only one of the sources of the paficii-
$taka (the Sivadharma) it is likely that this was in fact the original name and that
Vastrapada, the name found in most other sources, is an early corruption. This is
suggested by the combined evidence of the Sivadharma, the SkandapuraI)a and a
passage from Bhojadeva's Siddhantasarapaddhati: see the annotation ad SPs 54c.
Note that the Nisvasamukha and the Nisvasaguhya vary between Vastrapada and
Bhadrapada.
32 Introduction
• Mahalaya KiVP, MatVP, MrgVP, NiM, RauKP, SiDh, SoSP, SvSuS, SvT,
TSa
Maha:kllla MaViT; Mahabala NiG
• Gokan;a KiVP, MiWiT, MatVP, MrgVP, NiG, RauKP, SiDh, SoSP, SvSuS,
SvT 1 TSa; illegible NiM
• Bhadrakan;a KiVP, MaViT, MatVP, MrgVP, NiM, RauKP, SiDh, SoSP,
SvSuS, SvT, TSa; illegible NiG
• Suvarl!akf?a MatVP, SiDh; KiVP, lvDiViT, MmVP, NiM, RauKP,
8oSP, SvT; illegible NiG; SuvarI).a SvSuS, TSa
• Sthal,lu KiVP, MaViT, MatVP, MrgVP, NiM, RauKP, SoSP, SvSuS, SvT,
TSa
illegible NiG; SthaI).vfsvara SiDh
It can be observed from the list above that there is considerable variation
in the names of the forty sites as in the names given to the individu-
al ogdoads. In general, I have given preference to the presumably older
sources 76 or the ones that are transmitted in old Nepalese palm-leaf manu-
scripts.
One of the most striking features of the paiicii1;Jtaka list is the inclusion
of the names of the four incarnations of Siva at KarohaJ.1a, as narrated in
SPs 115 If., viz. Bharabhilti, DiJ.1<;1imuJ.1<;1a, and Lagu<;li (Lakuli).
These are the only names among the paiicii1;Jtaka that are not so much
toponyms as personal names. 77 The four do not occur together as a group
in any other source, and it is conceivable therefore that these names have
in fact been adopted from SPs. 78 If the list was indeed already current
in or soon after the sixth century, as Sanderson has suggested (d. above,
p. 28), this would strengthen the hypothesis that the original Skanda-
puri4la was composed in the sixth century. Alternatively, both traditions
may go back to an earlier source. It seems significant that KarohaJ.1a,
evidently an important toponym in early Saivism, is absent in these lists.
For a discussion of particular names of the paiicii1;Jtaka occurring in SPs I
refer the reader to the annotation on the relevant passages in SPs.
The distribution among the five ogdoads of the names of sites not
mentioned in the iiyatana account of the Skandapuralfa is unequal. Of
the pavitrii1;Jtaka only Amaresa is absent in SPs, and among the guhyiiti-
guhyii1;Jtaka only Jalpa is not mentioned. Moreover, it is very likely that the
latter is identical with or an early corruption of Japyesvara, a sanctuary
mentioned in the iiyatana account (see the annotation ad SPs 62a). If
so, among the first two ogdoads only Amaresa is absent.79 Among the
guhyii1;Jtaka Nakhala, Vimala, Attahasa and Bhlma do not occur in SPs.
Of these the name Nakhala, which is not known from other sources, gives
the impression of being a mere alliteration of Kanakhala, with which it is
always found juxtaposed. The names included among the pavitrii1;Jtaka are
all mentioned in SPs. Of the five names of the sthii7Jva1;Jtaka that are not
mentioned in SPs - Dural]<;Ia, Makota, Mal]<;Ialesvara, Sthillesvara and
Sthalesvara - two occur in other chapters of the Skandapuralfa. One is
Mal]c;lalesvara on the Vindhya mountain (see above, p. 19), the other one,
Sthillesvara, is identified with Sailkukarl]a in SPBh 73.62, at the confluence
of the Sindhu and the ocean (ef. annotation ad SPs 177e). Note, however,
that Sailkukarl]a is included as a separate name among the same ogdoad
as Sthillesvara.
Sanderson (2003-04, p. 405, n. 201) has suggested that six of the pa-
vit.,.ii#aka, the first ogdoad, namely Bharabhuti, Dil]c;limul]c;li,
Lakuli, Amaresa and Prabhasa, were Pasupata strongholds. The first four
because they are the four sites at Karohal]a near the Narmada, where Si-
va incarnated in the four respective yugas (SPs 112 ff.) and the last two
because of inscriptions surviving at these two sites. It seems that we can
go one step further and conclude that in fact most, if not all, of the forty
sites listed belonged to the Pasupata tradition. Thus, for example, Avi-
mukta is evidently an important early Pasupata site, and is described as
such in the SP's Viil'iilJ8Sfmiihii:tmya (SP 26-31.14).80 Mahalaya, heading
the list of mundane sanctuaries in SPs, was probably a Pasupata site, as
its occurrence in the list of avatiims preceding Lakulfsa 81 suggests; other
descriptions of this site confirm this impression (ef. annotation ad SPs
30b and SPRA 2.6). According to the miihiitmya of Srfparvata told in
SP Bh 70.42 ff., yogasiddhas are practising Pasupata yoga at SrTparvata.
The SkandapuriilJa and the Mahiibhiil'ata exhort one to bathe in ashes
at the sanctuary G,dhrakutesvara or G,dhravata in Gaya (see ad SPs
167). All of these references point to a Pasupata background of these
places.82 In addition the inclusion of the name Vimala could indicate a
connection with the mysterious Vaimalas, an early group of Pasupatas
known La. from the Svacchandatantra (SvT 11.72)83
CAMBODIAN INSCRIPTIONS
Another important source for the study of early Saiva topography consists
of the epigraphic records of the Khmers from ancient Cambodia (Kambu-
jadesa).84 Among these records those dating from the pre-Angkorean
period, Le. before the ninth century, are evidently the most crucial. They
present a picture of early India which in India itself has sometimes become
6 N aimisesvara; 92
7 (or 93
9 Prahasi tesvara; 95
10 Bhadresvara; 96
11 Rudramahalaya; 97
12 -trapadesvara (= Bhastrapadesvara?);98
The Pasupatas
100 The Picumata gives a list of sixty-eight Rudras of the nine cremation grounds
of the inititation MaDqala; one of the Rudras is called Prahasita. Cf. Sanderson
2003-04, p. 407£, n. 208, quoting the relevant passage from the Picumata's earliest
surviving manuscript, dated 1052 AD. For a short description of the contents of
the Picumata see Goudriaan & Gupta 1981, pp. 42-44. The names of sixty-eight
Rudras are for the major part identical with the names of Siva at the sixty-eight
Siva sites taught in SkP Nagarakha.r;J(;ia 108-109 and DeP 63. See above, p.24.
101 TVK p. 89, 1. 17 (quoting a 'LiilgapuraQa'), SkP KiiSfkhaQ<;Ia 2.85.22 and SkP
KasTkhaQ<;Ia 2.97.167.
38 Introduction
The study of the Pasupata cult has received a great impulse in the first half
of the 20th century with the publication of two of its principal texts: the
GalJaicariica with the Ratnatlica attributed to Bhasarvajiia, and the Pasu-
patasutra with the Paiicarthabha?ya by KaUl).c;linya. Prior to the publica-
tion of these two texts the only available textual source for the doctrine of
the Pasupatas was the sixth chapter (Naicullsapasupatadarsana) of Ma-
dhava's 14th century Sarvadarsanasarpgraha. 103 The credit of disclosing
this material goes, to a large extent, to Minoru Hara, who has published
several articles dealing with various aspects of the doctrine, continuing the
subject of his *1966 Harvard thesis 'Materials for the study of Pasupata
Saivism.' 104 At the end of the introduction to this thesis Hara makes a
remark which is relevant for the study of the Pasupatas in the PuralJas:
Often we meet the term pasupaia-yoga in Pural).ic literature. The
meaning assigned to it, however, is not that of the original Pasupata
scriptures. The pasupaia-yoga of Pural).ic literature is a system of
disciplines such as breath-control, sitting postures, etc., and some-
times a group of supernatural powers. The yoga as defined in the
Pasupata Sutra and Kaul).c;linya's Bhii?ya, on the contrary, is the
union of the individual soul with God, and this original concept
seems to be foreign to the Pural).ic literature. 105
When we consider the use of the term yoga and more particularly Pa-
supata yoga in the SkandapuralJa, this remark for the greater part holds
true. The final section of the SkandapuralJa (SP Bh 174-183), which is
devoted to PMupata yoga, indeed contains quite a number of descriptions
of various yogic practices. However, especially in SPBh 180 we come across
statements that yoga is not only something to practise, but also something
which is given by Siva, or which can be attained by the initiate.106 Such
a usage of the term yoga comes relatively close to the yoga of the Pa-
iiupatasiitra and its commentary. On the other hand, one of the most
peculiar practices described in the Sutras, namely the acting in such a
way as to court dishonour (avamana) in the second stage of the aspirant's
ascetic career,107 is conspicuously absent in the last ten chapters of the
SkandapuralJa. 108 An earlier passage, in which Siva is teaching the gods
the PMupata vrata (SPBh 122.81-84ab), has a verbal parallel to one of
the Sutras concerned with this peculiar practice,109 but significantly the
text does not mention how this dishonour is to be brought about by the
ascetic, as in the case of the PiiSupatasiitra. Rather dishonour seems to
be a kind of inner virtue pertaining to the PMupata ascetic as such, and
it has nothing to do with the particular ritual practice in the second stage
of the PMupata's career. One likewise looks in vain for this practice in
the sections in other Pura1Jas that purport to give a teaching of PMupata
yoga. llO This may reflect a difference between the esoteric doctrine of the
PiiSupatasiitra and its commentary, on the one hand, and the exoteric
aspects of the teaching in the Pura1Jas on the other.
106 Cf. e.g. SPBh 180.4 8vaT'{t yoga1Jt pradadau te§arp, tadti deva umtipatily, I sarviisa'f[L
mok§avidyanii:T}t vat tattvam atiricyate IIi SP Bh 180.13 dharmadharmaparityaktas
caret paBupataT'{t vratam! tato yogam ima'T[/> prapya mahad aiSvaryam iipnute II;
SP Bh 180.22cd-23ab dzk§ito brahmabhasmabhyii:rJ1. iivayiijf munir bhavet II sfghram
eva paraT'{t yogarp. prlipya mucyeta bandhaniit I. Cf. also e.g. SPs 125c and SP s
129c.
107 Cf. Ingalls 1962.
108 The anukrama'l}ika of the A recension adds a hemistich - thought to refer to
the PMupata yoga section (SP Vol. I, p. 55 and p. 64, n. 10) - of which the
first pada actually suggests a reference to the practice of courting dishonour: kiNe
viriJ,pakara1Jarr- yogasya ca paro vidhilJ, 1 'the putting on of a deformed figure (dis-
guise) in an act of deception and the supreme procedure of yoga.' There is, how-
ever, nothing in the last ten chapters that corresponds to this.
109 SPBh 122.81-84ab: yantrii'T}-ii'f{L parama'f{L yantram anyad yasmiin na vidyate 1
$a¢ariga1J1- sarvakiimfya1J1- sarvalokanamaskrtam II suci sarvapavitra1J1- ca sarvapiipii-
paha1J1- subham 1 tapaso vardhana'f}b caiva te carantu samiihitii/:t 1 yantra'f}b tv apiipa1J1-
vijiieya1J1- prakrtil;t sii ca sa1J1-smrtii II bandhas ca trividhaly, sanyiisaphalam
eva ca 1 tasmiit sarvaprahii1)iirtha'f}b yantram etat samiicaret II asa'f{Lmataly. sadii
loke yenajrryen na karhicit I. The reference to the 'supreme stratagem' (parama1J1-
yantram) calls to mind PilSfi 4.9: asa'f}bmiino hi yantrii1Jii1J1- sarve$iim uttamaly,
smrtaly,.
110 E.g. VaP 1.11-20, LiP 1.88-91 and KiiP 2.11.
40 Introduction
111 For a description of the Pa8upata vrata see especially SF Bh 180. This chapter
shows some similarities with LiP 1.34. Compare e.g. SPEh 180.18 prabhavanti
sura!y, somat pitaro vahnisaTJlbhavli?t I vahnisomiitmakarp. tasmiij jagat sarvaTJl prati-
$phitam II and LiP 1.34.6 fi$mapii?L pitaTO jiieya deVil vai somasafJlbhavii/:t I agnf-
$omiitmakarp, sarvaTf1 jagat sthavarajangamam II.
112 Cf. also SP 29.62d, where the Fasupatas in VaraI)-asT are described as 'having
a white radiance due to smearing [their bodies] with ashes' (bhasmiibhyarigasita-
prabhiiM·
113 '0 Kalakarl)l turn back, do not strike the best of the gods! They have become
Pasupatas, with their bodies smeared with ashes.' ktilakaT'f/,i nivartasva ma vadhi{t
surasattaman I ete pasupatfbhuta bhasmana digdhamurtayalJ II (SP Bh 32.115). The
word rtisi in SP Bh 32.1l3c may be significant. It could be a cryptic reference to
Pasupata initation at which one would bathe in ashes and receive a new name
ending in -rtisi. Many Pasupatas with names ending in -rtisi are known from
inscriptional records: see n. 259 on p.204 below.
114 SPs l30ef (see below) makes explicit that the four pupils have received initiation.
A central element of the Pasupata is the application of ashes: d. e.g. SPBh
180.22.
The PMupatas 41
The designation 'Pancartha' is significant, for this is the very name given
to the PMupata doctrine in KaUJ:'H;linya's PaiicarthabhEi{lya (cf. annotation
ad SPs 130b), and indicates familiarity with the Pancartha tradition.
Moreover this is the only place in this part of the text where the name
Lagw;li, viz. Lakulfsa, appears.115 It is noteworthy that the term Panca-
rtha as well as the incarnation's name have been left out of the parallel
version in SPRA 5.87 (see annotation ad loc.). The closest connection
between our text and the Paiicarthabha{lya is the story about the Lord's
descent (avatiira) as a teacher in Karohal,la (PBh: Kayavatara) and his
acceptance of Kusika as his first pupil at the cremation ground (smasiina)
of Ujjain. To this subject we will now turn our attention. In the following
paragraph I discuss some related material about Siva's descent as LakulIsa
at Karohal,la in other Pural,las.
115 Lakulfsa is mentioned a second time in this chapter, in the additional list, in SP s
169d (LagUt;lfsvara).
116 On these lists see also Meinhard 1928, pp. 20-34, and Dviveda 1982. A number
of manuscripts of the Gayamahatmya incorporate a similar list in place of Ga-
yamahatmya 2.33-34, in which a group of mental priests (miinasiin rtvija/:l-) is
enumerated. Jacques (1962, p. 42, n. 61) remarks: "Est-ce un souvenir d'une
influence pa§upata it Gaya?" Another, unpublished, source is the Jayadratha-
yamala. In it the list of twenty-eight avatiiras is grouped together with a list
of thirty-eight Rudras, together constituting sixty-six embodiments. See Bakker
2000, p. 11, quoting from an unpublished edition by Sanderson.
117 The following are the names of the sons or pupils of the twenty-eight incarnations
according to YaP 1.23 (variants in LiP 1.24 in parentheses): 1) Sveta, [Sveta-jsikha,
42 Introduction
27 SOffiaSarman Prabhasa
28 Nakulin / Lakulin Kayarohal}a / Kayavatara
118 Cf. e.g. the phrase bhasmasniiniinulepanli!y. in YaP 1.23.145b. The yoga prac-
tised by the sons of the twenty-eight incarnations is described as miihesvara yoga
(132a, 142c, 174c, 178c, 182c, 190a, 193e, 199c, 201e, 205c, 213c). The phrase
punaravrttidurlabha, a stock-expression in this chapter, qualifying either rudraloka
or matsamfpa (61d, 67d, 88d, 110d, 114d, 117d, 123d, 132d, 213d), likewise seems
to be characteristic of the Pasupata tradition: cf. Bisschop & Griffiths 2003, p. 337,
n.113.
119 Cf. SP Vol. I, pp. 20-25, and SP Vol. IIA, pp. 8-9, on the intertextual relationship
of the Pural,lic text corpus.
120 Note that the list has no parallel in the BrabmaIJ-gapuraIJ-a, which is an indication
that it may be a later addition to the text. Cf. Kirfel 1927, pp. XII-XIV.
44 Introduction
Dilf<;limUl,l<;la cut off the heads - no victims specified - in the Treta yu-
ga, and 3) granted his favour by dancing in the Dvapara yuga (SPs
115-116).
The part about Siva's three previous incarnations at Karohal/a is not
known from other sources, with the exception of the Karava1,lamahatmya.
This text teaches that Siva descends in Karohalfa every yuga, yet the
names given there are different and no events are linked with these pre-
vious incarnations (d. annotation ad SPs 117a). The list of twenty-eight
incarnations, on the other hand, does seem to presuppose some of these
names. The 25th incarnation is a certain Dal/<;lin MUl/<;llsvara, which is
suggestive of Dil/<;limulf<;li. 121 The 27th incarnation is SomaSarman in Pra-
bhasa, in whose house the Lord takes birth according to SPs 124. Possibly,
this SomaSarman is somehow linked to for SomaSarman is listed
in SPBh 131.2ab together with the Galfapas Dilf<;li and Bhiirabhuti, where
one would expect 122 and is the avatiira preceding Lakullsa
in Karohalfa according to SPs 116cd. In the Jayadl'athayamala, where
the twenty-eight form the first group of a total of sixty-six Rudras, the
name is in fact included among the last four incarnations, the
other ones being Tridalf<;li, Somesa and Lakullsa. '23 Finally, Lakullsa, the
28th incarnation, is identical with Siva's incarnation in the Kali yuga in
Karohalfa according to the Skandapura1,la (Lagu<;li). Thus, three of the
four incarnations at Karohalfa in the Skandapura1,la seem to be included
among the last of the twenty-eight incarnations of the Vayu and other
Puralfas.
The names of the four pupils of LakullSa, whose name is only mentioned
once in this passage (SPs 129d: liiguifiJ;.), are in agreement with the names
of the pupils in the list of twenty-eight avatiiras. According to SPs 122-
123 their names are Kausika, Gargya, Mitra and a fourth, anonymous
one, said to be a brahmaciirin from the country of the Kurus. He may be
121 Instead of Dal).<;lin MUl).Q.lsvara (LiP 1.7.34b, SiP Satarudrasaiphita 5.37, YaP
1.23.197: da1)¢f mU1)dfSvaraM, KuP 1.53.9b has Dil)<;lin lvIul)<;lin (di1)df mU1)df),
while two early manuscripts of the Vayupura.Qa (Vll V 2) have DaI).QimuI).qIsvara
(darru)imu'T[L9-fsvara) instead (Vl folio 54r, l. 8j V 2 folio 43v, L 8). The name is
clearly liable to alternation. Cf. also Dviveda 1982, p. 9: "itfmiini piithiintarii'(l.Y
atra drsyantel di'l:uJ,imur:uJ-ir itijalzyaT[L bhuvanaTJ1- svacchandatantre (101854) I iena
di'[tif,imu7!4fsa ity eva nama pradhiinaiveniiira sthiipyatel. '1 The passage from the
Svacchandatantra to which Dviveda refers concerns the see p.29
above.
122 Cf. SP Vol. I, p. 102, n. 137.
123 JRY 4.452d in an unpublished edition by Prof. Sanderson.
The Pasupatas 45
124 Cf. also SP Vol. IIA, p. 30, n. 100: "The doctrine of a guru LakulIsa and his
four pupils seems to be an example of 'invention of tradition' (Gupta period), in
order to account for several guru lineages that all claimed to go back to Siva's
incarnation, an avatiira who since then received the name L[kulin I LakulIsa.)1
125 With the exception of the descent in a brahmin body this account is strikingly
similar to SFS where it is said that the Lord went to UjjayanI, entered
a cremation ground, covered himself with ashes, took a firebrand in his left hand
and sat down to initiate his first pupil Kausika. See annotation ad loe. The
episode of LakulIsa)s descent in the body of a (dead) brahmin is known from many
other sources) in particular the account of the 28th incarnation in the avatiira lists
mentioned above. See also Lorenzen 1991) p. 176.
126 Cf. also PBh p. 3, II. 13-15 (Kausika) and PBh p. 102, I. 5 (Kusika). For a new
interpretation of the latter passage see below p. 49.
127 This was already noted by Schultz (1958, p. 6f).
46 Introduction
was the first to suggest that this Knsika is identical with the pupil of'
Lakulfsa. '28 On the basis of this inscription, which mentions among oth-
er things that one Uditacarya was tenth in line from Bhagavat Kusika,
Lakulfsa has been dated to the second century AD. However, the name
Lakulfsa does not appear in this inscription and there is no evidence that
the notion of Lakulfsa as an incarnation of Siva existed at the time. In
fact I am unaware of any attestation of the name Lakulfsa, or a variant
of that name, preceding the SkandapuraJ.la (SPBh 166.25b, SP Bh 166.29a,
SPs 129d, SPs 169d).'29 The earliest known images of Lakulfsa date from
about the same period (ca. the sixth century). 130 The term Pasupata does
not occur in the Mathura inscription either: the group of devotees men-
tioned are Mahesvaras (I. 11). We do, however, have epigraphic evidence
for the existence of the Pasupatas by the time of Samudragupta (ca. 335-
376 AD), for Pasupatacaryas are mentioned as the recipients of grants in
the copper plates of the mahi"iraja Bhulu,)<;la of the Valkhas, who was a
contemporary of Samudragupta. 131
128 Cf. also Sircar 1965, p. 277f. This identification has been accepted by many subse-
quent scholars, but not unanimously. Thus e.g. V.S. Pathak (1960, p. 9, ll. 3) rather
wants to identify him with Kusika II (Aparakusika) in the list of eighteen avaUiras
transmitted in Jaina sources (on which see below, p.49). Cf. also Lorenzen 1991,
p. 180f.
129 The first two instances occur in a passage explaining Siva's different names. He is
called Uiguqin (8 1 ; lakulf 8 2 ) because he carries the lagur.j,a (8 1 ; lakula 8 2 ), just as
he is said to be Bulin because he carries the silla (166.25a). One who praises him by
this name is said to attain supreme worlds (166.29ab). Only the two instances from
SPs clearly connect this name with Siva as the teacher of the Pasupata doctrine.
130 Cf. Shah 1984, p. 97 (with plates 81 and 82), referring to two images from Mathu-
ra. It is noteworthy that in one of these two images (pI. 81) LakulTsa is not flanked
by four pupils, as is the standard in later iconography, but only by two. Does
this reflect an early tradition in which the doctrine of the four pupils was not yet
fixed? LK. Sarma (1982, p. 84f) has identified a four-faced squattish sculptured
pillar from Mukhalingam, tentatively dated to the third-fourth century AD, as a
representation of LakulTsa, but the iconography is rare and from the photographs
supplied (pI. 50-53) it does not look like the figure really has a lakuta in his
left hand, as Sarma suggests. He is himself in doubt about the identity of the
object in the right hand: "It could be a Matulunga fruit or even a liriga. We
cannot be certain since the object is broken." (p. 95, n. 62). The main figure
is oriented towards the east, two-armed, seated in a cross-legged yogic posture,
has a yogapatta bound around his knees and has a rosary in his right hand. The
image is interesting for other reasons: although much more crudely carved, the
joint hairdress of the four heads, bound together across the cranium, looks very
similar to that of two four-faced Vakataka images from Mandhal, dated to the
last quarter of the fourth century, identified repectively as Mahesvara and NandT-
svara by Bakker (1997, p. 95f, plates I, II, III and IV). These images seem to be
iconographically related. Cf. also Brown 2004, p. 67f.
131 These copper plates were found at Bagh in Madhya Pradesh, and have been pub-
lished by Ramesh & Tewari 1990. Pasupatas are mentioned in plate nrs. III, V,
VI, IX, X, XII and XIV. Significantly, no mention is made of LakulTsa in these
The PiiSupatas 47
Although this is not the place to reconsider the identity of the line of
Saiva teachers of the Mathura inscription, 132 I would like to mention here
an additional epigraphic reference which could be important for the recon-
struction of the earliest phase of PiiSupata history. It concerns a Prakrit
inscription from Pac;lana hill, about eleven miles north of Bombay, six miles
west of Kanheri, which mentions a saint called Kosikaya (Kausikeya). 133
Possibly, he is identical or affiliated with the siddha called Musala men-
tioned in three other inscriptions at the same site. l34 The name Musala
is known from the Pasupata tradition as well: in his commen-
tary on the SvacclJandatantra distinguishes between Lakula PiiSupatas,
whose tradition had been founded by Lakullsa (kiiroha7}asthCiniivati'l"(!a) ,
and Mausula Pasupatas in the line of a pupil of Lakullsa, called Musula
or Musulendra. 135 The site where these inscriptions are engraved is said
to be the residence of Siddhas in another inscription,136 which apparently
refers to a group of ascetics living there.
The two inscriptions mentioning Kosikaya and the residence of Siddhas
are dated by Indraji to the first century AD, the ones referring to Mu-
sala to the third century AD. These inscriptions suggest the possibility
that Kausika or Musala was in fact an early Brahmanical sage who was
later incorporated in the Pasupata tradition by making him the pupil of
Lakullsa. Something similar seems to have happened with SomaSarman,
who was likewise a predecessor of Lakullsa but has been made into the
first recipient of his grace by the Skandapura'!a authors. 137 It is curious at
least that the line of teachers mentioned in the Mathura inscription does
inscriptions either.
132 See Acharya ferthe., for a new reading and interpretation of the Mathura
inscription.
133 Indraji 1881-82, p. 322, Inser. B kosikayasa udao tirtimo ca. Indraji translates this
as "And the eastern pleasure seat of Kosikaya. '1 (Skt. kausikeyasya udaya iiriimas
ca). Prof. Bakker first drew my attention to this inscription and suggested the
possibility of connecting the name of this person with that of the pupil of Laku-
lIsa. When I inquired at the Department of :lvluseurns & Archaeology in Bombay
for the present state of this site (December 2003), I was told by Dr. B.V. Kulkarni
that it has been turned into a stone-mine.
134 Indraji 1881-82, pp. 323-325, Inscr. E sadhamusala (Skt. siddhamusalafy" assuming
that sadha is a scribal error for sidha), and F and I: musaladatta.
135 ad SvT 11.71cd. Cf. Bakker 2000, p. 4f. Sanderson (2002, p. 30. n. 32
[4]) quotes a passage from the * Jayadrathayamala according to which LakulIsa
(laf.,,'"Ulapa1}i) taught his own pupil Musallndra the Svaccllanda and Musallndra in
his turn extracted and taught its essence. The last remark could imply that he
was held to be the author of the Svacchandasara, the name of a Svacchanda work
known from a list of 24 Tantras in the SrTka1)tbTya (ibidem, p. 20, n. 20).
136 Indraji 1881-82, p. 323, Inscr. D parvato abhu7ftto sidhavasati "The moun-
tain, the residence of Siddhas (monks) all about." (Skt. parvato 'bhyantafy, (7)
siddhavasatify,). Translation by Indraji.
137 Cf. Bakker 2000, p. 13f.
48 Introduction
not go back to Bhagavat Lakullsa himself but stops with Bhagavat Kusi-
ka. Given the fact that the earliest evidence for the name Lakullsa stems
from at least two centuries later it seems more likely that these Saiva
teachers were in fact followers of a saint Kusika, who was not yet held
to be the pupil of LakulIsa. '38 All this suggests that the only probable
historical figure of the earliest stage of the PMupata history is Kusika,
while the identification of his teacher - 'the Lord' (bhagavant) according
to Kau'!<;linya's - as Lakullsa was established afterwards and is
therefore a later invention. 139
The name Kusika is known from three other inscriptions, in which the
descent of Lakullsa in Ka(yava)roha,!a is mentioned: 1) the Eklingji Stone
Inscription dated Vikrama Sal]1vat 1028 = ca. CE 971 (Bhandarkar 1904-
07); 2) the Pa)<;ll Inscription of Guhila Arisil]1ha dated Vikrama Sal]1vat
1173 = ca. 1116 AD (EI 30 [1933-34], pp. 8-12); 3) the Cintra Prasasti
of the reign of Sarangadeva, dated Vikrama Sal]1vat 1343 = ca. 1287 AD
(Ell [1892]' pp. 271-287). These inscriptions refer to the story of Laku-
lIsa's descent and his four pupils, known from the Pura,!as and date from
a much later period than the Mathura inscription. Concerning the other
three pupils of Lakullsa, we only hear about a lineage of teachers who
trace their origin back to Gargya, from the twelfth century onwards. 140
No lineages of the third and fourth pupil of Lakullsa are known from
epigraphic sources. 141
Although epigraphic sources do not help us much in reconstructing the
supposed lineages of the four pupils of Lakullsa there is a literary reference
138 About the exact identity of these Miihesvaras we know next to nothing, except
that the names of two of the gurus mentioned in the inscription, Kapilavimala
and Upamitavimala, end in -vimala, which suggests a connection with the Vaima-
las, a division of PMupatas known i.a. from the Svacchandatantra (SvT 11.72).
Cf. Bakker 2000, p. 6.
139 The view that Lakulfsa was a real, historical figure is widespread. Cf. e,g. Bhan-
darkar 1913 ("Fl'om all this it appears that there lived a certain person of the name
of Lakulin (the holder of a lakuta, or lagu<;la, or lakula, i.e. a club) who founded a
PMupata system."), Hara 1958, p. 9f ("He seems to have lived about the beginning
of the Christian era.") and Lorenzen 1991, p. 175 ("LakulIsa was in alllikelyhood
the founder of the Pasupata order.").
140 This lineage was based in Prabhasapatan (Somanatha). The earliest inscription in
which this tradition is recorded is the Somnathpattan Praiiasti of Bhava Brhaspati,
dated ValabhTSaqlVat 850 = ca. CE 1169 (Ozha 1889; Peterson 1895). Cf. Bisschop
& Griffiths 2003, p. 321£, n. 33.
141 The attempts to connect Kauru9ya, the fourth pupil, with the Kalamukhas, which
are still sometimes encountered in secondary literature (e.g. Choubey 1997, p. 87-
88) are not based on facts: cf. Lorenzen 1991, p. 182. Choubey's statement that
in the Tadikonda (sic!) inscription "the KaJamukhas are said to have descended
from the third pupil of LakulIsvara by the name Kauru9a or Kalana," cannot be
corroborated: no such name is mentioned in the Tal).r;likOl)-qa grant of Ammaraja
II, dated Saka 880 = ca. 959 AD (EI 23 [1935-36], pp. 161-170).
The Pasupatas 49
to the four lineages that has not been noticed so far. In his commentary on
PaSu 4.10 (indro vii agre asure§u pasupatam iicarat) KaUl;l<;linya comments
upon the word agre as follows:
atriigra iti purvakiilam adhikurute I kusikeiiinasambandhiit priik I
prathamam amarais cf",am I krtatretiidviipariidi§u yuge§v ity
arthal;!l
Here 'in the beginning' denotes the past: before the lineage of Kusi-
ka and Isana. First it was practised by the immortals: in the yugas
beginning with K,ta, T\.·eta and Dvapara.
142 The relevant passage from the $ar;ldarsanasamuccaya is printed as the 8th
parisi§ta in the GOS edition of Candrananda's Vrtti on the VaiSe$ikasiitra
(pp. 220-224 [not pp. 120-124, as erronously reported in Bisschop 2005, p. 547,
n. 107]): te?li:rp. ea sarikaro devaf:t sr#isarp,harakiirakaf:t I tasyiivatiiTiiJ;, sara ye
te '$ttidaia II te§iiT[t nfimiiny atha bromo nakulfso 'tha kausikaJ;, I
giirgyo maitryaly, kaUTU$aS ea fsiina!y, $a§tha ucyate II saptamaly, paragiirgyas tu
I aparakusiko 'tris ea pirigaliik$o 'tha pU$pakaJ;.1I brhadticiiryo
'gastis ea santlina!y, $or,lasaf:i, smrtalJ I riisfkaraly, saptadaso vidyiigurur athiiparaly, II
ete '$tiidasa tfrtheiiis sevyante pade pade I (p. 222, vss. 90-94b). The pas-
sage from GUI).aratna's Tarkarahasyadfpika is printed as Appendix II in the
GOS edition of the RatnatTka, where it is erronously attributed to Haribhadra's
$aq.dadanasamuccaya, on which it is a commentary. It has Kusika instead of
Aparakusika, Pingali instead of PingaHikr;;a and Brhadarya instead of B:rhadacarya.
143 This interpretation is based on a joint reading of the Paiiciirthabha?ya with Prof.
Bakker. Cf. also SP Vol. IIA, pp. 29-30, n. 100. A passage added after the
final colophon of a newly identified manuscript of the Paiiclirthabhi!i:?ya from
Benares has a similar list, to which are added the five ktirar:ta deities SadMiva,
Isvara, Rudra, Vi?l).u and Brahma, and the names of six successive heads of the
Mal).ik(arl).ik)esvara Matha in Benares: see Bisschop 2005, pp. 545-549, for details.
The reconstruction of four lineages proposed here receives some support from this
passage.
50 Introduction
Editorial Principles
A reader familiar with the editorial principles and presentation of the text
adopted in SP Vol. I and IIA cannot fail to notice a strong difference
of approach in the two editions printed here. In SP Vol. I and IIA the
editors have constituted a single text on the basis of the three available
recensions, with the important restriction that it is "in principle the recen-
sion of the SP transmitted in the Nepalese manuscripts" (p. 41) that is
edited. '47 The main critical apparatus in SP I and IIA, at the bottom of
the page, is a positive apparatus in which deviations from the Nepalese
manuscripts and the editio princeps of Bhattaral are given as well as sigla
for all the manuscripts - including the A and R manuscripts - that
share the adopted reading. Above the lacuna register two additional lay-
ers reporting the variants of the manuscripts of the AmbikakhalJr;la and
RevakhalJr;la are placed. In this way the readings of all three recensions
can be reconstructed from the apparatus accompanying the edition.
The above procedure could not be adopted in the case of the material
edited for the present work. While the three available recensions (S, R, A)
transmit on the whole the same text up to adhyaya 162 - apart from
numerous individual variants, omissions and additions - the situation of
the subsequent chapters is significantly different. After this adhyaya, the
147 The importance of this restriction needs to be stressed. While BhaHaraI conflated
the text of the Nepalese manuscripts and A 3 , the single Ambikakhal}.Qa manuscript
available to him, in the critical edition preference is given to the readings of the
oldest manuscripts, viz. the Nepalese recension. The advantage of this procedure
for future research is undeniable: the attestation of a certain phrase, word, name,
concept, idea, etc., in the main text of the critical edition of the Skandapura1)a
testifies to its existence before 810 AD, the date of the earliest manuscript. In this
regard it is quite astonishing to read the following words in Bailey's review of SP
Vol. I: (IThere are four levels of critical apparatus and, to me, this seems to be
extraordinarily complicated, and though it stands as an exemplary specimen of a
text transparent to all of its sources, this is somewhat at odds with the indigenous
understanding of Pura1).as where multiple versions of a text of the same name are all
regarded as legitimate within the tradition of that name. But where does the text
lie? In its collection of manuscripts - mistakes, corruptions and all recensions
- or in the artificially constructed text of the scholar who lives outside of the
culture where the text is used?" (Bailey 2005, pp. 227-228). While the editors
of SP Vol. I (and IIA) have tried to make the multiple versions of the text as
transparent as possible, Bailey apparently wants to do away with the apparatuses
reporting these different versions. The result in that case would indeed be an
artificially constructed text, for which there would be no way of checking what the
indigenous manuscripts themselves actually read.
52 Introduction
148 Cf. now also Harimoto 2004a, 2004b, and T6rzs6k 2004. Already before adhyaya
162 major revisions have taken place, but not to such an extent as in the subsequent
chapters.
149 It should also not be forgotten that especially in the case of R such a transcription
suggests an objectivity which does not correspond to the real state of affairs.
Reading the R manuscript necessarily requires an element of interpretation, as the
transcription of many of its characters is based on considerations of sense. Cf. SP
Vol. I, p. 302, and the apt characterization on p. 34: the manuscripts we used,
R is perhaps the most difficult to read and transcribe. Even after considerable
experience, the form of early Bengali script used by this scribe presents difficulties.
He appears to write several in more than one way, and partly due to this
there are many ambiguities." In Dimitrov's classification of Bengali scripts the R
manuscript would probably have to be defined as 'premodern Bengali' (d. Dimitrov
2002, p. 29).
Editorial Principles 53
I have opted for a different procedure and prepared two separate editions,
one on the basis of the S recension, the other on the basis of the R and A
recensions. Although this procedure may at times have led to conflation
of readings from the R and A recensions, it has some clear advantages.
First of all, one must bear in mind that the relationship between the
three recensions is unequal. Although the R and A manuscripts repre-
sent recensions in their own right - as their difference in name
already indicates - they clearly go back to a common ancestor. As I have
demonstrated elsewhere (Bisschop 2002), this ancestor must have been in
existence at least before the time of (beginning of the 12th
century), who quotes three verses from the ayatana list, without parallel
in the Nepalese recension, in the TIrtlJavivecanakaJ;lQa of his K,tyakalpa-
taru. Tiirzsiik (2004) has pointed out various aspects in the additional
material of the R and A recensions which tend to show that this ancestor
may in fact have to be dated quite early. The exact relationship between
the two recensions needs further investigation, but for the moment we
can safely assume that in the case of the ayatana list what we seem to
have are basically two sub-recensions of the same archetype. 150 By a well-
grounded, eclectic choice between the two recensions we are in a position
to restore a text that is still relatively intelligible, and which comes close
to what the ancestor of the two recensions may have looked like.
150 The unequal relationship between the three recensions is already evident from the
fact that while the ayatana list is transmitted in one chapter in the S recension,
it is divided into five sub-chapters in both the A and the R recension. Cf. also
the tables of adhyaya colophons of Bhattarru's edition, the AmbikakhaI;u;1a and the
RevakhaJ;uja in Harimoto 2004b.
151 With the addition of A7 to the group of available manuscripts of the Ambika-
kha1)ga recension, it has become clear that this manuscript is not exactly on the
same level as the other A manuscripts. A few preliminary observations may be
54 Introduction
A few examples from the edition will serve to show that R and A presup-
pose a common, already corrupt archetype. A telling instance is the loss
of text in SPRA 4.51cd. All manuscripts read the word sahasravanditar[!
together with the next verse, which obscures the aparavaktrii metre in
SP RA 4.52. Because SPRA 4.51ab is incomplete and we expect the men-
tion of the result of visiting A.mratakesvara, the iiyatana under discussion,
we must assume that sahasravanditar[! belonged to padas cd of the same
verse, of which the remainder was already lost or damaged in the archetype
of both recensions. It is precisely for this reason that reconstruction of
made here about the nature of this manuscript and its position among the other A
manuscripts. More research is certainly needed, but see now also the discussion in
SP Vol. HA, pp. 10-12, which confirms the following observations for another part
of the text, its Vara{lasfmaha:tmya. First, it can be observed that All AZl A3 and
A4 all descend from a common hyparchetype, which had a number of errors that
are not shared with A 7 • cr. e.g. the variants in SPRA 1.2c (tan me A7j tathii Ac
A4), 1.16a (manunii A7i yatvanii ACA4)' 1.48d (dhiimrikrtam A7i dhustrikrtam
ACA4), 2.37d (0 saTflghayofy, A7i 0saTflsayo!y, ACA4); 3.28b (tridr$.tyiiyatane$u A7;
dr$Pyiiyatane$u ACA4 [unmetr.]), 3.63c (bahulingasya tallingarp, A7; bahulirp,garp,
tasya liTf'yaTf' A,-A, [approx.]), 3.78e (brahmalokiiya A7; brahmaloko A,-
A 4), 3.101a (iisviisitlis A7; lisvliditlis A 1-A 4), 4.11b and 4.18a A7;
kuiayu§tau ACA3; kusapu§tau A 4), 4.21b (priyiin asun A7; priylin abhun ACA4)'
4.3ge A7; A,-A,), 5.102a A7; A,-A4)' 5.103a
(jajiie A7; yajiie ACA4)' This hyparchetype cannot have been identical with A 7,
for the latter has wrong readings that are transmitted correctly in the other A
manuscripts: e.g. SPRA 1.2d (suvrata ACA4; suvrata A 7), 1.25c (tapodhaniis ca
ACA4; tapodhana na A 7), 2.54b (patir A 1-A 4; pavir A 7), 3.18a (pramatha ACA4;
prathamii A 7), 3.67b (kamakathasrayiiJ:!, ACA4; kiimayathasrayaJ:!, A 7), 4.32b (satya-
vatfsuta ACA4; satyavatfsutaJ:!, A7)' 4.44a (naganlisoru ACA4; niiganiisaura A7)'
5.51c (avatf'l'l}o ACA4; atfva'r'l}o A 7), 5.87a (kanyakubje ACA4; kanyaggambhe
A7)' 5.93b (dharmasarp,jiiitam AI-A4; dharmasangitarp, A7)' More importantly,
the other four A manuscripts contain a few verses, also transmitted in R, that
are missing in A7: SPRA 2.57-58, 3.62 and 3.70. The presence of these verses in
the A manuscripts cannot be explained if A7 (or, for that matter, a copy of A7)
was the hyparchetype of the other A manuscripts. The reverse is also true: A7
contains a few padas, shared with R, that are not transmitted in ACA4: SPRA
1.28, 3.68ed and 5.38ad. Both A7 and the hyparehetype of A,-A" on the other
hand, presuppose a common hyparchetype, since they share so many errors and
omissions that are too specific to be coincidentaL Significant is a case such as
SPRA 1.15b, where all the A manuscripts report a lacuna, while R preserves the
supposedly original reading. The examples of involuntary omission of verses are
too numerous to mention, but let me just draw attention here to the large omis-
sion of SPRA 4.52f-5.30b in all the A manuscripts, a part only preserved in the
R manuscript, but supported by the presence of parallels for many of its verses in
SPS' On the whole, the scribe of A7 seems to have been a faithful copyist, for on
virtually every page of the manuscript we encounter gaps that are clearly marked
as such. On the other hand, the number of hyper- and hypometrical readings in
this manuscript suggests that he was often careless. In the present edition I have
noted two instances of additional padas only in this manuscript which are both
due to dittography: after SPRA 1.28f and 4.33b.
Editorial Principles 55
152 This symbol does not belong to the edition proper. It only serves as an indicator
that something may have been lost. Placement of the symbol is often arbitrary,
because in the case of presumptive loss of text it is mostly hard to figure out what
exactly was the chain of events and where the loss may have occurred.
153 If any general answer must be given it will be that each individual case should
be considered in its own right. cr. Housman's dictum (1988, p. 326): "... every
problem which presents itself to the textual critic must be regarded as possibly
unique." For a discussion of a similar issue in a different class of literature, that
is the transmission of the Atharvaveda in the Saunaka Sakha and the PaippaIada
Sakha, and the problems it poses for the editor of the PaippaIada transmission,
see Griffiths *2004, pp. xlvi-xlix. Griffiths, following Hoffmann and Witzel, distin-
guishes between the 'original' reading of the text when it was composed and the
56 Introduction
One example comes from the first sub-chapter (SP RA 1.23cd) and is an
instance where all readings diverge: R has dhiirayanti dhrtiiciirii Mzrbhii;,
pariviiritiii;" A has a hypo metrical reading dhrtiihiisa Mzrbh* sa katha1J!
raviprabhai;, 154 and the parallel passage in SPs llcd reads dhiirayanti
mahiitmiina iiszrbhii;, saratha1J! ravim. The verb dhiirayanti, missing in A,
is clearly original since it is shared by Sand R. The reading mahiitmiina
(S) on the other hand cannot have been the original reading of the R/ A
recension, because R's dhrtiiciirii is partly supported by A's non-sensical
dhrtiihiisa. Finally, A's approximate sa katha1J! raviprabhai;" with the extra
syllables added to fill up the line (cf. A,!), indicates that saratha1J! ravim
(S) must have been in the archetype and that R's pariviiritiii;, is therefore
corrupt. These text-critical observations lead us to the conclusion that the
archetype differed from SPs llcd only with respect to dhrtiiciirii (instead
of mahiitmiina). This particular example shows that, although the paths
of transmission may be complex, the original reading can be restored with
relative certainty. On the other hand, it also illustrates the difficult posi-
tion in which we find ourselves in the majority of cases, where there is no
parallel for a verse in SPs.
In many cases I have felt confident to emend the text on the basis of
SP s , e.g. diiravai;, in SPRA 2.60d (diinaviii;, R/ A), tathaivotkutukeivare in
SPRA 3.27b (tM ca vandetukeivara R, tathaiva kukkuteivare A) and yatra
vaikhiinasii in SPRA 3.38a (yatriivakhiinasii R, yatriivaisviinarii A" yatra
vaisviinarii A,/ A3/ A" yatra vaisviinaro A4)' In order to draw attention
to the fact that an emendation or conjecture is supported by SPs the
symbol < S has been introduced in the edition. This symbol does not
necessarily imply that the entire reading is actually present in SPs nor
that the emendation is only based on the parallel in SPs, but merely
indicates that there is some support for an emendation or conjecture in
the Nepalese recension.
On the other hand, there is always the danger of incorporating readings
of SPs where the archetype in fact did have something different. For
example in SPRA 1.50c, although a dat. pI. tebhyai;, is expected and is
present in the parallel of SPs 26c, I have refrained from emending the
gen. pI. te;;ii1J!, because the evidence of the manuscripts suggests that this
was in fact the archetype's reading, probably considered as a genitivus
pro dativo. Such grammatically poor readings may in fact prove to be
characteristic of the predecessor of the R and A recensions (d. in this
particular case SPRA 5.ld with annotation) and one should be careful
therefore not to remove them. We do not know enough yet about the style
neuter for masculine: tam referring to lingam 1.lOb, 1.29b, 1.52c; 157 vano-
ddesarJl3.37c (conj.).
155 Compare the lists of so-called 'aiSa language' in Goodall 1995, pp. lxv-lxviii,
Goodall 2004, pp. lxxviii-lxxxv, Goudriaan & Schoterman 1988, pp. 44-109, and
Torzs6k *1999, pp. xxvi-lxvii.
156 Oberlies's recent Grammar of Epic Sanskrit (2003) is a valuable addition concern-
ing the language of the two Epics. He observes (Oberlies 2003, p. xxxi): "Decisive
for the use of grammatically incorrect forms is in the overwhelming majority
of cases - their position within the piida." In the words of Hopkins (quoted by
Oberlies 2003, p. xxxi): "Metre surpasses grammar." The same observation holds
true for the language of the SkandapuraI}a in its various recensions.
157 This phenomenon is also quite common in SPs: e.g. 1ge, 23c, 7ge, 173d. Possibly,
it expresses the identity of God with the liriga; in other words, it might reflect a
kind of personification of the liriga. In most cases the word liriga is simply treated
as neuter.
158 Brockington (199S, p. 93) refers to this phenomenon in the Epics and calls it the
use of the present tense as the tense of narration, the present taking on a general
58 Introduction
lyap for ktvii: manasii cintya 3.50a; tiin arcya 3.83a; riidhya varJasata,!,
somam 5.36c.
na-vipula: 1.3a, 1.10c, 1.11a, 1.14c, 1.16c, 1.36a, 1.37c, 1.39c, 2.11c, 2.19a,
2.22a, 2.28a, 2.30c, 2.31c, 2.50c, 2.56a, 3.11a, 3.32a, 3.33a, 3.47c,
3.79c, 3.81c, 4.44c, 4.48a, 4.48c, 5.1c, 5.5c (conj.), 5.7c, 5.21c, 5.22a,
5.35c (conj.), 5.54a, 5.68c, 5.74a.
bha-vipulti: 1.64a, 2.29a, 2.37c, 3.19c, 3.80a, 3.86a, 4.14c, 4.23c, 4.43a
(conj.), 4.45e, 5.52c, 5.56a, 5.69a, 5.111a.
ma-vipula: 1.7a, 1.18c, 1.25a, 1.27a, 1.38c, 2.35a (conj.), 2.48c, 3.34c (cor-
rupt), 3.41a (conj.), 3.46c, 3.66a, 3.70a, 3.87a, 3.94c, 4.34c, 4.49c,
5.29c, 5.32c, 5.48e (conj.), 5.59c, 5.71c, 5.72c, 5.87c, 5.109c (irregu-
lar), 5.113c.
From this list we can deduce that the use of vipulas must be considered
a hallmark of the style of the redactor(s) of SPRA 1-5. In addition to
the frequent use of vipulas the attention paid to metrical variety is also
indicated by the presence of non-sloka metres at the end of each sub-
chapter in SPRA. The following non-sloka metres can be identified: upa-
jati (1.66-68; 2.63-65; 3.103; 3.105), malinf (1.69), indravajra (3.102),
va,!,sasthti (3.104; 3.106; 3.107; 3.108; 5.115), (3.109), apara-
vaktra (4.52) and mtilatf (5.116). Among these, SPRA 3.103, 3.109 and
5.116 are uncertain because the transmission ofthese verses is too corrupt
to reconstruct them with satisfaction. With the possible exception of the
malatf metre which concludes the ayatana section, all these metres are
well-attested in other Purii1:ws as well. 163
163 Cf. the list of metres in PuraJ).as, prepared by Hohenberger (1965, pp. on
the basis of twenty PuraQas including the Harivarpsa. No malafi is recorded by
him, but I can refer at least to SP 30.30 (= lvItP 180.31 / LiP 1.92.19).
60 Introduction
In general, I follow the standard of the presentation of the text and crit-
ical apparatus in SP Vol. I and IlA, the principles of which have been
laid down in SP Vol. I, pp. 45-54, to which I refer here for the sake of
brevity. However, the different editorial procedure adopted here, the fact
that two separate editions have been prepared instead of one, has a few
consequences for the presentation of the edition. First of all, the only
available evidence for the text of SPs are the manuscripts S, and S" and
for that of SP RA the R and A manuscripts. Consequently, the two top
layers of the apparatus in SP Vol. I and IlA - reporting variant readings
in R and A - are absent in both editions. The main critical apparatus
is the lowest one at the bottom of the page and it is a positive one, just
as in SP Vol. I and IlA. This apparatus contains one new symbol « S),
used only in the edition of SPRA. It serves to indicate that an emendation
or conjecture is supported by a parallel in SPs (cf. above, p. 56). Directly
above the main apparatus is the lacuna register, as in SP Vol. 1. These
are the only two layers in the apparatus of SPs. The edition of SPRA has
two additional layers on top of that. One is a layer recording the relevant
verse and pada numbers in case there is a parallel with SPS. '64 The other
one, a fourth layer, is present only in SPRA 4, where a testimonium from
the Krtyakalpataru is cited for three verses.
In the main text of SPRA the lines printed in smaller type indicate
the absence of the line in question in either one of the two recensions,
the Revakh8.I,lQa or the AmbikakhalJQa. The symbol [... J at the end of
a line has been introduced in this edition to draw attention to the fact
that part of the text may be lost. These are the only deviations from the
presentation of the text and apparatus adopted in SP Vol. I and IlA. For
easy reference the list of symbols and abbreviations in the apparatus of SP
Vol. I has been reproduced (with a few adaptions) on the page preceding
the editions of SPs and SPRA (p.89).
In the colophon of SPs the adhyiiya number 167 is printed with no
name, following the policy of SP Vol. I and IlA. 165 The number 167 itself
does not occur in the manuscripts, but there can be no doubt that this
is the correct adhyiiya number in the S recension. The colophons in the
two available S manuscripts are hard to read, although S, seems to give
the adhyiiya name as TlrthavarlJana. Bhattarm has reconstructed Siva-
yatanavarlJana. In the edition of SPRA I have followed a different policy,
because there is no certainty about the adhyiiya number (ef. above, p.5).
A colophon name has been edited on the basis of the colophons in the
166 The presentation in SP Vol. IIA is slightly different, in that there the synopsis and
annotation are integrated.
62 Introduction