HISTORY THROUGH TEXTUAL CRITICISM

in the study of Saivism, the Paiicariitra and

the Buddhist Yoginftantras

ALEXIS SANDERSON

My purpose is to communicate a number of conclusions concerning

the histories of Saivism, the Paiicaratra and the latest stratum of Esoteric
Buddhism reached through the identification and text-critical examination
of extensive parallel passages linking the scriptural literatures of these
traditions, first those of the various Saiva traditions with each other and
then these with those of the Paiicaratrika Vaisnavas and the Buddhist
Yoginitantras of the Sa:rp.vara corpus. ｔ･ｸｴＭ｣ｲｩｾ｡ｬ＠ analysis has achieved
this end by revealing the direction of redaction, so providing evidence of
relative chronology.
Subsumed within this purpose is that of drawing attention to certain
aspects of the practice of textual criticism. By establishing that one text
has incorporated passages from another we obtain a powerful means of
selecting from among the variants transmitted by the manuscripts of the
borrower. A redactor will tend to revise the text he is incorporating,
especially when he is taking it from the corpus of another religion, since
then he must take care to remove all that would betray its origin. But
whatever in the transmission still matches the source must be the derived
text as it was at the moment of its redaction before the development
of variants, except in cases in which the agreement is likely to be the
result of random variation or secondary scholarly editing. In a closed
recension, in which the readings of the surviving manuscripts of a borrower
allow the editor to discern the text of their archetype, the source-text will
be a means of evaluating that archetype and may alert the editor to its
limitations as the representative of a small, conveniently coherent area of a
wider and less straightforward transmission. When a recension is open_the
source-text will reveal this through the inconstancy of the groupings of the
manuscripts in agreement with its readings; and when the contamination
that has produced this openness is not so complete that it has obliterated
all evidence of vertical structure the source text may be crucial in guiding
2 Sources and time History through Textual Criticism 3

the editor to a clearer perception of this concealed order, one that he can
rii.sakii.rikavrtti), the latter's pupil Bhaga Vidyakai_ltha (Bhii.vacu(!.ii.mal)i on Maya-
extend to the text beyond the boundaries of the parallel. sarp.graha), Abhinavagupta ( Tantrii/oka, Parii.trirp.sikii.vivaral)a etc.) and his pupil
Source-texts, then, are the most valuable of all testimonia. For unlike ｋｾ･ｭ｡ｲｪ＠ (Svacchandoddyota, Netroddyota etc.). The only absolute dates associated
others they point directly to the state of the text at its inception. The with these are those given by Abhinavagupta for the completion of certain of his works:
Ｈｓ｡ｰｴｲｾｩｳｶ＠ 40]66 for his Kramastotra, Ｈｓ｡ｰｴｲｾｩｳｶ＠ 40]68 for his Bhairavas-
search for them should therefore be a primary concern of anyone who un- totra, and Ｈｓ｡ｰｴｲｾｩｳｶ＠ 40]90/Kalisarpvat 4115 for his isvarapratyabhijiiavivrtivi-
dertakes the critical study of such literature. As my examples will show, marsin>, that is to say, A.D. 990/1, 992/2 and 1014/5. Bhatta Ramaka'.'tha predates
this may require one to extend one's reading beyond the boundaries of the Abhinavagupta because in Tantrii.loka 8.428-434b the latter quotes (without attribu-
text-group or Sastra within which one commences one's work. But such tion) the former's summary of Matanga, Vidyii.pii.da (hereafter VP), Patala 23. But he
does not predate him by much because ad Mrgendra 1.11 Bhatta Ramakantha's father
breadth is in any case the royal road to success in scholarship. For though Narayai_lakai_ltha quotes the lsvarasiddhi of Utpaladeva, who ｴｾｵｧｨ＠ ａ｢ｨｬｾ｡ｶｧｵｰｴＧｳ＠
the critic may be assisted by excellent testimonia for parts of certain texts teacher ｌ｡ｫｾｭｮｧｵｰｴＮ＠ From M!J-lava we have the Prii.yascittasamuccaya, a digest of
he will more commonly find himself confronting problems which only the chapters on penance found in the Saiva canon, by one Hrdaya.Siva. According to metric-
ｾｬｹ＠ inaccurate verses by another found at the end of this text Hrdaya.Siva was a pupil of
cultivation of this breadth can equip him to recognize and solve. He will
l§vara.Siva in the lineage of Lambakari_la, an ascetic of the Rai_lipadraka (/ ａｲ｡ｾｩｰ､ｫＩ＠
have on occasion to choose between rival readings that are equally well · Matha of the Mattamayura lineage, who moved to the Goratikamatha and became the
attested; and even where the transmission can be reduced to a single read- ｄｩｫｾ｡ｧｵｲ＠ of king Siyaka of Dhara (University Library, ｃｾｭ｢ｲｩ､ｧ･＠ [hereafter ULCJ,
ing he ought still to submit that reading to his judgement, and if he has Add. 2833, f. 128rl-vl): ·
grounds to suspect that it is spurious he should aspire to undo the cor- 1 mattamayuravarp.saje srfral)ipadra(kasrame)
ruption through emendation. The extent of his success in these tasks of lambakaTI)a iti nama prthivyii.rp. prathito munif;.
2 nikhilasaivasiddhii.ntasrotaf;.sarp.skii.mpii.vanaf;.
selection, diagnosis and emendation will depend on the level of his know-
candrii.nkitajatajutaf;. kailii.sastheva sankaraf;.
ledge of the restraints imposed on language, style, and sense by a wide 3 srfral)ipadmkii.Sramii.d gorii#kamathe (' )gamat
range of contexts. Most importantly, to master texts of. this kind, written dhii.rii.yii.rp. siyakam iti anujagriiha pii.rthivam
within a highly complex and multiform world of religious practice and doc- 4 iisit tatsantatau munif;. .Sri-iSvarasiva iti
jagatfpatibhir nryaif;. pujitapiidapankajaf;.
trine and written for persons engaged in it, the critic must work towards 5 tacchi§yo nikhilii.gamii.rthavimalii.darso munif;.
an ever more thorough understanding of that world; and this will lead him srrhrdayasiva iti vidyesvaratulyakirtif;.
from one area of the Tantric tradition to another and will also require him, 6 yasyocchrito dik§u tapaf;.prabhiiva udiryate sii.dhujanair jagatyii.m
like the Tantric scholars before him, to have a grounding in the domains heliid uddhrtagamasiistmsii.raf;. sa jhatity iikha7)(!.itaviidaka7)(!.uf;.
that underly and inform the Tantric, such as those of Vaidika observance ld prthivyii.rp. prathito conj.: prathitaf;. prthivyii.rp. cod. 2b srotah
and hermeneutics in the case of the Saiva and Paiicaratrika systems, and em. : srotra cod. 3d piirthivam em. : piirthivaf;. cod. 6a ｡ｰｦ［Ｎｲ｢ｨｩｶｾ＠ｴ
corr. : tapaf;.prabhii.vo cod.
of Abhidharma and Vinaya in the case of Tantric Buddhism. In this way
the training of the textual critic is nothing less that the intimate study There were two Siyakas of Dhara. The first ruled c. A.D. 863-890 and the second,
of the civilization that produced and understood the documents he con- the son of Vairisirpha, came to power some time before 949 (the date of the first of
two known epigraphs from his reign [Epigmphia Indica [hereafter EI] 19, p. 236)]) and
fronts. Nor can that study proceed without textual criticism, since that is was still in power in 969 (the date of the second [EI 19 p. 177]); see D. C. Ganguly,
the art of reading the documents which are its richest and most numerous History of the Paramii.ra Dynasty (Dacca, 1933), pp. 30 and 37. The first grant of
witnesses. his successor Vakpati II is dated Vikrama 1029 (=A.D. 972/3); see Georg BUhler, 'The
Udepur Pra.Sasti of the kings of Malva' in EI 1, p. 222-238. The Cambridge palm-
The knowledge of the history of this area that we can obtain by means leaf manuscript of the Prii.yascittasamuccaya is dated (Nepala]sal)lvat 278 (=A.D.l158).
other than textual criticism is meagre. For the most part we are reduced to There must have been at least two Gurus between Lambakari_la and Hrdaya.Siva's guru
cautious generalizations. Concerning the chronology of the early scriptural Isvara.Siva. Consequently the outer limits of the date of his work are 1159 and either
949 or 863.
sources of Tantric Saivism we can do little more than assert for most of the
The early Kashmirian authors show that they have direct knowledge of the following
texts known to us that they predate the citations that appear in the works· Tantras:
of the earliest datable commentators, that is to say, in works of the tenth to 1. Saiddhantika: Anantavijaya, Kacabhiirgava, Kamika, Sardhatrisatika Ka/ottara
early eleventh centuries from Kashmir and Malava, 1 and for a few of them, (/ Kalajiiii.na, Kii.lapii.da), Sapta§atika Kiilottam, Trayodasasatika Kiilottara, Kiral)a,
that they go back at least to the early ninth century since they survive in Devyii.mata (Prati§thatantra of the Nisvii.sa), Nandikesvaramata, Nisvii.sa, Nisvasakii.-
rikii., Nisvii.sakii.rikii.-Dik§ottaro, Nisviisottaro, Parii.khya (/ Saurabheya), Piiramesvara,
1 These early datable commentators are (for Kashmir) Bhaga Naraya.I)aka'.'tha Pau§karapiiramesvara, Prati§thapii.mmesvara, Bhiirgava (=Kacabhiirgava?), Bhii.rgo-
(Mrgendravrtti), his son BhaHa Ramaka'.'tha (Kiral)avrtti, Matangavrtti, Kalottara- vottara, Matangapiirame8vara, Mayasarp.graha (/Maya), Mukuta (/Makuta/ Makota),
vrtti, Nare5varaparfk§ii.pmkii.sa, Bhogakii.rikavrtti, Mok§akarikavrtti and Paramok§ani- Mukutottara, Mrgendra (/ Mrgendrottam), Yak§inipii.ramesvara, Raurava, Rauravasu-
4 Sources and time HistQry through Textual Criticism 5

Nepalese manuscripts of that date, 2 are recognizably paraphrased in the Haravijaya of Ratnakara composed in Kashmir around 830, 3 are listed in
.J' •

tro.saf[tgraha, Lalita, Saruajiianottaro. (of the Vathula), /Siik§majsvfi.yambhuv.a, Svaya- 1. Paramesvara (/ Pau§karaparamesvara): (a) ULC:: Add. 1049, dated year 252'[of the
'r[tbhuvasiitrasaf!tgro.ha, H af[tsaparamesvara. era of Manadeva, alias ａｩｰｾｵｶ｡ｲｭｮＩ＠ (A,D. 827/8), incomplete; {b) a few folios found
2. Mantrapitha: Svacchanda (/ Svatantra), Aghoresvarisvacchanda, Lakulasvaccha- with, and roughly contemporary with, the Skandapurai;>a of year 234 (A.D. 819/810),
nda. - . National Archives, Kathmandu (hereafter NAK) MS. 2-229. Saiddhantika. ·
3. Vidyapitha and Kaula (including K:ama): Nisatana (/ NiSisaf!tcO.ra/ Nilacaro.), The colophons of ULC Add. 1049 call the work Paramesvara. There are no colophons
Anandagah11ara, Ananda, Anandesvaro., Anandabhaira11a, Ucchu§mabhaira?J_a, Utphu- in the NAK 2-229 fragments. Extensive quotations from the Pau§karapaiamelvara
llakamata, ｏｲｭｩｫ｡ｵｬｾ＿ｊＬ＠ Kalikfi.krama, Kallkula, Kallmukha, Kulakaii?Jidhi, Kula- in Hrdaya8iva's Prayascittasamuccaya, Rajanaki ｾ｡ｫｶｩｴＧｳ＠ NityadisarPgraha and
gahvara, Ｎｋｵｬ｡｣ｩＨｭｴｾＬ＠ Kulapaiicasika, Kulayukti, Kulasara, /'J'rikajkularatnamala, shorter quotations in the Kashrriirian commentators agree with ｭｾｴ･ｲｩ｡＠ in ULC
Kularna11a Kaulottara Kramarahasya, Kro.masadbha11a, Khecarimata, Gama, Gah?Ja- Add.1.049 and NAK 2-229. · Palaeographica!' considerations indicate ti{at the unstated
ra, ｇｾｰｴ｡ＬＧ＠ Guhya, Guhyayoginl, Catu§kapaiicasika, Candragarbha, Caryakula, .Qama- era of the dates of both the Parameiuara and the Skandapurai;>a ms. can only be that
ra (/ Vadya), Tatt?Jarak§O.vidhi, Tantrarajabhattaraka (=Jayadrathayamala), Tantrasa- which is found in Nepalese inscriptions that begin during the reign of king Manadeva,
dbhavf!., Tantrasamuccaya, Timirodghii.ta, Tilaka, Trikakula, Trika[tantrafsara (/Mali- alias ａｩｰｾｵｶ｡ｲｭｮＬ＠ and ran:ge in their dates froin 30 to '194. The era of these dateswas
ni'sara), Trikavif[tlatika, Trikasadbhava/Trikahrdaya, Tri.firobhairava,. De?Jipaiica.fa- assumed by some to be that of ｈ｡ｲｾ＠ (A,p. 606) but Luciano Petech lias 'shown from
taka, De?Jyay'amala, Nandi.fikha, ｎｩｴｹ｡ｮｲｾＬ＠ Nirmaryada, Nisakula, Paiicamrta, Tibetan evidence that it is A.D.576 ('The Chronology of the Early.Kings of Nepal',
Paratrif[tlika (/ Paratrisika, Anuttarasiitra, Trikasutra), Picumata (/ Brahmayamala), East and West, n.s.12, 1961, pp. 227-232). That this is the era of the date of the
Bhargasikhakula, [Trika/bhaira?Jakula, Bhairavayamala, Bhogahastaka, ｍ｡ｴｾｨｲﾭ Skandapura'l)>a ms. was recognized by M. Witzel (Indo-Iranian Journal (heareafter IIJ)
ka Matatrimsika Matsyodarimata, Madhavakula, Malinivijayottara (/ Piiruasastra), 29 (1986), p. 259, n. 9) and accepted by R. Adriaef!sen, H. T. Bakker and H. Isaacson in
ｙｾｧ｡ｳｦＡｴ｣ｏＮＬ＠ Yoginrkaula, ｙｯｮｹ｡ｾｶＬ＠ Ｍｌ｡ｫﾧｭｩｵｬｾｶＬ＠ Vajasaneyasa'r[thita, 'Towards a Critical Edition of the ｓｫ｡ｮ､ｰｵｲｴｾＧＬ＠ 'IIJ 37 (1994), p. 326.
Vijiianabhairava, {Siddhajvrravalr[kulabhairavaf, Vlravalmrdaya, Sarvavrra (Sarvaca- 2. Sarvajiianottarottara (col.: paramelvare mahatantre pitaputriye sarvajiianottaro-
ra), Saruacii.rahrdaya, Siddhayoge.fvarimata (/ Siddhamata), Siddhii.mrta, Skandayama- ttare. .. ). ULC Add. 1049, incomplete. Saiddhaniika.
la, Haf[tsayamala, Hrdayabhattaraka (/ Hardesa), Hai(lara. · 3. Bhairavivardhamanaka. ULC Add. 1049, incomplete. N on-Saiddhantika.
4. GaruQ.ika/Bhautika: Totula, ｋｲｩｹ｡ｫｬｧｵｴｾｯＬ＠ ｃ｡ｴｾＨｬｳｩ､ｨＮｲ＠ 4. ｄ･ｶｹ｡､ｴｾＨｬｫＮ＠ ULC Add.1049, incomplete. A hymn to the Goddess attributed
5. Universal: Pingalamata (/ Jayadrathadhikara), Netra (/ Mrtyujit, Amrtesa; Sarua- to the ｊｩ｡ｮｾｶｭｨｴｲ＠ (col.: jiianaTt>a?Je mahatantre ...•). ｎｯｮｾｓ｡ｩ､ｨｴｫＮ＠
srotalj.saf!tgraha), Saruasrotalj.saf[tgrahasara. 5. KalottaraiVathula. A fragment of an as yet unidentified recension found with the
6. Of unknown affiliation: Candrajiiana, ｓｮｫ｡ｴｾｨｩｹ＠ (/ ｓｲｩｫ｡ｴｾｨＩＬ＠ Saruamanga/a Skandapura'l)>a MS of A.D. 810. Saiddhantika. .
(Mangala, Mangala), ｂｨ｡ｲｵｴｾ＠ (?{Tantraloka 25.14c]) · 6. Saruajiianottara. NAK 1-1692; incomplete; assignable to the 9th century on
The Saiddhantika authors rely for the most part on Saiddhantika Tantras ·but also palaeographic grounds; see T. Goudriaan in T. Goudriaan and S. Gupta, Hindu Tan-
quote as authorities the S11acchanda (Mantrapitha), the Malinivijayottara Ｈｖｩ､ｹ｡ｰｾｨ＠ tric and Sakta Literature (Wiesbaden, 1981) (hereafter Goudriaan 1981), p. 38, n. 27.
(Trika]) and the Pingalamata, which is based in the ｖｩ､ｹ｡ｰｾｨ＠ as an affiliate of the Saiddhantika.
Picumata but is concerned with installation rituals in both Saiddhantika and non- 7. Nilvasatattvasaf[thita. NAK 1-277; complete; c. A.D. 900; see Goudriaan 1981, p. 34,
Saiddhantika systems (Varna, ｄ｡ｫｾｩ＠ (t- Picumata) and Trika). The non-Saidhantika n.l. Saiddhantika. An apograph of this manuscript prepared in A.D.1912 (London,
authors (Abhinavagupta and ｋｾ･ｭ｡ｲｪＩ＠ rely on almost all the sources listed, both Wellcome Institute Library, Sanskrit MS !.33 [B)) has independent value, since the
Saiddhantika and non-Saiddhantika. exemplar had deteriorated further by the time it was photographed in the recent Nepal-
ｈｾ､｡ｹＸｩｶ＠ of Malava has brought together the penance chapters of the following: German Manuscript Preservation ｐｲｯｪｾｴ＠ (hereafter NGMPP).
1. Saiddhantika: Kamika, ｋｩｲ｡ｴｾＬ＠ Tilakaparamesvara, Devimata ( =11yamata), It comprises the Nilvasamukha (ff.1v-18v6), the NiSvasamiilaiMiilasii.tra (ff.18v6-
Nih.fvasakii.rika, Ni/j.lvasottara, Paramelvara, Pu§karaparamesvara, Prati§thaparame- 23v1), the Nisvasottara (/ Uttarasii.tra) (If. 23vl-29r5), the Ni.f11asanaya (/ Nayasiitra)
ｬｶｾｲ｡Ｌ＠ Paraka, Matangaparamel11ara, Mrgendra, Mohacii.rottara, Raura11a, Vathu- (If. 29r5-42;r5), and the Nisvasaguhya (/ Guhyasiitra) (If. 42r5-114v). In the last verse of
la, Vamadeviyakriyasaf!tgraha, ｖｩ､ｹ｡ｰｵｲｴｾＬ＠ Sapta11idyadhara, Sarvajiianamahodadhi, the work Siva invites Devito ask him to go on to teach the fifth Sutra, the Nisvasakarika
Sahasra, Svaya'r[tbhuvasiitrasaf!tgraha, Haf[tsaparamelvara, Paiicalatika Haf[tsapara- (apograph, f.ll4v3): cat11aro kathitii siitra *ye mukhadya (conj. : sa + lcadya)
melvara. 11aranane I paiicaman tu para1p siitraf[t karika nama namatalj. I siicitii ｳｩｴｲ｡ｭ･ｾ＠
2. Mantrapitha: Aghorelvarisvacchanda (Dvadalasahasra-), Svacchanda (Dvadasa- kiirika puna/]. prcchatha. That work too, then, which has survived in South-Indian MSS
sahasra-), Rasasvacchanda ($atsahasra-). {Institut ｆｲ｡ｮｾｩｳ＠ de ｐｯｮ､ｩ｣ｨｾｲｹ＠ (hereafter IFP], MS T. 17, T. 150), which include the
3. Vidyapitha: Picumata, Siddhayoge.fvarimata-Bijabheda ( Caturui'r[tlatisiiha- Dlk§ottara as the last part of the text, was in existence at this time. .
sra-), Siddhayoge§varimata- ｃｩＨｬ｡ｭｴｾｮｲＭ Uttarottara-Jiiiiniidhikara- Bhairavodya- 8. ｋｩｲ｡ｴｾ＠ NAK 5-893; complete; dated (NeplUa]saipvat 44 (A.D. 923/4). Saiddhantika.
na ( Siirdhatrisiihasriko.-), Siddhayogelvarimata- Trikasarottara ( Caturvi'r[tlatisiiha- 3 The Haravijaya {hereafter HaVi) contains two learned hymns, the first addressed
sriko.-). by Spring to Siva (6.5-170), the second by the Siddhas and Sa.dhyas to ｃｾｑＮｉ＠ (47.2-
4. Universal: Pingalamata. 168). The poet praises these deities in the terms of the principal soteriologles of the
Thus he adds the following to the list of Saiddhantika Tantras known from the works day. In keeping with the allusive style favoured by the Sanskrit poets he seldom names
of the early Kashmirians: Tilakapiiramesvara, Paraka, Mohaciirottara, Vamade?Jiya- the schools whose language he deploys. But many of the stanzas, particularly in the
kriyasaf[tgraha, ｖｩ､ｹｰｵｲ｡ｴｾＬ＠ Saptavidyadhara, Sarvajiianamahodadhi, Sahasra, and a hymn to Siva, draw on the terms and concepts of the Saiddhantika Saivas; and in
Paiicasatika Haf[tsaparames?Jara; and the Rasasvacchanda to that of the Mantrapltha. some of them one can recognize echoes of specific scriptural sources. These are the
2The following Saiva Tantric texts are completely or incompletely preserved in S?Jayambhuvasiitrasaf!tgraha (hereafter Sva8uS), the Rauravasiitrasaf!tgraha {hereafter
Nepalese palm-leaf manuscripts of the ninth and early tenth centuries: RauSuS) and the Matangaparames?Jara(hereafter MaPa). Svayambhuvasiitrasaf!tgraha,
6 Sources and time History through Textual Criticism 7

the text of the Skandapurii:T,La preserved in a manuscript completed in 810, 4 or. are mentioned as having been studied or practised during this period in
the Matangapiirametvara, and theRauravasiitrasa7]>graha. (1) HaVi 6.126 +- SvaSuS
Saiva inscriptions from Cambodia. 5
VP 1.10-11 (ed. Filliozat); (2) HaVi 6.139 +- SvaSuS 3.11-13; (3) HaVi 6.121 +- brahma'!> svii.yambhuva7]> caiva kaumiira7]> yamala"f!' tathii
RauSuS1.17; (4) with HaVi 6.147cd cf.RauSuS 4.48 (variant Cited ad loc.by Alaka) 129 sii.rasvata"T[' sagiindhii.rain aisiina7]> nandiyii.malam
and Parii.khya, ed. D. Goodall (forthcoming) 1.5 (but the latter is a less probable source tantrii.?}Y etani yu§miika7]> athanyani sahasrasaJ;.
since it shows signs of being th<;! latest of the pre-tenth-century Saiddhantika scriptures; 130 bhavi§yanti narii yais tu yu§mii.n yak§yanti bhaktital],
see Goodall, op. cit. xx, xlviHv); (5) HaVi 6.84 +-MaPa VP 8.65ab; (6) HaVi 6.127ab
+- MaPa VP 11.33; (7) HaVi 6.115 +- MaPa VP 7.70c-71; (8) HaVi 6.162 +- MaPa propose that yamala7]> in 128d is to be understood with all (sarvase§ataya), so
VP 9.29b-31. .. 35-35; and (9) HaVi 47.124 +-MaPa VP 7.31ab. that the titles indicated are Brahmayii.mala, Svayambhuvayiimala, Kumarayii.mala (=
The colophons of the Haravijaya describe Ratnakara as a courtier of king Cippata- Skandayii.mala), Sarasvatzyii.mala, Gii.ndhiirayiimala, Uanayiimala and Nandiyiimala.
J ayapi<;Ia, alias Brhaspati, when the latter was a juvenile (snbiilabrhaspatyanujfvino For of these seven all but the Sviiyambhuvayamala and the Giindhiirayii.mala are found in
... mahiikave rii.jii.nakasrfratnii.karasya). Kalhal).a tells us ( Riijatarangi'(lf, ed. Stein the scriptural lists of Yamalatantras known to me, namely (1) in the list of the 64 Bhair-
[hereafter RaT], 4.703 and 687) that he was murdered inLaukika [38]89 (A.D.813/4) avatantras (sii.diisiva'!> cakram) given in the Srika!lthfya's account of the canon Rajanaka
after· a reign of twelve years. But this date is too early, because ｋ｡ｬｨｾ＠ describes j Takeyakavarta in Nityiidisa7]>grahapaddhati, Oxford, Bodleian Library (hereafter Bod!.),
Ratnakanl. as an illustrious figure in the court of Avantivarman, who ruled half a cen- ﾷＮｾ＠ MS Stei!l Or. d. 43 (hereafter Snka'(lthiya), f.19r3-6: Brahma-, Vi§'(IU-, Skanda-, Uma-,
tury later, from 855/6 to 883) (RaT 5.34), at a time for which Kalhal).a's dates are more l Ruru-, Atharvarw-, Rudra- and Vetala-; cf. the version of the same quoted by Jayaratha
reliable. Seeing this Georg BUhler pointed out (A Detailed Report of a Tour in Search 1 in Tantriilokaviveka 1 (1) pp. 41-43; (2) in the version of the same list in Satka 1 of the
of Sanskrit Manuscripts in Kasmrr, Rajaputana, and Central India, Bombay, 1887,
pp. 42-43) that the date given for Cippata-Jayapi<;Ia must be corrected by the addition
' Jayadrathayiimala,_ NAK 5-4650, { 184v6-8 (44.5-6): Brahma-, Rudra-, Vi;'(lu-, Skan-
da-, Gautamlya-, Atharva-, Vet{lla-, and Ruru-; (3) in the Siddhayogesvarimata, Asiatic
of about twenty-five years to c. 826-838. Chinese evidence adds weight by diverging Society of Bengal (hereafter ASB), MS 5465 G, f. 69vl-5 (29.16-20): Rudra-, Brahma-,
to the same extent from Kalhal).a's dates forCandrapi<;Ia, an early king of this dynasty. Vi§'(IU-, Skanda-, Uma-; ( 4) in Jayadrathayiimala, $atka 1, f. 169r7-v6 (36.16-25):
Kalhal).a assigns the reign to 686-695 (RaT 4.44 + 4.118), but the Tangshu records Rudra-, Brahma-, Vi§'(IU-, (S)kii(n)da-, Auma-jGautami"ya-; (5) ibid,, f.182v8-183r3
that a king Chentuobili of Kashmir, who can only be this Candrapi<;Ia, sent an embassy (42.1-7): 5 Upayamalas: Vetala- (t-Brahma-), !sana-(+- Vi§'(IU-), Atharva- (t-Rudra-),
to the Chinese capital in 713 and was granted the title of king on the imperial rolls in Sarasvatf- ( +-Skanda-), Soma- ( +- Umii.- ); (6) Picumata, NAK 1-363, .f. 199r2: Rudra-,
720. See A. Stein, Kalha'(la's Rii.jatarangi'(lf. A Chronicle of the Kings of Kasmfr, 1900, (S)kanda-, Brahma, Vi§!lU-, Yama-, Viiyu-, Kubera-, Indra-; (7) in Jayadrathayiimala,
ad 4.45 (with bibliography). ｾ｡ｴｫ＠ 3, NAK 5-1975, f. 170r8 ( Yoginfsa7]>ciir;tprakarana, Yoginfcakresvarotpattipatala,
We may add the Kiilottara to the list of Saiddhantika Tantras current in the ninth v. 37d), ·listing fifty-eight Tantras in which Siva has taught ﾷｹｯｧｩｮｦｳ｡ｔＨｉＮ｣ｲｾＺ＠ Ruru-,
century if the Guhyasiddhi of Padmavajra in the Buddhist Tantric system of the Guhy- Rudra-, Umii.-, Gaur!-, Skanda-, Bhairava-, Vi§'(IU-, Nandi-, Sukra-, Sakra-.
asamiija is from this period, as it may well be. For verses 8.11-16 instruct the initiate Kotivareya is the site also called Devikotta, Devikota, Sripitha, Srikota, or Sol).itapura.
to disguise himself as a Saiva, win the confidence of a family of untouchables, teach A Vieyaya of the Pul)<;iravardhana Bhukti (EI 14, p. 325) it may be identified with
them the Siddhanta, give them Saiva initiation, receive the ｄ｡ｫｾｩｬＩＮ＠ and then exchange Bangarh on the bank of the Punarbhava river in the west Dinajpur District of Bengal;
it for one ·of their unmarried girls, thus obtaining the consort necessary to the practice see D. C. Ganguly, 'Yadavaprakasa on the ancient Geography of India', Indian His-
of his Vidya-observance ( vidyii.vratam). Padmavajra specifies the Saiddhantika texts to torical Quarterly 19 (1943), p. 224; J. C. Sengupta, West Dinajpur, Calcutta: State
be ·used: Editor West Bengal District Gazetteers, 1965, pp. 186-187, 219-220. It is one of the set
of eight ｋｾ･ｴｲ｡ｳＬ＠ or primary sites of the Goddess, venerated in our earliest sources,
12 darsayec ca tatas te§ii.'!> dharma'!> siddhantapilrvakam the other seven being Prayaga (mod. Allahabad), Varai;ta (mod. Benares), Kollagiri
kalottaradisa7]>siddha7]> no cen ｮｩｾｶ｡ｳｭ｢ｨ＠ (mod. Kolhapur in Karl).ataka), AHahasa (in the Bhirbhum district of Bengal), Jay-
anti (mod. Jayantipur in the Ganjam District of Orissa?), Caritra (in the Puri Dis-
12b dharma7]> siddhanta em. : dharmasiddhanta ed. 12c sa7]>siddha7]>
trict of Orissa) and Ekamra (Bhubaneswar in the same District) (Madhavakula and
conj. : sa7]>suddha7]> ed.
Kiilasa"T['kara§a'(lyii.T['-saktinirdesa [both in Jayadrathayiimala, ｾ｡ｴｫ＠ 4], Guhyasiddhi,
Then he should reveal to them the Siddhantadharma established in such Nityii.hnikatilaka etc.) or these with Ujjayini in place of Ekamra (Nisisamcii.ra NAK 1-
[scriptures] as the Kiilottara, or that whose source is the Nisvii.sa. 1606, f. 14r [3.5-6]). It is also one of the eight Sma8ana-sites worshipped in the 'Mandala
of the Picumata, Patala 3, the others being Prayaga, Vara!fasi, ｖｩｾ｡ｪ＠ (mod. ｊｾｪＮｰｵｲ＠
4 NAK 2-229. This Ur-Skandapurii.?}a lists seven Tantras of the Mother Goddesses in Orissa), Kollagiri, Prabhasa (mod. Somnath-Pattan in Surat), Ujjayini (in Malwa),
( matrtantra?}i) in Adhyaya 171. Siva has declared that henceforth ｋｯｴｩｶ｡ｲｾ＠ will Bhutesvara Ｈｩ､･ｮｾｴｹ＠ uncertain) and Ekamra. According to Keyemaraja it is one of the
be a place sacred to the Mothers (171.121), that he will be present there as Het- sites of the eight Sma8anadhipatis listed in Svacchandatantra 2.177-180b, the other be-
ukesvara to lead them (171.122), and that whoever. drinks the water of the SU!akunda ing Kamarupa, Ujjayini, Kasmira, Kaner, KaravTra, U<;l<;liyana and Hiral)yapura. Het-
in that place, the site where Bahumarpsa has just slain the Danavas with her ｴｲｩ､ｾＮＬ［＠ uka/Hetukesvara is named as the Siva or Ksetrapala of .this site in 'Picumata Patala
and offers obeisance to Bahumarpsa herself, will be protected from all harmful spirits 3, NisisaT(I.cii.ra Patala 4, Kubjikamata ｐ｡ｴｬｾ＠ 22, and the Buddhist Dakiirnavata;.tra
(171.124-125). He continues (171.127-130b, ed. Kreyl)aprasada Bhattarar, Kathmandu: Patala 50, Prakara1.1a 3 (NAK 3-447). An ｩｾｳ｣ｲｰｴｯｮ＠ of the Pala king Nayapala (EI 39,
Mahendrasaf:!1skrtavisvavidyalaya, 1988): · no. 7) from the middle of the ninth century records (v. 28) the construction of a temple
of Hetukesa at Devikota. The rebuilding of a temple of HetukaSulin is mentioned in a
127 aha7]> brahma ca vi§'(IUS ca f§ayas ca tapodhanal;. copper-plate inscription of the Mleccha king Vanamalavarman of Assam (r. c. 835-860)
matrtantra'(li divyii.ni matrYajii.avidhi7]> param (EI 29, no.20, v.24.)
128 pu'(lyll'(li prakari§yamo yajana7]> yair avapsyatha 5 An inscription of the reign of the Cambodian King Rajendravarman (944-968)
8 Sources and time His.tory through Textual Criticism 9

, Going back further than this we lose sight of titles and can only es-
reign, while staying in the village Marrura in the Va.Ip.guva9i distrkt I
tablish that Tantric Saiva texts of certain familiar kinds must have been have on the full moon day of VaiSakha, to 'attain merit and fame for my
present and that these or some of these were probably works among those parents and myself, donated with [the due pouring of] water the· village_
that were current later. Thus.I propose that a scriptural corpus of the kind ｉｰ｡ｾｵｩＮｫｬ＠ with all exemptions to Sudarganacarya as Guru's fee on the
occasion of [my] initiation into the ｍｾＴ｡ｬ＠ of Siva.'
we find later in the Saiddhantika scriptures must have been_ in existence by
the beginning of the seventh century. There survive inscriptions recording For the expression ﾧｩｶ｡ｭＡＧｬ､＾｣ｾ＠ 'initiation into the Mai)Qala of Siva.' cf. the ref-
erence in the Cambodian inscription at Bantay ｋ､ｾｩ＠ cited above to nai§vii.samar>(lali
the Saiddhantika Saiva initiation of .three major kings during the second ､ｩｬ｣ｾ｡＠ 'initiation pertaining to the Mai)Qala of [of Siva] taught iii' the Ni§vii.sa.'
half of that century, 6 and during its first half the Buddhist philosopher The initiation of bevendravarman was ·recorded in a copper-plate grant dated
[Gailga]sa.Ip.vat 184 (EI 26, no. 5, ed. R. K.Ghoshal, ll. 7-10) found during excavations
at Bantay Kdei (Bulletin de !'Ecole ｆｲ｡ｮｾｩｳ･＠ d'Extreme Orient [hereafter BEFEO] near the Dharmalhig$Vli.I:a temple in the village Sudaba near Parlakimidi, close to the
25: 359) says about a certain Sivacarya who ､ｩｾ＠ in· c. 890 a.D.d was the Hotar of Orissa-Andhra border,· near Mukhalingam, the ancient Ganga capital in what was then
this king and his predecessors beginning with ｉｾ｡ｮｶＮｲｭ＠ II (900-928), that he know as Kaliilga. It records that this village was gifted to the Guru ｐ｡ｴｮｧｾｩｶ｣ｲｹ＠
ha.d mastered the 'Saruajiiii.nottara and all other Salva scriptures' '(v. 38ab: ta$yO,sye as his dalc§i!'ii. for performing the king's initiation:
sanhitii.s sarvvii.s saruajiiii.nottarii.dayalj.) and been consecrated as a Sa.iva _·officiant
viditam astu bhavatii.1[t yathii.smii.bhir aya1Jt grii.mas saruvalcarail]. pa-
through initiation into the ｍｾｴ［Ａ｡ｬ＠ taught in the Ni§vii.sa (36: nai§vii.sama!J.{lalln dlk§ii.n
ｲｩｨｴｹ｡｣ｮ､ＮｬｰｾＱｊｭｯ＠ atmana§ ca pu!}yii.bhivrddhaye
nai§thilcii.caryyatii.rppa!J.i1Jt §ivacii.ryyii.bhidhii.(lhyii.fJt yo bhi§ekavidhau dadhau).
. .. bhagavatpatanga§ivii.cii.ryii.ya gurave ､ｩｬ｣ｾｯｴ｡ｲＮ･＠ gurupiijii.yai _dattas
The Sdok Kak Thoip. inscription -of A.D.1052 reports 'that the Sira§cheda, the
... pravarddhamii.navijayarii.jyasa7[!vatsara§ate caturii.§ite 100 80 4 utkl-
Samrrioha, the Nayottara, and the Vinii.§ilcha ( = Vi'!J.Milcha), · the '[teachings of] the
T!'!J.a1Jt ... ｳ｡ｲｶ｣ｮ､･ＡｊＮｾ＠ ' ·
ｦｯｵｾ＠ faces of Tumburu', were transmitted by one ｈｩｲｾｹ｡Ｎ､ｭｮ＠ to Sivakaivalya, the
Hotar of king Jayava.rman II (A.D. 802-850), founder of the kingdom of Angkor, and Be it known to you that for the increase of the merit of my parents
that this Hiral)ya.daman performed a ritual following these texts to protect the· king- and myself I have gifted this v.illage with exemption from all taxes for
dom and to guarantee its continued independence. These are the principal Tantras of as long as the moon and sun shall remain, to His Holiness, the. Gurti
the Varna branch of the ｖｩ､ｹ｡ｰｾｨＬ＠ teaching the cult of Tumburu and his four sisters ｐ｡ｴｮｧｾｩｶ｣ｲｹＬ＠ ... , in order that I might make a worshipful offering
J aya, Vijaya, J ayanti/Ajita and Aparajita. · to hhn now that [my) initiation has been completed . . . . Engraved by
Indirect evidence that this cult was already well-established in India at lealit as early Sarvacandra in the year 184 of [the dynasty's] prosperous and victorious
as the beginning of the eighth century is provided by the incorporation ofTumburu and rule.
his sisters in the Esoteric Buddhism that reached China in the middle of that centucy.
The date is A.D. 682/3 if we accept 498/9 as the epoch of the Ganga era as argued
Chapter 15 of the tantricized Adhyardha§atilcii. Prajiiii.pii.ramitii., alias *Nayasiitra, in the by Mirashi (EI 26, pp. 326-336; EI 27, p.192; EI 28, pp.171-174) and now widely
Chinese translation completed by Amoghava.jra (TaishO 243, of A.D. 774), speaks of the accepted (see Richard Salomon, Indian Epigraphy, New York, 1998, pp.187-188.)
four sisters among deities making offerings to the Buddha (tr. Ian Astley-Kristensen, The initiation of the third king, Narasiip.havarman II of Kanci, also called R.ajasiip.ha
The Rishulcyii, Tring, 1991, p.159). In his Chinese commentary on this text (Ta.ishO and Atyantakama, took place at some time between c. 680 and c. 731, those being
1003) he identifies them as the sisters of Tumburudeva (ibid., 'p.160). The five also the earliest and latest dates that have been assigned to the beginning and end of the
appear in the Maiijugri enclosure of the Ｊｇ｡ｲ｢ｨＮ､ｴｵｭｾ［Ａｬ＠ in the version known much disputed period of his rule. It is indicated by a contemporary inscription in
as the Genzu brought from China to Japan by Kiikai in A.D. 806. There the central Kaiici on the king's eponymous R.ajasiip.hesvara shrine (South Indian Inscriptions I:
ｍ｡ｩｪｵｾｲ＠ configuration is flanked by, among others, Domoro, Shaya, Hishaya, Ajita and
24). Though not stating explicitly that he underwent Sa.iva initiation (divadik§ii.) it
Aharajita; see Adrian Snodgrass, The Matrix and Diamond World Mandalas in Shingon conveys the same by describing him in v. 5cd as §aivasiddhii.ntamii.rge §riman atyanta-
Buddhism, Volume 1 (New Delhi, 1988), p. 391; and Ulrich Mammitzsch, Evolution of
lcama(IJ.} k§atasakalamalo ... , 'Atyantakama, all of whose Impurity has been destroyed
the Garbhadhiitu Ma!J.{lala (New Delhi, 1991), pp.150-151, 288-293
by following the path of the Saivasiddhanta.' For in- Saiddhantika Salva. doctrine Im-
6The three kings are the Calukya Vikramaditya I of Badami, the Eastern Ganga purity is what impedes the soul's liberation and it is held that the only way in which
Devendravarman, and the Pallava Narasiip.havarman of Kaiici. The earliest of these a soul can be freed from its effect is to undergo Salva initiation. A similar expression
initiations is that of Vikramaditya I. It was received while he was in the Kollapuram occurs in an inscription of the Kalacuri king Yuvaraja.deva II (r. c. 980-990), referring
Taluk of the Mahbubnagar District in Andhra and recorded in his Amudalapii.9u plates to the Salva initiation received by king Avantivarman in c. 825 (EI I, pp. 251-270,
announcing a gift to his initiator made on 30th April, 660 (EI 32, no. 21, ed. D.C: v. 49): mattamayiiranathal]. / nil].§e§akalma§ama§>m apahrtya yena sanlcrii.mitam para-
Sircar, lL 25-30): maho nryater avantel].. ' ...... [Purandara,] the abbot of Mattamayiira, entirely removed
viditam astu vo 'smabhiQ. pravarddhamii.navijayarii.jyapaiicamasa"l[lvatsa- from King Ava.nti the stain of Impurity and transmitted to him the supreme radiance
re marfiiragramii.dhivii.salcal]. ｶ｡ｦｊｴｧｵｲＨｬｩｾｹ･＠ matapitror ii.tmanal ca [of Sivahood].' That the inscription is referring to Sa.iva initiation here rather than to
putiyaya§ovii.ptaye- §iv.ama!'!laladi'k§ii.yii.1[1SUdar§anii.cii.ryaya ｧｵｲ､｡ｫｾｩｻＧＡＮﾭ simple instruction is proved by an inscription found in the ruins of the Sa.iddhantika
rtha1[1] vai§alchapauf!tl'amii.syam iparu1f!lcalnamagrii.mal]. saruvaparihii.ram monastery of AraJ;lipa.dra (/Ra.I]ipa.dra, -t mod. Ral)od, 25° 5 N, 71° 5 E). For that
udalcapiiruvo dattaiJ. .. narrates the same occurrence more directly (EI 1: 41, vv. 10-,-13):
vo 'smabhil]. em. Sircar : vii.smii.bhil]. ep. grii.madhivasakal]. em. Sircar : 10 tasmii.t purandaragurur guruvad garimr>a/J.
'grii.madhivii.salcail]. ep. prajiititirekajanitasya babhiiva bhumil].
yasyii.dhunapi -vibudhair itilcrtyalafJtsi
'Be it .known to you that in the fifth year of my prosperous and victorious vyii.hanyate na vacana1[1 naydmii.rggavidbhil].
10 Sources and time ｈｩｾｴｯｲｹ＠ through Textual Criticism 11

Dharmakirti (c. 600-660) goes to the trouble of attacking the Tantric that Tantric Saivism of this relatively public and strongly soteriological
Saiva practice of initiation as the means to liberation. 7 These facts reveal variety was not merely present in the seventh century but well established.
And this implies the existence of Tantric Saiva scriptures. For while in-
11 ｶ｡ｮ､ｹｾ＠ ko pi cakiisty acintyamahima tulya"f(L munir bhiisvatii
novation in religious practice must have preceded its scripturalization, it
riijann uttamasabdapurvvasikhariibhyar7J-7J-am prakiT1}7J-adyutil),
dik§iirthfti vaco nisamya sukrt> caroktam urvvipatir could not have survived without it, far less reached such prominence. As
yasyehanayaniiya yatnam akaroc chr?:man avantil]. purii · is exceptionally clear in the case of the Buddhist Yoginitantras, the need
12 gatvii tapasyantam upendrapurvve pure tadii srfmadavantivarmmii to display a scripture as the proof of validity could be greater than the
bly.rsa"f(L samiiradhya tam atmabhumi"f(L kathaiicid an'iya cakara putam need that such scripture be fully intelligible and coherent.
13 athopasadyiipya ca samyag aisf"f(L d'ik§ii"f(L sa dak§o gurudak§i7J-iirtham We also have some evidence from this period of the existence of texts
nivedya yasmai nijarajyasiira"f(L svajanmasiiphalyam aviipa bhupaly. belonging to more esoteric, private and Siddhi-directed forms of Tantric
Then came the Guru Purandara, who as befitted a Guru had the gravity Saiva practice. In the first quarter of the eighth century Bhavabhuti's
that comes from the highest wisdom; whose teachings concerning the du- picture of Kapalika observances and doctrine in his drama Miilatzmiidhava
ties [of Saiva initiates] are observed even today by scholars learned in the
way of discipline; whom the glorious and virtuous king Avanti[varman]
corresponds closely with what is seen in parts of the Jayadmthayamala; 8
made efforts to bring to this land because he desired to receive [Saiva] the poet Vakpati, his contemporary, knows of the Kaula worshippers of the
initiation and had heard from one of his agents that there was a cer- Goddess; 9 and in the first half of the seventh century Dharmakirti testifies
tain holy ascetic in the vicinity of Uttama.Sikhara shining in unimaginable that among holders of the soul-doctrine there existed Mantrakalpas, 'texts
glory, shedding his radiance like the sun. Avantivarman then went to
[Purandara], who was practising austerities in Upendrapura, and having
of Mantra ritual', which taught procedures involving the taking of life,
striven to win his favour succeeded in bringing him back to sanctify his theft and sexual congress. He cites the :Qakinitantras and Bhaginitantras
kingdom. Then, having served him with devotion he properly received as examples. 10 His contemporary, the poet Bai_la, mentions a Mantrakalpa
Saiva initiation [from him]. The wise king then presented him with the
best part of the wealth of his kingdom as Guru's fee (gurudak§i7J-ii) and so
suffice in itself to show that he has the Saivas in mind, since the Paiicaratrikas too had
brought his human birth to fulfillment.
this belief (see, e.g., Jayakhyasa"f(Lhita, ed. Krishamacharya, 16.44a, 60; Sattvatasa1[!hita,
That these initiations were Saiddhantika Saiva rather than in the Varna, Dakl)i!fa or ed. Dwivedi, 19.4; Pau§karasa"f(Lhita, ed. Sampatkumara Ramanuja, l.llc-12, 34-41).
some other Saiva system of ritual is a strong probability in the light of the respect- What narrows the field is that in 261c-262b Dharmakirti attacks the claim that the ef-
able and catholic character one would expect of the publicly proclaimed initiation of ficacity of initiation (and therefore the validity of the scriptures that teach it) is proved
a king. This, for example, was the tradition followed by the Kalacuris of Cedi, the by a variety of the ritual called tuliidfk§ii 'scales-initiation', in which the initiand is
Paramaras of Malava, the Colas, the Kakatiyas, the Khmer kings of Cambodia, and weighed before and after the performance. Those present, we are asked to believe,
others. However, it would not be a certainty without further evidence. In the case of see that the person's weight is greater before than after, and are thereby made to
the Pallava king, that his initiation was Saiddhantika is clearly stated. In the other cases understand that the initiation has succeeded in destroying the Impurity ( malaly.) that
it is not; but it is, I propose, implied by circumstantial details. That the Calukya's initi- impedes the soul's omniscience and omnipotence, that being of the nature of a material
ation was Saiddhantika is suggested by its description as an initiation into the Ma!f<;iala substance (dravyam), albeit imperceptible. Now, this variety of initiation is Saiva. It is
of Siva (Sivama7J-4aladik§ii), and while his Guru Sudar§anacarya appears not to have cited by the Saiddhantika Sadyojyoti in Tattvasa"f(Lgraha 36: suddhi"f(L vrajati tuliiyii"f(L
a Saiddhantika initiation-name Ｈ､Ｇｩｫﾧｮｭ｡Ｉｾｳ｣ｲｰｴｵｬ＠ texts indicate that the taking dfk§iito brahmahatyato mukhyiit I pratyayato jiinfyiid bandhanavigama"f(L vi§ak§ayavat,
on of such a name is optional in most ｣｡ｳ･ｾｷ＠ do find such names among those 'By initiation on the scales he is purified even of such great sins as brahminicide. One
of three of the twenty-seven Brahmins listed in the inscription as indirect recipients may know that his bonds have been destroyed through the evidence [of one's senses],
of shares ( iiciirye'l}iipi briihmanebhyo dattalf.): Rudra.Sivacarya, Gayatri§iva and Para- just as [the validity of the Garu<;latantras is proved] by the cessation of poison [brought
ma§iva (11. 32-34). That Devendravarman's initiation· was Saiddhantika is indicated by about by their Mantras]'. It is the subject of the twentieth chapter of Abhinavagupta's
the name of the officiant Patanga.Sivacarya. For a Saiddharitika abbot with this name Tantriiloka, where he draws on Pat ala 9 of the Trika's Tantrasadbhiivatantra. His com-
see the Gwalior Museum stone inscription of Patanga.Sambhu published by V. Mirashi mentator Jayaratha on 9.15b quotes the Tantrariijatantra. The passage he cites is Jaya-
in Journal of the Madhya Pradesh Itihasa Parishad 64 (1962), pp. 3-12. drathayamala, ｾ｡ｴｫ＠ 3, Patala 20 (ghoraghoratariicakre catuscatvari1J?.satidik§iividhil},
7 See Pramii7J-aviirtikakiirika, ed. Yushi'i Miyasaka (1973-75), Pramii7J-asiddhi, variety of initiation no. 21: ghatadfk§ii), f. 153vl-2.
vv. 259c-269c, in which Dharmakirti attacks those non-Buddhists who claim on the 8 See my 'Purity and power among the Brahmans of Kashmir' in M. Carrithers, S.
sole authority of their scriptures ( iigamamatre7J-a) that it is initiation ( d'ik§ii) that brings Collins and S. Lukes (ed.), The Category of the Person (1985), p. 213.
about liberation. See his commentator Manorathanandin (ed. R. Sanktyayana) ad 259c- 9 Gaurj.avaho, ed. Shankar Pandurang Pandit, v. 319: visasijjantamahiipasuda1J?.Sa'l}a-
260b: nanuktam isvare'l}iigame 'sty iitmii mok§aS ciisya dfk§iividhineti 'Surely Isvara Sa"f(Lbhamaparopparurf.hii I gaya'l}e cciya ga"f(Ldhaurf.i1!' ku7J-a"f(Lti kaula7J-iirzo 'The Kaula
(Siva] has said in his [Tanttic] scriptures that the soul exists and that it attains liberation women seem to form a shrine in the air as they clamber over of each other in their
through the ritual of initiation' and the rebuttal of DharmakTrti's criticisms put forward eagerness to watch a victim of human sacrifice being carved up.'
by ｋｾ･ｭ｡ｲｪ＠ in the '.inquiry into [the theory of] initiation' (dik§iivicaraly.) with which 10 Dharmakirti, Pramii7J-aviirtikasvavrtti, ed. R. Gnoli, p. 163, II. 3-5: na, dharmavi-
he concludes his commentary on Svacchanda, Patala 5 (Svacchandoddyota 3 [5]73, 10- ruddhiiniim api krauryasteyamaithunahinakarmiidibahulanii"f(L rf.iikinfbhaginftantriidi§U
84, 9). The fact that he attacks the idea that initiation is the way to liberation does not darsaniit 'No. [It is not the case that Mantras are effective only if they involve virtue];
12 Sources and time History through Textual Criticism 13

manuscript for the propitiation of the ｍ｡ｨｫｬｾ､ｹｭｮｴｲＬ＠ in his de-

for there is the evidence that there are also sinful [but nonetheless effective observances
( vratam) which are] full of violence, theft, sexual congress, base acts and the like in the
scription of a fictitious Tantric Sadhaka. 11 The terms Mantrakalpa and
"QakinTtantras, Bhaginitantras and others'. Kalpa are used in exactly this sense in Tantric literature, to denote the
Our attention has previously been drawn to this passage by R. M. Davidson ('The manuscript of a text setting out the procedure for the propitiation of a
Litany of Names of Manjusri. Text and Translation of the Manjusrinamasaipgiti', in Mantra. 12 Finally, the jurist Bharuci, who may also belong to the first
M. Strickmann (ed.), Tantric and Taoist Studies in Honour of R.A. Stein, Vol 1,
Bruxelles, 1981, p. 8). He took Dharmakirti to be referring to Buddhist Tantras here half of the seventh century, refers to the Bhutatantras in his commentary
and wondered whether these might not be the same as the Yoginitantras known from
later evidence. But it is certain that he is mistaken and that Dharmakirti refers to The forms ajitam and vijayam here are evidently bhfmavat contractions of ajitiitantram
Tantras of the Saivas. To be convinced of this it should be sufficient to read the passage and vijayiitantram. The same source also lists a Catu§ka, which is said to comprise
in the light of the following remark earlier in the same discourse (ed. Gnoli, p. 123, four Tantras, ·one ｃ｡ｴｵｾｫ＠ for each of the sisters: (f.l8r14-v1) catu§karrt tu tato va-
ll. 22-24: bauddhetarayor mantrakalpayor hi'IJ'siimaithuniitmadarsaniidaya 'nabhyuda- k§ye tac caturdhii prakfrtitam I jayiicatu§karrt prathamarrt vijayiikhyarrt tathaiva ca 1
yahetavo 'nyatha ca var'l}-yante. 'Violence, sexual congress and belief in the Self are jayantyiikhya'IJ' trtfyarrt. tu tathii. caiviipariijitam. In the list of the Vamatantras given
taught to be cause of suffering and bliss in texts of Mantra ritual that are Buddhist and in the account of the Saiva canon prefixed without notice to the Jiiiinapaiiciisikii at
non- Buddhist [respectively].' the beginning of NAK 5-4632 (Kiilottaratantra) are a Jay{iitantrja, a Vijayfiitantrja, an
Karl).akagomin's commentary on the later passage associates the "Qakinitantras with Ajit{iitantrja and an Apariijit{atantr}a: (vv. 22-23b [f. 2r6]) nayarrt nayottararrt mukarrt
the taking of life, the Kambukinitantras with theft, and sexual congress with the sa'IJ'moharrt mohanamrtam I karapujavidhana'!' tu Vi'(tatantrarrt taduttaram I jayarrt
Bhaginitantras, which he identifies as the Caturbhaginitantras 'the Tantras of the ca vijayarrt caiva ajitarrt cii.parajitam. The first $atka of the Jayadrathayamala lists a
Four Sisters'. See p. 578, ll. 7-9 (ed. R. Sa1pk:rtyayana): if.akinibhaginrtantradi§u da- bhiigineyarrt tantram, i.e. a Tantra of the Sisters in the Varna division as one of the
rsanat. if.akinitantre caturbhaginftantre, adisabdac cauryahetu§U kambukini:tantradi§u Tantras of the cycle of the M ahiisarrtmohanatantra (f. 178r7-40.lllc-112): *bhii.gineya'!'
dar§aniit, "'For there is the evidence in the "Qakinitantras, Bhaginitantras and oth- (em. : bhogineyarrt cod.) visalar[! ca bhratrtantriidikii.ni tu I mahiisarrtmohajaladhes
ers.", i.e. because there is evidence in the "Qakinitantras, in the Caturbhaginitantras tantraratniini kotisaly, I nirgatani mahiidevi bhuktimuktikarii'l}-i ca.
and, considering the words "and others", in such as the Kambukinitantras that instig- 11 Har§acarita, ed. FUhrer, p.l61, ll.ll-14 (the words of Bhairavacarya): bhagava-

ate robbery'; and ibid., 1!.14-17: if.akinitantre samayavyavasthii, yadii prii'l}-ina'IJ' ha- to mahiikii.lahrdayaniimno mahiimantrasya kf§'l}-asragambariinulepanenakalpena kalpa-
tvii khiidati tadii mantrasiddhim iisiidayati, tathii kambukinftantre steyiicara'l}-iit siddhir kathitena mahii.smasiine japako!yii kfiapurvasevo 'smi tasya ca vetiilasiidhaniivasii.nii
uktii, tatha maithuniicara'l}-iit siddhiprada kii.cid devateti bhaginftantrantare kvacit sa- siddhily,, 'I have completed the preliminary service (purvasevii) of. the great Mantra
ｭ｡ｹｾ｝Ｌ＠ 'In the "Qakinitantras the rule of postinitiatory discipline is that one attains called the Heart of Mahakala by muttering it ten million times in a great cremation
the power of the Mantra if one kills and devours a living creature. In the Kam- ground while wearing a garland of black flowers, a black robe and black unguent, with
bukinitantras one's attainment is said come about through the practice of robbery. all the adornments prescribed in the Kalpa.' Neither the Mantra nor its Kalpa can be
And somewhere in the Bhaginitantras the postinitiatory rule is that a certain goddess identified now bu_t the details of procecfure given by Ba'.la tally closely with what is laid
bestows the attainment on one if one practises sexual intercourse'. I have seen no other down in Tantric Saiva sources. With this account of a purvasevii, also called purascarya
reference to the Kambukinitantras ('Tantras of the Robber Wives'). No "Qakinitantra or vidyiivratam, i.e. the initial period of ascetic japaly, etc. to be undertaken after one
survives, to my knowledge; but they are referred to elsewhere, and in terms that ex- has received a Mantra, whereby one becomes able to accomplish feats (karmii'l}-i) with
actly support Kar'.lakagomin's claim. See Keyemaraja, Netroddyota ad 20.39, referring that Mantra (see, e.g. Nisvasaguhya, f. 80v3: japamiina-m eva miisena purvaseva kfiii
to methods by which YoginIs kill their victims ( tattaif.if.ii.kinftantroktii.nii.m iicarii'l}-ii.m); bhavati), we may compare, e.g., Picumata 21.84-87 (f. 100v3-4): caturthan tu prava-
Somadeva, Kathii.saritsiigara 20.102-114 (/Keyemendra, Brhatkathii.maiijan 3.368-374); k§yiimi kf§'l}-abhinnaiijanaprabham I duscararrt ghorariiparrt ca citrascaryadidipitam 1
and Bhasarvajna, Nyiiyabhii§ana; p. 395, ll. 18-21 (ed. Svami Yogindrananda): yathiibhi- mahiisiddhikara'IJ' nama mahocchu§marrt prakirtitam I kr§'l}-iimbaradharo mantri kr§'(ta-
cii.rii.kar§a1)-iidikarmii.nu§.thii.yinii.'IJ' if.ii.kinftantrii.di§u krauryasteyamaithunii.dyatinindita- sragdamabhii§italy, I valayabhara'l}-an divyarrt piidau nupurabhii§itau I keyuraka!isutrarrt
karmopetavratanu§thayinii.'IJ' dhanesvarii.desena tacchatru'IJ' brii.hma'l}-iidikam api visvii.- ca padau laktakaraiijitau I kar'(tiilankiiramakuta'!' valikiibhara'l}-an tathii 1 kapiilarrt su-
sya ghii.tayatii.'IJ' cauryii.dikarmakiiri'f}-ii'IJ' ceti. Note also that Muslim accounts of Indian lakha!viingan dhanur niiriicasarrtyutam . . . . A Kalpa of Mahakala (worshipped with
religion going back to the report of an eighth-century AbbasTd envoy refer to prac- the eight Mothers Brahm! etc. as his retinue) is given in Patala 52 ( mudriipfthiidhikiiro
titioners of human sacrifice whom the manuscripts call Tahkiniyya, Dahkiniya, etc. nama) of the Picumata.
W. Halbfass (Tradition and Reflection, 1991, p. 105, n. 85) wonders whether these have 12 Svacchanda 4.498abc: ｭ｡ｮｴｲｫｬｰｩＮｾｳｵＧｉｊ＠ ca khatikii'!' chattrapiiduke j ｕｾＧｬｽＭｦ｡ﾭ
anything to do with thakasii.stra/Thuggee. I judge it more probable, though by no means rahita'!' dattvii ... 'Having given [the newly consecrated Sadhaka] the Mantrakalpa, the
certain, that they are referring to practitioners of the "Qakinitantras ( *if.iikinfyii.IJ,). For rosary, the chalk, the umbrella and sandals, but not the turban[,which is given to the
these accounts see B. B. Lawrence, Shahrastii.nf on the Indian Religions, 1976, pp. 54 Guru alone] .... '; Svacchandoddyota ad Joe.: kalpaly, sii.dhyamantrapustikii 'The Kalpa is
and 226-237. a book whose subject is [the procedure for propitiating] the Mantra he is to propitiate';
As for the BhaginTtantras, if Karl)akagomin is right that Dharmakirti is referring to Sviiyambhuvasutrasa'IJ'graha, Mysore ed., 19.33cd: kalpokta'!' tu japa'IJ' krtva devaya
the Tantras of the Four Sisters, then the four, since no other set of four sisters appears in vinivedayet 'When he has completed the Mantra-recitation laid down in the Kalpa he
Saivism, are Jaya, Vijaya, Ajita/JayantT and Aparajita, the sisters of Tumburu, whose should offer it to the deity'. The term is transferred•to the book from its subject; see
cult .is the signature of the Tantras of the Varna Stream ( vii.masrotaly,). One may support Svacchanda 9.lla ... 12a: asya ｭ｡ｮｴｲｾ｝Ｌ＠ puriikhyiito ... tasya kalpa-qt pravak§yiimi 'I
this identification by pointing out that the Srrka'l}-!hfya (see above, n. 4) has among its have taught the Mantra of this [deity ｎｩｾｫ｡ｬｳｶ｣ｨｮ､｝＠ already ... I shall now teach
twenty-four Varna tantras an Ajitiitantra and a Vijayiitantra: (f. 1 7v4-6) udaksrotasi its Kalpa. ｋｾ･ｭ｡ｲｪ＠ comments ad loc.: kalpyata ii.rii.dhyate 'neneti kalpo vidhiinam
viimii.khyarrt caturvirrtsatibhedataly, I ｴ･ｾｩＧｉｊ＠ madhye mahiitantra'!' nayasutra'IJ' nayotta- 'Kalpa means the ritual procedure, [literally,] the means of propitiating [the Mantra],
ram I mahiisarrtmohanarrt ciinyad vf'l}-atantra'!' sikhottaram I ajitarrt vijaya'!' hy anyac. from the verb y'kfp in that sense.
14 Sources and time History tbrough Textual Criticism 15

on Manu as sources teaching rites for the mastering of Vetalas. 13 of prophecy, which would have allowed reference to the past without this
So much for the scanty evidence of the limits before which Tantric consequence, is rare in this predominantly prescriptive literature. 14 So
Saiva literature can be seen to have existed. For evidence of what it one is reduced to trying to get the better of the redactors by identifying
postdates, we must look to the texts themselves. Since they have been elements in .their texts which they probably considered timeless facts but
composed as scripture, that is to say, as transmissions of a timeless revel- whose introduction can nonetheless be dated, if only approximately. Thus
ation, they are less than generous in this respect. Their redactors seeni to there .are elements of Greek astrology and chronometry. in some of the
have been careful to avoid references to historical persons and events that early Saiva scriptures, elements such as the signs of the zodiac ( rasaya}J),
would undermine faith by implyinga termi'fl:lf,S post quem; and the device their subdivisions into horii]J (wpm) and drekkar;Ji]J (bEKavoL), and the listc
ing of the Grahas in the order of their lordship of the weekdays. Texts with
13 Bharuci ad Manu, ed. D. Derrett, 9.290 (prohibiting sorcery). For Bharuci's date
see the edition's introduction. Canonical lists of twenty exorcistic Bhiltatantras·,
the Tantras of the :pascimasrotai} 'Western Stream' of Siva's revelation, are found 14 The few passages known to me in the Saiva Tantras which refer to events in the
in the accounts of the canon given in the Srika1Jthiya and the text prefixed to the
future tense of prophecy concern the history of the transmission of the text in which
Jiiiinapaiicasikii. But I have seen no evidence of any surviving manuscript of any of
they are contained and lack references to datable persons or events. Examples are
these titles,· not even of the Ca'(Lif,iisidhiira, mentioned by the Mrgendra ( CP 1.36ab)
the Kularatnoddyota's ｐ｡ｾｬｳ＠ 8 (pariiparasantiinanir'(Layai}), 9 ( ｣｡ｴｵｲｶｩ､ｨｳｾｹﾭ
as the exemplary scripture of this class. The only early scriptural work of this kind
rasrfmiilama'(Lt)alapiijiividhiinam) and 10 Ｈ｢ｨ｡ｶｩｾｹｰｲｭｮｴｳ｣Ｉ＠ and the
that I have encountered is the Kriyiikiilaguij,ottara, which is"quoted by ｋｾ･ｭ｡ｲｪ＠ (ad
first ｐ｡ｾｬ＠ of the Pic?Lmata. The last is· exceptional in that the Gurus and disciples
Netratantra 19.62c-64b, 69, 172-178b, 178cd, 179a, 179c-180b, 180cd and 182) and
survives in Nepalese manuscripts, the earliest (NAK 3-392) dated Nepaladesiyasa:qlVat to whom it refers seem to be historical rather than mythical, since it gives not only
their names but also the places and castes of their birth, their Vedic Sakha in the
304 (A.D. 1184/5). This covers both exorcism (the domain of the Bhiltatantras) and
case of brahmins, and in some cases the names of their parents: However, it appears
the curing of snake-bites and the like (the domain of the Garu<;latantras). That its
that nothing in this interesting account is datable. It is more common for accounts
exorcistic sections are nilated to what has been lost is suggested by the fact that it
of the transmission or descent of a revelation to be given in the past tense, but then
teaches the worship of such Siva-forms as ｋｨ｡＼［ｬｧｲｶｾ＠ and Cal)<;lasidhara. Tantras
they generally attribute the events to mythical, prehistorical time, such as the begin-
bearing their names are in the canonical Bhiltataritra lists:
ning of the Kaliyuga. I know of one exception. The Yonigahvara (ASB 1000 G), a
text-bricolage in the tradition of the Krama, claims to have been revealed on earth by
Srika1Jthiya, f. 12v5-9 Text before Jiiiinapaiiciisikii, NAK 5-4632, JiiJinanetraniitha (f.lv3): yonigahvaram uddhrtya mantrasadbhiivam uttamam I srf-
. f. 2r8-v1 (vv. 26-28) jiiiinanetraniithena bhiitale sa1)'LprakiiSitam), who probably flourished towards the be-
ginning of the tenth century since he is reported to have been at three removes from
1. *Hela (em. : hola cod.) [1] Halahala Cakrabhanu, the Guru of the Bhiltiraja who was one of the teachers of Abhinavagupta
2. Hayagriva [2] Hayagriva (fl. c. 975-1025); see Arl)asiqtha, Mahiinayaprakiisa, NAK 5-358, f. 119vl-5 (vv. 154-
3. ｋ｡ｲｮｫＯｾ＠ [3] ｋ｡ｲｫｯｾ＠ 157): tataiJ- srimacchivanandanathaguruvara1}'L sada 1 ｬ｡ｫｯｴｲｭｨｪｩｮ｣ｾｵＱｽＧｌﾭ
4. ｋ｡ｾｫ＠ [4] ｋ｡ｾｮｫ＠ Pm1Jamamy aham I srimatkeyuravatyiikhyii pr_thajajiiiinapiiragii I khacakraciiri'(Lf ye-
5. KaroF [5] ｋ｡ｲｯｾ＠ ya1}'L tiim aha1]'L naumi bhaktitai} I mahiivfravaro yo 'sau srfmadviimanasa1}'Ljiiakai} I
6. Mu!f<;\amala [6] Mul)<;lamala devidhiimni sadiiriitjhas ta1}'L vande kramabhiiskaram I tatai} srimaccakrabhiinur deva-
7. Karkota/Karkota [7] ｋ｡ｲｫｯｾ＠ tiinayadesikai} I mahiivratadharo yas ta1}'L vande 'ha1]'L paradhiimagam; arid Jayaratha,
8. Kha<;lgaraval)a [8] Kha<;lgarava(va em. :rna cod.)!fa Tantriilokaviveka ad 4.173ab, p. 193,13-16. Elsewhere in the Yonigahvara a passage
9. Cal)<;la in barbarous Sanskrit appears to report composition by Oghananda, the disciple of
10. Cal)<;lasidhara [9] Cal)<;lasidhara Rupananda, who had received the Krama (Mahanaya) from VIrasiqthasvamini; see
[10] Huqtkara vv. 493-495 (ff. 34v4-35r2): devl srlvlrasi1]'Lhiikhya ｳＮｶ｡ｭｩｮｾ＠ prasphuta bhuvi 1 tadpada-
[11] Hahakara padmayugalat priipta1}'L caivam mahiinayam I srfriipiinandiibhidhii devl datta1}'L tasya
[17] Vimala prasiidatai} I tayii datta1}'L ｳｶ｡ｩｾｹ＠ oghiinandiibhidhiinatai} I datta1}'L pariimrtara-
11. Vikata [18] Vikata sa1}'L mahiisiisanam uttamam I teneda1]'L racita1}'L sarva1}'L yathii priipta1}'L guror mukhiit
12. ｛ａｭｾｲ｡Ａｦ｝ｫＭｊ［ｬ＼＠ I sarnpradiiya1}'L susambaddha1]'L sarahasya1}'L mukhiigamam I racita1}'L gur?Lviikyena sva-
13. Bhiltatrasa [16] ｄｵｾｴ｡ｲｳｫ＠ santiinahitiiya ca. This VIrasiqthasvamini is the Yogini otherwise called Mangala, whom
14. Sikharava [12] Sikharava the followers of the Krama claimed to be the source of their tradition as the teacher of
15. Gharghara [15] Gurghura the Siddha Jiiananetranatha; see MahiinayaprakiiSa (Trivandrum Sansk. Ser. 130)" 7.85:
16. ｓｩｱｴｨ｡ｫｯｾｲ＠ vzrasi1]'Lhiisanastheya1}'L devf paramamangala; Ari_lasiqtha, Mahii.nayaprakiiSa, f. 117v2-3
17. Ghorattahasa ｾＱＳ｝＠ Ghorattahasa (v. 135abc): ida1]'L srivlrdsi1]'Lhiikhyasviiminyii samprakiisitam I srzjiiii.nanetranii.thasya;
18. ｕ｣ｨｩｾ｡＠ [14] ｕ｣ｨｩｾｴ｡＠ cf. ibid. f.ll9r5-vl (v. 152): udg'ithapztha *jii. (em.: yii. cod.) srimanmangalii.nanda·
18 bis. U ｣ｨｩｾ｡＠ nirbharii I sadasadbhiisanii.vesavarjitii tam aha1]'L sraye; ｓｩｴｫ｡ｬＩｾｨＬ＠ Old Kashmiri
19. ｙ｡ｭｧｨｬＩｾ＠ [20] ｙ｡ｭｧｨｬＩｾ＠ Mahii.nayaprakiiSa 9.5ab: harasiru jina gangi avatii.rana tii. janu mangala pzthadisiina
20. ｇｨ｡ｾｯｴｫ｣＠ [19] Mahotkata 'As the Ganges descends to earth from the head-of Siva, so from Mangalathe lineages
of the ｐｉｾｨ｡＠ [U<;l<;liyana]'.
 ｈｩｳｾｯｲｹ＠ through Textual Criticism 17
16 Sources and time

these elements can hardly be earlier than the fourth century A.D., though (c. A.D. 550-700). 18 And the Brhatkiilottara, a rather late, eclectic text
they may be three or more centuries later. 15 influenced by non-dualism, echoes the Spandakarikii of Bhatta ｋ｡ｬｾＬ ＱＹ＠

As for hard evidence of dependence on datable literary sources, I have kalpaniipotf.ham given by Dignaga in the pratyak§aparicchedah of his Pramii'!'asamuccaya
｡ｾ＠ ｾ･ｴ＠
ｬｩｴｾ･＠ to offer. The Matangapiiramesvara paraphrases the Sa:rpkhya- (Masaaki Hattori, Digniiga, On Perception, 1968, pp. 25, 82, note 1.25). For Dignaga's
date see Erich Frauwallner 'Landmarks in the History of Indian Logic,' Weiner Zeits-
kanka ｯｦｬｳｶ｡ｲｫｾＡ＠ (c. A.D. 350-400), 16 and echoes the definition of sense- chrijt fur die Kunde Sud- und Ostasiens 5 (1961), pp. 134-137. Cf. Dharmakirti,
perception formulated in the Pramii'(Lasamuccaya of the Buddhist Dignaga Nyayabindu 1.4: pratyak§am kalpanapotf.ham abhriintam, 'Sense-knowledge is free of
(c. A.D. 48R-540) and ･｝｡｢ｯｲｴｾ､＠ in the Nyayabi'T!_du of Dharmakirti (c. A.D. imagination and non-erroneous'. For Dharmakirti's date see Frauwallner op. cit.,
pp. 137-9. The Matanga also knows the Mahayanist doctrine of consciousness-only
600-660). The Jayadrathayamala echoes the Agamasiistra of GaucJapada
( vijiianamatratii); see VP 6.19c-20.
' 8 This influence is apparent in the nondualistic 'Teachings on the Waveless'
IS see D. Pingree, Ｇｒ･ｰｲｳｾｴ｡ｩｯｮ＠ of the Planets in Indian Astrology', IIJ 8 (1964- Ｈｮｩｳｴ｡ｲｧｯｰ､･ﾧｾＩ＠ found in the third §atkam of that work. (1) Jayadrathayiimala,
5), PP· 249-267; The YavanaJii.iaka of Sphujidhvaja (1978), vol. 2, pp. 197-98. Though $atka 3 (NAK 5-1975, ff. 107vl-118r3), Patala 15 ＨｫｵＧＡｴｦＮ｡ｬ･ﾧｶｲｩ､ｨｾＩ＠ (NU), v.136ab:
Greek astrology entered India through ｙ｡ｶｮ･ｾｲＧｳ＠ Sanskrit translation of an un- kalpayiimy iitmanatmiina'f!' svatantre'!'aiva hetunii/; cf. Agamasastra, ed. Christian
k?own ｇｲ･ｾ＠ work in A.D. ＱＴＹＯＵｾ＠ subsequently rendered into verse by Sphujidhvaja in Bouy (2000) (AS), 2.12ab: kalpayaty iitmaniitmiinam atmii ､･ｶ｡ｾ＠ svamiiyayii; (2) NU
his YavanaJataka of A.D. 269/70, It was not established outside this Hellene-Indian cor- 137c-138b: cijjyotirupam atmana'!l tad aha'f!' sarvadrk sadii I amiitro 'nantamiitriikhyo
ｾｵｳ＠ before t:he ｇｵｰｾ｡＠ period_ (A.D. 320+ ). We see these elements of Greek astronomy miitriibhavantare sthitah; cf. AS 1.12d and l.29a: tur'iyaQ. sarvadrk sadii ... amatro 'na-
I? _the f!•s.vasakanka, the Kalottara, the B;hatkalottara, and major works outside the ntamiitras ca; (3) NU 178cd: ｡､ｶｹｾ＠ sarvabhutii.nii'!l svaya'f!' ｶｦｲ｡ｾ＠ svaya'!l ｰｲ｡｢ｨｵｾ［＠
Saiddha_:rtika corpus: the Svacchanda, the Picumata, the Pingalii.mata and the Jaya- cf. AS l.lOcd: ｡､ｶｩｴｾ＠ sarvabhiiviinii'l!' devas turyo ｶｩ｢ｨｵｾ＠ ｳｭｲｴ｡ｾ［＠ ( 4) NU 182c: ajena
､ｾ｡ｴｨｹｲＺＧｬ［＠ _see, e.g., N•svasakarika-Dfk§ottara, Patala 17 (kalacakraprakaranam); tu aja'f!' *buddhvii (em.: buddhyii cod.); cf. AS 3.33d and 3.47cd: ajeniija'f!' vibudhyate
S/ardhatnsat1kalottara 11.6a (sa'f!'krii.ntiQ., in the same context); Brhatkalottara: gra- and ajam ajena jiieyena sarvajiia'f!' paricak§aie; (5) NU 184a: ｮｩｶｲｴｾ＠ ｳ｡ｲｶ､ｵｾｫｨｩｮＮＧｦＡ［＠
ｨ｡ｹｧｰｾｬ＠ (34th Patala); Svacchanda 7.26c-208 (in the same context); Picumata cf. AS l.10a: nivrtteQ. ｳ｡ｲｶ､ｵｾｫｨｩｮＧｦＡ［＠ (6) NU 196c: *ajam (em.: ajoy cod.) anidram
f. ＳＱｾＴ＠ (m_ ｴｾ･＠ same ｣ｯｮｴｾｸＺ＠ ｲ｡ｳｹｾ＠ ﾧ｡ＧＡｦｨｶｲｪｩｴｾＬ＠ 'the [twelve] vowels minus the asvapnam; cf. AS 4.81ab: ajam anidram asvapnam; (7) NU 238: sarvarambhavini-
mfertile [l!JJ.H] equal the signs of the zodiac); Pingalamata, Pratimii.dhikara, 438-45lc §kranta'f!' citta'f!' yasmin suniscalam I *aninganam (em.: alinganam cod.) aniibhiisa'f!'
(NAK 5-1929, f. 16r6-ll). The last details the iconography of the Grahas. However brahmaiva paramiirthataQ.; cf. AS 3.46: yada na lryate citta'l!' na ca vik§ipyate ｰｵｮ｡ｾ＠ I
when the planets are worshipped in a circle (grahacakram) their distribution around aninganam anabhii.sa'f!' ni§panna'f!' brahma tat tadii.
the circle is not usually in the order of the days of the week with Ra.hu and Ketu at As for the date of the Agamasii.stra, it is quoted by ｍ｡ｩ｟ｬ＼［ｮｭｾｲＬ＠ Sankara, and
the end. We see the Greek order in the B;hatkii.lottara; but in Nisvasakii.rikii.-Diksottara ｓｵｲ･ｾｶ｡Ｌ＠ which provides a somewhat uncertain posterior limit of c. A.D. 700. A more
17.19c-22 (IFP MS T.150, p.160) and Kira'!'a, Patala30 (grahayii.gah) it is Sun (centre certain limit is provided by his being known to ｓ｡ｮｴｲｫｾｩＬ＠ c. 725-788, for whose
[Sunday]), Moon (E ｛ｍｯｮｾ｡ｹ｝ＩＬ＠ Mars (SE [Tuesday]), Mercury (S (Wednesday]), Sat- dates see Helmut Krasser, 'On the relationship between Dharmottara, ｓ｡ｮｴｲｫｾｩ＠ and
urn (SW ｛ｓ｡ｴｵｲ､ｾｹｊＡＬ＠ Jupiter (W [Thursday]), Rahu (NW), Venus (N [Friday]) and Kamala.Slla', in Tibetan Studies, Proceedings of the 5th Seminar of the International
Ketu NE). Worship m the order of the weekdays is seen in the grahayii.gapatalah of the Association for Tibetan Studies, Narita, Vol. 1, 1989, p. 151. A prior limit of c. 550 is
B;hatkii.lottara. · · established if, as seems probable but is not certain, the source of Agamasii.stra 4.24 is
A somewhat later limit is probable for the B;hatkii.lottara, the Kii.lottara in thir- Bhavya (c. 490-570); see Ch. Lindtner, 'Remarks on the Gau<;lapadiya-Karikas (GK)',
teen hundred verses ( Trayodasasatika-) and the Svacchanda because they have in- IIJ 28 (1985), pp. 275-279. For a judicious review of the evidence see Bouy, op. cit.,
ｾ･ｧｲ｡ｴ､＠ the duodecimal and sexigesimal cycles of one and five Jovian Great Years
pp. 17-21.
Into their Yoga. The second of these cycles is unlikely to have entered our literat- 19 See * Brhatkiilottara, ｄｩｫｾ｡ｰｴｬＬ＠ NAK 1-89 (B), f. 41r6: iti yojanikii. kiiryii citspa-
ure before A.D. 500. See Trayodasasatikakii.lottara quoted at Sii.rdhatrisatikalottaravr- ndiinandamandiTe; Yogapatala, B f. 268v3, NAK 5-778 (A), f. 303v3-4: asatya'f!' satyam
tt•, P· 102: ﾧ｡ｴｳＧｦＡｶｲｩｾ＠ proktii ahoriitre'!'a yoginii.m; Brhatkii.lottara NAK 1-89 abhati tiivat sarva'f!' cariicaram *yiivat sva (conj. : yiivatsu B : yiiva'f!'ta A)spandaru-
f. 178r4 Ｈｋ｡ｬｪｩＮｮｰｦｾＬ＠ v. lOcd:' tenii.horii.tramiisii.s ca ､ｶ｡ｳｴｩｾｲＮｨﾷ＠ ｓｶ｡｣ｾ＠ pa'f!' tu na df§ta'!l guruvaktrataQ. I dr§te tasmin pare tattve *asad(A : susad B)riipa'f!'
chanda 7.130c-138, ｔ｡ｮｴｲｾｬｯ＠ ＶＮＱＲｾＭＷ＠ The duodecimal syste,;,_ ｦｩｾｳｴ＠ ｡ｰ･ｾ［ｳ＠ in in cariicaram; Mantrarpai_lapatala, A f. 314r2-3, B f. 179v2: *cit(B :vi A)spandamandi-
the late ｟ｦｩｴｾ＠ century A.D .. m mscnptwns of the Parivrajaka feudatories of the Guptas. re sthitvii hiiniidiinavivarjite j bhiivabhiivavinirmukte dhya*nadhyeyavi( em. : niidheyavi
The sexigesimal_ brhaspatzcakr_am was in use in 530; but it may go back further, since B : niic ca pari A)varjite I etat tat parama'f!' jiiiina'f!' rahasya'f!' te prakii.Sitam. Cf. also
so,me south Indian astronomical tables have it commence in 427; see Renou et a!., Brhatkiilottara, Antaryagapatala, A f. 217r3-5, B f. 190v4-5 (B) [I] with Spandakiirikii
L ｉｮｾ･＠ ｟ｃｬ｡ｳｺｱｵｾ＠ ＨｾＹＸＵＩＬ＠ vol. _2, pp. 725-26. Its earliest attested use in epigraphy is in a 1.23-2.1 [II]:
Mahakuta mscnpt10n dated m the 53rd year of the cycle, which probably = A.D. 602;
see Salomon, Indian Epigraphy (1998), p. 198. II
16 Matanga VP 6.63c-64b (on the Saipkhyas' view of bondage): pradhiinasye§yate
sarga'! sa ca ｾ｡Ｚｖ｟ｴｲ＠ Sa'f!'sthttaQ. I na kascid badhyate loke na ca kascid vimucyate. 1.23 yiim avasthii'f!' samiilambya
ｃｾＮ＠ Sa'f!'khyakank_a 62 (as quoted ad_ Joe. by Bhatta Ramaka'!tha): tasman na badhyate yad aya'f!' mama vak§yati
nap1 mucyate na1va Sa'f!'sarat. kasctt I samsarati badhyate sii vimucyate ca n- ·s - tad avasya'f!' kari§ye 'ham
prakrt*. . . ana raya iti sa'f!'kalpya ti§thati
17 Mt'
a anga, ｾＭ ogapada 4.15c-16a I : amrdesyam · asa'f!'digdha'f!' kalpaniipodhagoca- 24 tam ii.Srityordhvamiirge'!'a
rarr: I ｰｲ｡ｾｹｫﾧｭＬ＠ 'Sense-knowledge iS inexpressible, certain and free of imagination.' miinasasya padaT[t priipya candrasuryiiv ubhiiv api
This, specifically the term kalpaniipotf.ha-, is derived from the definition pratyaksam brahmii.dfniim agocaram sau§umne 'dhvany astamito
18 Sources and time ｈｩｳｾｯｲｹ＠ through Textual Criticism 19

who according to Kalhal).a flourished in Kashmir during the reign of Mantrapitha and the Krama, and of the majority of titles known to us in
Avantivarman (A.D. 855/6-883). 20 all categories, we have to wait until the second half of the tenth century.
So by the beginning of the seventh century at the latest there existed a But, of course, we cannot make the absence of earlier evidence of these
Tantric Saivism of the kind known from the early surviving literature, by divisions and texts the basis for an inference that they were not yet in
which I mean a Saivism comprising both of the two kinds of system whose ･ｸｩｳｴｮｾＮ＠ In other words, while we have established the approximate limits
scriptures the Saivas distinguished as the ordinary (siimiinya-) and the of the development of the Tantric Saiva literature, we have not succeeded
extraordinary ( vaise!}ika-), that is to say both Saiddhantika Saivism and in demonstrating, except in the case of the Brhatkiilottani, that any text
non-Saiddhantika Saivism of some sort. Inscriptions indicate the existence or ､ｩＬｾｳｯｮ＠ is earlier or later thq,n any other. The task before us now is to
of the first and Dharmakirti and Bal).a that of the second. It is quite apply text-critical methods to obtain evidence of such relative chronology.
possible that by the seventh century most of the literature available to The Tantric Saiva scriptural literature is divided in Saiva -sources in
Saiva scholars in the tenth was already in existence. But it is not until the three ways: (I) into Saiddhantika Tantras on the one hand and tht) Man-
beginning of the ninth that we have firm evidence of specific texts. Our few trapitha and Vidyapitha divisions of the Bhairava Stream ( bhairavasrota/;L)
witnesses from that time show us the principal among the Saiddhantika on the other, (2) into Tantras of the Saiddhantika (Urdhva), Varna and
scriptures known later, and also texts of most of the major divisions of ｄ｡ｫｾｩｬＩＮ＠ Streams, and (3) into Saiddhantika Tantras and Bhairavatan-
the non-Saiddhantika tradition: the Varna, the Yamala and, probably, tras.22 In the first classification the Mantrapitha is principally the Svac-
the Trika. 21 For our earliest evidence of certain divisions, notably the chanda and such lost ancillary texts as the Aghoresvar'!svacchanda 'imd
the Rasasvacchanda, and the Vidyapitha comprises the Viimatantras, the
37 prii.1,1odaya'f[' samii.sritya hitvii. brahmii.ttdagocaram
trimu§tyanta'f[' kramii.d vrajet 25 tadii. tasmin mahii.vyomni Yamalatantras and the Saktitantras, the last of these containing texts per-
somasii.ryalaya'f[' yatra pralznasasibhii.skare taining to both the Trika and the Krama. In the second classification, the
tatra sthitvii. yathii.sukha'f[' sau§uptapadavan ｭｵｾｨ｡＠ Vamatantras are put in a class of their own as the Tantras of the Northern
38 tato bhrudvii.ram ii.sritya ｰｲ｡｢ｵ､ｨｾ＠ syii.d ｡ｮｩＮｶｲｴｾ＠
mantravirya'f[' samii.kramet
Stream ( viimasrota/;L, uttarasrotal;L), revealed from the effeminate and gentle
tad ii.kramya bala'f[' mantr< 2.1 tad ii.kramya bala'f[' ｭ｡ｮｴｲｾ＠ Vamadeva face of Siva, while the ｄ｡ｫｾｩｬＩＮｴｮｲｳＬ＠ the Tantras of the South-
sarvaisvaryapada'f[' labhet ｳ｡ｲｶｪｩ｢ｬＮｮｾ＠ ern Stream ( dak!}i'!tasrota/;L), revealed from his ferocious Aghora/Bhairava
face, cover the territory of the Mantrapitha and that of the Tiika-related
36c mii.nasasya em. : mii.na-
masya AB 37b kramii.d B
and Yamala divisions of the Vidyapitha. In the third classification, the
kramii. A 37c somasii.rya Bhairavatantras, conventionally sixty-four in number, comprise Tantras
B : somasurya'f[' A 38a bhro-
dvii.ram B : bhudvii.ram A (tattva'f[' bhavii.ni narasaktisivii.tmaka'f[' yad etat tridhii. sthitam usanti parii.varajiiii.M.
On the level of doctrine it is this that distinguishes the Trika from the other Siva
2o Rajatarangi1,12 5.66. systems. Abhinavagupta, when teaching that the seed-syllable of Para may be HSAUH
21 I am not aware of direct evidence of the existence of specific Tantras of the Trika reports that this (and S-HAU!f) are the forms prescribed in the ｔｲｩｳｯ｢ｨ｡ｶｴｮｾ＠
at this date. But, as I have pointed out elsewhere ('MaJ!<;Jala and Agamic Identity in
( Tantrii.loka 30.28d and Jayaratha ad Joe.). It is at least probable, then, that Ratnakara
was aware of this scripture of the Trika.
the Trika of Kashmir' in Mantras et Diagrammes Rituelles dans l'Hindouisme, Paris:
Furthermore, the Bhairavfvardhamii.naka, the hymn to the Goddess of which some
CNRS, 1986, p. 169, n. 1), Ratnakara shows technical knowledge of the Trika in Hara-
folios are preserved with the Pii.ramesvara codex of A.D. 827/8, seems to know the Trika's
vijaya 47.99:
defining triad, the goddesses Para, Parapara and Apara. The text is corrupt at this
cakrii.sritatrisikhakotigatii.ravindasatkaM)ikii.sanagatam iha kaulikas tvii.m point but the view that the passage refers to these goddesses gains in plausibility from
dhyii.tvaiti saktipatitii.'f[' sthirabhairavankavrtti'f[' parii.parabhedavati'f[' jano the fact that the goddess is identified immediately after it with the signature goddesses
'yam of another Tantric system, the four Sisters of Thmburu (f. [51]r3: tparii.para sakti tva
yiiparii.. parii.parii.tatva jayii.parii.t. jayii. vijayii. ajitii.parii.jitii. The hymn also knows a
By meditating on you here, visualizing you enthroned on the beautiful
Brahmayii.mala, ｖｩｾＡＧｵｹＮｭ｡ｬ＠ and Rudrayii.mala, since it refers to the Goddess as the
centre of the lotus above the cusps of the trident that is within [your]
embodiment of these {f. 53rl): tvam brahmayii.malii. tva'l[l vi§'l,!uyiimalii. tva"[' rudrayii.-
MaJ!<;iala, we Kaulas become Lords of the Powers, attaining the continuous
malii..
state of existence in Bhairava that embraces both the transcendent and 22 Here I exclude from consideration only the peripheral and purely instrumental tra-
the immanent aspects [of his nature].
ditions represented by the iatro-magical {Garu<;la-) and exorcistic (Bhilta-, Bhautika-)
To this I add 47.96, 47.97 and 47.112. I propose that in the first two of these verses Tantras. They are accommodated in the second classification as the revelations of
Ratnakara is giving us encrypted descriptions of the syllables HSAU!f and SAU!f, two the Eastern and Western Streams respectively: Garu<;ia E Ｈｔ｡ｴｰｵｲｾＩＬ＠ Bhautika W
forms of the seed-syllable of Para, the principal goddess of the Tfika. In the third he Ｈｓ｡､ｹｯｪｴｾＬ＠ Varna N (Vamadeva), ｄ｡ｫｾｩｊＡ＠ S (Aghora), and the Siddhantatantras
refers to those who hold that reality has three aspects, namely ｮ｡ｲｾＬＮ＠ ｳ｡ｫｴｩｾ＠ and ｳｩｶ｡ｾ＠ above (Mana).
20 Sources and time History through Textual Criticism 21

that according to the other classifications are (1) revelations of the Man- of transmissions of the text in Nepal and South India, especially since
trapitha, (2) Yamalatantras, (3) the Trika-related. Vidyapitha and (4) the these can be shown to independent of the Kashmirian; and by the pres-
Varna Stream. 23 '· ence of Svacchandabhairava worship to this day among the Newars of the
In all these classifications the special standing of the Svacchanda on Kathmandu Valley. 26 It is therefore of particular interest to determine its
the non-Saiddhantika side of Saivism is clear. For both the classifications 26 See my 'MaiJ.<;iala and Agamic Identity in the Trika of Kashmir', p. '1.82, n. 6B.
whiCh do not mark out the Svacchanda-dominated territory of the Man- Svacchandabhairava is the Agamic identity of the Pacalibhairav in the south of
trapitha as a distinct entity but subsume it within the larger categories Kathmandu, as is evident from unpublished inscriptions at the site. One of 924
of ｴｨｾ＠ Daksina- and Bhairavatantras, nonetheless place the Svaccha)ida at (A.D. 1804) in a pilgrim shelter (New. phalca) facing the front of the shrine refers
to the place as the seat of Svacchandamahabhairava: (ll.1-2): Of!' lv.bharr-. sarr-vat
their head.24 The importance of this text is also indicated by the exist- 924 magha ludi 10 lnsrilri (2) svaccharr-damahabhaimvasthane. Another, 'in a
ence of a number of extensive ancillary Svacthanda scriptures; 25 by the shelter facing the lateral entrance, begins with obeisance to this deity (1. 1: O"Tfl.
attention it received from the learned in Kashmir in the tenth to eleventh nama bhagavate svaccharr-damahabhairovaya) and then refers to the place in the de-
claration ofpious intent ( sarr-kalpa!t) as the seat of Svacchandabhairavabhattaraka
centuries; by the fact that it was the •basis of standard Saiva practice iri (11. ＵｾＺ＠ lrimatkii."f[l.ti(6)nagare sri 3 svacchamdabhairavabhattarakasthii.ne · ihaiva
that region and has remained so •down to modern times; by the existence pv.f.1yabhUmav.). A folio at the end of a Nepalese ｾｳＮ＠ of the ｓｶ｡ｾｨｮ､｢ｩｲｴｯ＠
completed in A.D.1068/9 (NAK 1-224) records the installation of a Svacchandabhair-
23The reason that texts related to the Krama are absent from the second ·and third ava as Lord of the South in A.D. 1184/5, probably referring to Pacalibhairav, who is the
classifications is probably that they predate it. The only ｳｵｲｶｩｾｮｧ＠ Vidyapi¥ha text guardian of the southern district of Kathmandu (f. 163r: samvat 316 ... daksinesva-
that is Krama-related is the Jayadrothayamala and this. character belongs only to ｲｯｳｶ｡｣ｨＷｊＮ､ｭｩ･ｰｴｾＩ＠ Of two inscriptions of A.D. 1819 and 1847 ｲｾ｣､ｩｮｧ＠
the last three of its four sections of six thousand verses Ｈｾ｡ｴｬ｣ｭＩＮ＠ The' first $a¥ka, the dedication of pilgrim shelters in Kupu, about two miles SE of Pharping, the first
to which we owe our only full exposition of the classification into Mantrapi¥ha and begins with a verse of devotion to ·Svacchanda, and in the second, the donor 'refers
Vidyapi¥ha, is a text of the cult of ｋ｡ｬｳＮｲｰｫｾｩｊｉ＠ with no clear Krama superstructure. to the same as his personal deity (II. 2-3: sve§tadevatasvacchandabhairavapritaye ... )
The five-stream classification seen in the Srika'f.lthzya's account of the canon includes {see Hemanta ｒｩＮｉ［ｾ｡＠ and Dhanavajra Vajracarya, Ｇｻｩｾｴｵｬｧ｡Ｍｃｮ＠ bhekako aithii.sika
a number of works with titles resonant of the Krama in a list of subdivisions of the ,samagri (Kirtipur: CNAS, VS 2029, pp. 30-32, 34). Kamalaprakasa Malia reports that
last of its Bhiitatantras, the Ghatotkaca, e.g. Mahii.kalfkroma, Kromabheda, 'Sr§tikrama, these shelters in Kupu are at the temple of Svacchandabhairava (Purvaja Lv.ma7J7.-
Rajavataro, · Mantravataro, Kalikroma, ｋ｡ｬｳｲＭｫｾＧｦＮｩｯｭＬ＠ Sarr-hamkmma, and ka, 1980, p. 7). A puja and goat sacrifice to Svacchandabhairava are offered in Pharping
Bhavasarr-hamkmma. This is probably a secondary attempt to accommodate the new by the Karmacaryas of Kocchutva.J:i during three chariot processions ( rathayatraf) in
tradition while at the same time expressing the view that the Krama is a soteriological the course of the annual Navaratra (see Kulacandra Korrala, Ｇｄ｡ｫｾｩ［ｬ＠ [PharJ>iii]',·
development from within the exorcistic divisio:O. of Saivism. There is much to be said Pracfna Nepala 119 [August-September 1990], p. 30b). There is further evidence of
in favour of this view. the worship of Svacchandabhairava in Nepal until modern times in connection with the
24 A canon of thirty-two Tantras of the Southern Stream (which become sixty-four shrine of Pa.Supati, the national deity: a document records a gift of land in A.D. 1701
since each is supposed to have an -uttamtantra) is given vv.17 ff. of the text pre- to KeSavabhaHa, priest of the Pa.Supati temple, to provide for the puja and pasv.bali/t
fixed to the Jii.anapaii.cii.!ika: svacchandabhaimvarr- caf.1¢a(rr-) krodham unmattabhairo- of Svacchandabhairava (text in Axel Michaels, Die Reisen der Gotter. Der nepa/ische
vam . . . . These first four of the list comprise the four primary Siitras of the Man- Paiv.patinatha-Tempel v.nd sein ritv.elles Umfeld (1994], pp. 331-2); in A.D.1854/55
ｴｲ｡ｰｩｾｨ＠ in the first classification, which we know from the Jayadrothayamala and, in Jang Bahadur issues a letter fixing the rent of the tenants of the land of a Svac-
identical words, from the Sarvavim as quoted by ｋｾ･ｭ｡ｲｪ＠ ad Svacchanda 1.4c-7: chandabhairava of ｎｵｶ｡ｫｯｾ＠ to the NW of Kathmandu (see Dhanavajra Vajracarya and
svacchandabhairoval caf.1¢ai£ krodha unmattabhairavai£ I granthantarli.'f.li catvari ma- 'fekbaha.dur ｓｲｾｨ｡Ｌ＠ Nv.vakotko Aitihasik Ruparekhii. [Kirtipur: CNAS, VS 2032], p.
ntmpiJharr- varanane. Another list is found in the Srika'f.lthiya, according to which 107); a copper-plate inscription A.D. 1801/2 (text in Govind Ｇｦ｡ｩ［ｾ＼ＡｮＬ＠ ｐ｡ｬｶＮｰｴｩｫｾ･ｲｯ＠
there are twenty-four ｄ｡ｫｾｩ［ｴｮｲｳＮ＠ The first is the Mahaghom. This is not ex- Sarr-skrtika Adhyayana, vol. 2, inscr. no. 126, p. 338), recording provisions for the annual
actly a synonym of the Svacchanda, though Aghora = Svacchanda, but rather a class- mahapuja of Pa.Supati, includes the requirements for the puja of Svacchandabhairava
name for a series of 24 Tantra.S among which Svacchanda texts figure conspicuously. and other ancillary deities of the site; a document of A.D. 1915/6 provides for the annual
They are the Lakula (i.e. Lakulasvacchanda; cf. ｋｾｭ｡ｲｪﾷ＠ ad Svacchanda J.0.1131c), sacrifice of a goat each to GuhyeSvari and Svacchandabhairava on the new-moon day of
Aghorelvarisvacchanda, Vidyasvacchanda, Svacchandasara, ' Rasasvacchanda, Rajapv.- Pha.lguna {'faQQan, op. cit., vol. 1, p. 286).
trasvacchanda, Bindv.svacchanda, and Nadasvacchanda. The third classification, that There are numerous Nepalese mss. of the Svacchandatantm: NAK 1-11, 1-43, 1-
of the sixty-four Bhairavatantras, is available to us in two versions, the first in the 114, 1-224, 1-225, 1-177, 5-691, 5-4974, 5-6165, etc. South-Indian mss.: IFP T.
Srika'f.lthiya and the second in the Jayadrathayamala. Though they differ from each 507,1032,1077. Nepalese Paddhatis for the puja and homah of Svacchandabhair-
to some extent, they both have the Svacchandabhairova as the first text, beginning ava following the Svacchanda: Svacchandadevarcanavidhi ( NAK 5.744, 1.386) and
with the verse already cited, in which this text is followed by the Ca1.1¢a-, the Krodha- Svacchandadevalak§ahomayaga (NAK 5.744).
and the Unmattabhairova. The Svacchanda also heads the list of ｄ｡ｫｾｩ［ｴｮｲｳ＠ in [2001:] Other evidence of the strength of this tradition is that of its influence on the
the Picumata (f. 200r, 39.34c-36c); svacchandabhairavarr- devi krodhabhairavam ·eva Saiddhantikas. Since writing this essay in 1997 I have examined mss. of the unpublished ·
ca I v.nmattabhairovarr- devi tatha caivograbhairavai£ I kapalwhairovarr- caiva tatha Siddhantasarapaddhati or Bhojadevapaddhati, the Saiddhantika ritual manual attribute-
jharr-karabhairovai£ I lekharaii. ca tatha caiva vijayabhairovam eva ca I rv.drayamalam d to Maharajadhiraja Bhojadeva (NAK 1-1363 [A.D.1077/8], NAK 5-743 [A.D.llll/2]),
anyaii. .... probably the Paramara scholar-king Bhojadeva, who ruled c. A.D.1018-1060, and have
25See notes 24 and 27. concluded from evidence that must be presented elsewhere that this work drew oi:J.
22 ' ·Sources and time History through Textual Criticism 23

position in relation to the other parts of the Saiva corpus: to Saiddhantika Siddhanta as the teaching of this scripture, as though it knew no other; and
literature on the one hand and to texts of the Vidyapitha on the other. ·· the Guhyasiddhi of Padmavajra speaks of the teaching of the Siddhanta
as following either 'the Kiilottara etc.' or this text. 29 That the· text should
An intimate relation with the former has emerged through comparison
have fallen into obscurity in later times is not surprising. For it contains,
with the unpublished Nisviisa. This work, which is much the largest of the
as we shall see, features that link it closely with the earlier, pre-Tantric
Saiddhantika scriptures, 27 seems to have been of l.ittle importance .when
Saivism, too closely, I suspect, for the comfort of the later commentators,
viewed with the perspective of the Kashmirian commentators .and. their
who received and further moulded a Siddhanta more starkly differentiated
successors, since they seldom refer to it; But there are reasons to suppose
both from that Saivism and from that of the non-Saiddhantika Tantric
that in earlier times it was in the forefront of the tradition. In Cambodia,
systems of which the Svacchanda is exemplary.
where later Indian developments do not appear to have penetrated, it was
the principal authority of the Saivas; 28 the Variihapuriitta speaks of the In its tenth Patala the Svacchanda gives an account of the bhuvaniidhvii,

the Svacchanda extensively and deeply. The influence was passed on into subsequent II (1113-c.ll50) learnt all the Siddhantas beginning with the holy Guhya (G. Coedes,
Saiddhantika practice through the principal manuals of the late 11th to mid-12th ｣･ｮｾ＠ Inscriptions du Cambodge (1937-66] (hereafter IC]1, p. 29); and the Prasat Tor inscrip-
turies, the Pa,ddhatis of Som<Wambhu, Jfian<Wiva ( Jnanaratnavali[IFP MS T. 231 ]), and tion of 1189 or 1195 mentions a commentary composed by a Cambodian brahman (IC
Aghor<Wiva, all of which relied on Bhojadeva's, a fact evident from extensive unsignalled 1, p. 231, v. 5cd): sarp.sarasindhubhuvanoddhara'!'iiya guhyatfkii.patha yad akarod yama-
paraphrases in the first and, in the other two works, openly acknowledged. sadma ｬＧｩＮｮｹｾｭ＠ 'since he emptied the abode of Death by means of a commentary on the
27 See n. 2 for its division into two parts namely the Ni§vasamukha and the four Guhya (composed] to promote the raising of (souls] from the worlds within the ocean
Siitras (Mula, Naya, Uttara and Guhya) and the fifth Siitra, called the Ni§vasakarika of transmigration'. Finally, in the Prasat Khna inscription, probably of 1060 and ii:t
(whic1J, includes the ｄＡｫｾｯｴ｡ｲ＠ at its end). At the end of the latter we are told any case from the reign of Udaya.dityavarman II (105D-1066), the author Phalapriya
that the .whole Tantra in five Siitras is 12,000 verses long (IFP MS T. 17, p.ll52): describes himself as Jrilankarakave!J. praptaguhyajiiana!J. (IC 1, p. 206, v.108), 'one who
ida'T{L pancasutrojjvalarp. samudayena dvada§asahasrikarp. samaptarp. *nilvasakhyarp.- has received knowledge of the Guhya from Saitkarakavi', though it is possible that it
(em. : nilvasakarikakhyarp. cod.) tantram. As we have it, it is about 1,500 short of means only that he received secret (guhya-) knowledge of some kind from him.· ·
this total, the text of the first four Siitras together with the Nilvasamukha in NAK 'l- The hypothesis that the Saiddhantikas of Cambodia remained cut off from the
277 being c. 4,500 verses and that of the fifth c. 6,000, We may not. have the whole text, mainstream once their tradition had taken root is reinforced by .the persistence in
then, But it is possible that 12,000 is an exaggeration, since it seems to have been a Cambodia of the ritual ｴｾ｡､ｩｯｮ＠ of the Pii.rame§vara Ｈ］ｐ｡ｵｾｫｲｰｭ･ｬｶＩＬ＠ an-
｣ｯｮｶｾｴｩ｡ｬ＠ number for large-scale Tantras. Thus we have the 12,000-verse Picumata, other Siddhanta that had lost ground in India by the end of the first millenium.
the 12,000-verse Svacchanda (cited by Hrday<Wiva, Prayaicittasamuccaya, ff. 90r4- Its transcription is mentioned in in A.D. 967 at Bantay Srl!i (BEFEO 25, pp. 359
93r5; Manodadatta, ｋ｡ｬ､ｦｫｾｰｨｴｩＬ＠ Bhandarkar Oriental Research Institute, Pune, and 362, v. 30: parame§varap'ii.rviiii ca yo'likhac chivasarp.hitam -'who had also tran-
Ms. 157 of 1886-92, f. 9r8: iti Jrilalitasvacchandabhairavasarp.hitayarp. dvada!asaha- scribed the Pii.rame§vara§ivasarp.hita') and an inscription from Angkor Wat (A. Barth
sryarp. Jribahurupagarbhastavarajal;l.), the 12,000-verse Aghore§varisvacchanda (cited by and A. Bergaigne, Inscriptions sanscrites de Campa et du Cambodge (Paris, 1885-93],
Hrday3Siva, Praya§cittasamuccaya, ff. 84r2-90r2), and the 12000-verse Anande§vara pp. 570-571, no. 65) shows that it was till followed in the fourteenth century. Thus v. 30:
(Tantralokaviveka ad 15.139; cf.'Tantric Collectanea', Bod!., MS Chandra Shum tasmin kuru mahad yaga'T{L yathoktarp. parame§vare 'There make the great MaQqala as it
Shere e. 264, ff.l(=64]vl-13(=64]vl-13(=74]v4, ending: iti dvada§asahasre *anande§va- has been prescribed in the Pii.ramelvara'; and v. 36: krtavii.n sa mahad yagarp. kii.layagam
re (em.: anande§vara cod.) saptamapatale varrtadhyanarp. samaptam). In any case it iti Arutam / sarasvatfyii.gayutarp. lokapalasamavrtam 'He made the great MaQqala that
is much bigger than any other surviving early Saiddhantika Tantra. The only text of the scripture calls the Ka.Jayaga, including the Yaga of Sarasvati (i.e. ｖ｡ｧｩｾｶｲＩ＠ and en-
this class that approaches it in length is the late Brhatkalottara, which has about 5,000 closed by the Lokapalas.' I take the term kalayagaJ;!. to denote a ｍ｡ｲ［ｾ＠ gala in which Siva
verses. After that comes the Matanga with about 3,500. is worshipped in the form of the year ( sarp.vatsaratma, kalatma), i.e. as embodied in the
28 The Nilvasa's importance in Cambodia is seen in the tenth-century Bantay Kdei VYOMAVYAPIMANTRA when 360 of its 368 syllables, equal in number to the days in the
inscription cited in n. 5. Moreover, the Guhya that is mentioned in Cambodian in- year of the Indian calendar, are distributed along twelve equidistant spokes representing
scriptions in contexts that indicate that it is a Saiddhantika work is, I propose, the the months in a wheel representing the year, with the remaining eight occupying its
(Nilvasa]guhyafGuhya(sutra] that is the major part of the Nilvasa. Thus in the nave. See Svii.yambhuvas'ii.trasarp.graha 6.1-23 and 18.22, reading kii.latinabhyadhikani
Old Khmer section of the Sdok Kak Thorp inscription we are told concerning King with the Nepalese codex (NAK 1-348). The Ni§vii.saguhya teaches this Yaga in de-
Udaya.dityavarman II (r.1050-1066) (4.64-67): tail in its ninth Patala, calling its ritual the worship (yii.ga/1) (of Siva] as the year
(ff. 79v6-80r1): yagavidhi/1 prokta/1 sarp.vatsara§arfri'('a/1. The surviving fragments of
vra/1 pii.da kamraten an thve vra/1 ､ｦｫｾｩＮ＠ darp.nepra gi bhuvanii.dhva vra/1 the Pii.rame§vara (see n. 2) show that the Yaga of that early text does indeed utilize
vrahmayajna thve mahotsava p'ii.jii. toy vra/1 guhya
the VYOMAVYAPIMANTRA enclosed by the Lokapa.Ias and incorporates the worship of
Following the venerable Guhya His ｍ｡ｪｾｳｴｹ＠ undertook the venerable Diksa Sarasvatl, in the form of the fifty elements of the syllabary ( matrka).
of the Order of the Worlds (Skt. bhuvanii.dhvadfk§ii.), the Brahmayajfi;., 29 See Varii.hapurii.'('a, ed. A. S. Gupta (1981), 70.42-43, 71.52-53, but read sid-
. and the Mahotsavapiija. dhii.ntasarp.jnakarp. p'ii.rvarp. rather than nayasiddhantasarp.jnii.bhir in 70.42C and
The Guhyas'ii.tra is indeed distinguished by teaching the bhuvanii.dhvadfk§ii. form of ini- n*Jvii.sasarp.hitii. yii. rather than ni/llvii.sasarp.hitii.yii.rp. in 71.53a, following the citation
tiation. In fact, it knows no other, unlike later Siddhantas, for which this most complex ofthis passage ad Yajiiavalkyasmrti 1.7 in the commentary of Apararka (r. c.A.o.1110-
of procedures is but one option· beside several that are' easier, notably the kalii.dfksii.. 1114) (ed. A pte, 1903, vol.1, p.12). For the text of the reference to the NiAvasa in the
Likewise, the Phnorp Sandak and Prl!.J;i Vihil.r inscriptions state that King ｓｩｲｹ｡ｶｭｾ＠ Guhyasiddhi see above, n. 3.
24 Sources and time History through Textual Criticism 25

the ascending sequence of worlds to be transcended through initiation and padmasutrii.krf;is suklii.
Yoga. Its 1,265 verses are based on the 309 in which the Nisvii.saguhya sets saktibhir bahubhir vrtii.
out its own distinctive treatment of this subject. The expansion is achieved [302] padmahastii. sitii.rigf tu 1232cd svetapadmakarii. devf
largely through elaboration and the insertion of stereotyped descriptive padmiibharar;tabhii§itii. padmamii.lair vibhii§itii.
passages. The essential structure of the Nisvii.saguhya's bhuvanii.dhvii. has saktir bhuvanapariktzbhis
been preserved intact and the verses which convey it are still visible as samantii.d avabhii.sitii.
the backbone of the Svacchanda's text. The redactor has done his job [303]. niidabrahmabile lznii 1235c nii.lj,yii. brahmabile lfnas
with exceptional diligence. But it is still possible to detect his hand and catui},Saktibhir ii.vrf;ii.
therefore the direction of transmission. One example must suffice. lak§ar;takrtiriiper;ta
catur ++ vyavasthitii.
The Nisvii.saguhya's bhuvanii.dhvii. culminates in the world of the god-
[304] ｭ｡ｨｩｰ､ｯｲｾｴ＠ sii.
dess ｓｵｾｭＱ｟｡Ｎ＠ Above that it has only ｎｩｾｫ｡ｬ＠ Siva in formless transcend-
sivakiiyiid viniJ;,srtii
ence.30 But for the redactor of the Svacchanda this arrangement was
sa tu bhutagu'l}ais tyaktii
unsatisfactory. In the style of competitive extension that characterizes
miiyiivayavavarjitii
other new developments in.the Saiva systems he has inserted verses giving
[305] dhyiitavyii ｭｯｫｾ｡ｩｳ＠ tu 1233d dhyiitavyii. sii.dhakii.dibhiJ:t
three additional strata of worlds between Susum1_1a and Siva. ｓｵｾｭＱ｟｡＠ is
ｴｵｾｩｲ｡ｫＧｬｽ､ｨｳ＠ 1232b ｴｵｾｩＮｲ｡ｫＧＨｊＬ､ｨｳ＠
provided with a consort ｓｵｾｭＱ｟･ｳ｡＠ (10.123Q-31b) and above her in as-
cending order are sited the worlds (1) Brahmabila, (2) Saktitattva, and
Sf§tisa'T[!hiirakarttl},'l}ii'T[! 1257cd sarve§ii'T[! kiira'l}ii.nii'T[! tu
(3) Sivatattva. Then comes Samana, the ultimate power, in whom the
kartrbhutii vyavasthitii kartrbhiitii. vyavavasthitii.
universe and its cycles of time come to rest. In making this change the
[306] bibharty a'l}rf,iiny anekiini 1258ab bibharty a'l}rf,iiny anekii.ni
redactor has revealed his hand in two ways. Firstly the verses providing
phaliinfva vanaspatil}, sivena samadhi§thitii
these levels above ｓｵｾｭＱ｟｡＠ are inserted into the passage that describes her
tatordhvan ｮｩｾｫ｡ｬｦｯ＠ devo[
in the Nisvii.saguhya, dividing it into two parts. The first, which merely
vartula sa + ｮｩｾｴｨ｡Ｇｦ｛Ａ＠
describes the form in which the goddess may be visualized, now precedes
[307] anaupamyam aniikiiram
the inserted verses,. but the second, which describes her cosmic function
uhaviidavivarjitam
as the creative power of the universe and is therefore no longer applicable
sarvajiia'T[! sarvaga'f[! deva'f[!
to ｓｵｾｭＱ｟｡Ｌ＠ is put to work at the end of the insertion in the description
svayambhu'T[! bhuvaniidhipa'f[!
of the goddess Samana who has usurped her role. What is more, the re-
dactor has been careful to change one element in the passage of ｓｵｾｭＱ｟｡＠
which his revision has made inaccurate. The Guhya speaks of her visu- 303a ntidabrahmabile cod. (= ｎｩｾｶ｡ｳｫｲＭ S!GLA: N 1 = NAK 1-224; N 2 = NAK 1-182;
alization by liberation-seekers: ｭｯｫｾ｡ｩｳ＠ tu dhyiitavyii (7.305a). This ka 43.85c) : narfya brahmabile Svacchanda K = Ed.; G = IFP MSS. T. 507,1032 and
was appropriate when she was at the summit of the universe. But now 10.1235c ｾ＠ Tantraloka 8.393a 303b catu!y,- 1077. 1232c karii N 1 N 2 K : dhara G 1232d
that she has been demoted meditation on her must be for Siddhi rather saktibhir avrta conj. : catusaktiruddhrtaiy, padmamalair N 1 N 2 : padmamala KG 1235c
than liberation. The Svacchanda appears to have revised the text to take cod. 304a ｭ｡ｨｰ､ｯｲｩｾｴ＠ sa corr. : niirfya brahmabile lfna N 1 N 2 K : vinabhyabrah-
account of this: dhyii.tavyii. siidhakii.dibhil}, (10.1233d). ｭ｡ｨｩＮｰ､ｯｲｾｴ＠ sa cod. 304b viniiy,srta mabile lagna G 1257c kii.ra!'tinti'l!' tu Nl N2 :
Nisviisaguhya 7, ff. 68v5-69rl: Svacchanda 10 (Ed. 's numera- corr. : vinisrta cod. 306b vanaspatiiy, corr. : ktira!'tinti'l!' ca KG 1275d kartr N 1 N 2 K :
tion): vanaspate cod. 307c sarvajiia'l!' corr. : sarv- cakra G 1258a bibharty al'rfiiny K Ｈ］ｎｩｾｶｇｵ＠
[301] tatra madhye sthitii. devf 1229d tatal}, SU§Umr;tiibhuvana'f[! jiia cod. 307d svayambhu'!' (an Aisa form) 7.306a) : vibhaktiral'y N 1 : vibhaktiranyany :
su§Um'(J,ii. nii.ma nii.matal}, SU§Uffi'T}ii. tatra SU'f[!Sthitii. conj. : svayambhur cod. N 2 : bhavanty al'¢iiny G

3 °Cf. Nisvasakarika 43.85ab (IFP MS. T. 17, p. 410) / ｎｩｳｶ｡ｫｲＭｄｾｯｴ＠

19.54ab (ibid., p. 1151): ata!y, ｳｵｾｭＡＧｩ＠ vijiieya sarvadhvanasamaptaye, 'Know that
ｓｵｾｭｬｦＮ＠ follows, to terminate the entire bhuvaniidhvii..' This eliminates the possib-
ility that the lack of a passage in the Guhyasii.tra between the treatment of ｓｵｾｭｬｦ｡＠
and Siva is due to textual corruption.
II
If

26 Sources and time History through Textual Criticism 27

Here the Svacchanda has reworked the text of the Guhya thoroughly. (245] sufddhoj buddha!}, prabuddhas suddha buddha prabuddhas ca
The following is an example of lighter revision in the same Patala: ca
prasantal}, paramak§aral}, prasantal},paramak§aral},
siva§ ca susivas caiva [1110] siva§ ca. susivas caiva
dhruvas cak§ara eva ca dhruvam ak§arasa'f'(l-bhurat
ata urdhva ( 'T'fl-) sthita vidya 1087 ｮ｡ｾｹｶｩ､ｴｫＧｦＨｬＭ devi (246] avyayas ca samakhyata
｡ｾｴ､ｨｩ＠ .prakrtir dhruvii kathayami ataQ, param dasaite nirmitaQ, siva!}, dasaite siva jiieya
(237] ｩｾ｡＠ ca candri7y,f caiva i# ca candri7y,f gaurf kalatattve varanane
gauri §antis tathapara ｳｩｮｴｾｊＬ＠ santikari tatha ｰ｡ｲｭｫｾｪ･ｮ＠
mala ca maulinf svaha 1088 mala ca maulinf caiva samayf labhate pada'f'(l-
svadhii ceti prakirttita (1),) svaha svadhii tatha [24 7] hemabhas sankaraQ, proktaQ, [1111] hemabhaQ, sa'f'(l-karaQ, pro-
(238] [athataQ, ｳ｡｝ｭｰｲｶｫｾｹｩ＠ ｡ｴｯｰｲｩｾ､＠ devese sivas sphatikasannibhaQ, ktaQ, sivaQ, sphatikasannibhaQ,
vigrahan tu yathasthita'f'(l- ｶｩｧｲ｡ｨｾｴｫｭ＠ ucyate ekaikasya ｶｩｮｲ､ｾｴ｡ｭ＠ ekaikasya ｶｩｮｲ､ｾｴ｡ｊＬ＠
karyan duQ,khan tatha jiiana'f'(l- 1089 karya'f'(l- ca kara7y,afi cai- parivare yasasvini parivaro yasasvini
sadhanan tatvam eva ca va sukhaduQ,khan tathaparam [248] kotir eka samakhyata [1112] kotir eka samakhyiita
[239] sadhyaii caiva tathaisvaryan jfiana'f'(l- sadhya'f'(l- ca vikhya- sahasrii7y,i tu ｾｯ｡ｳ＠ sahasra7y,i ca ｾｯ｡ｳ＠
kara7y,aii ca tathii§tama'f'(l- ta'f'(l- sadhana'f'(l- kara7y,a'f'(l- tatha kurmakara7y,i ｳ｡ｲｶ･ｾＧｔｦｬﾭ kurmakarii7y,i sarve§a'f'(l-
ete ｡ｾｴｶｩ､ｨ＠ jiieya proktani bhuvanani tu proktani bhuvanani tu
nigrahanugrahe sthita [249] asadhyekapadam urdhvan (1113] atordhva'f'(l- hariharas cai-
[240] niyatikalaragaii ca 1106 atordhve niyatir jfieyas tato hariharau varau va ragatattve nibodha me
ｳ｡ＧｦＨｬＭｫｾ･ｰＷｹＬ＠ bravfme te ·tasya rudran nibodha me tasmad api dasesanal},
sukhaduQ,khaprada devf Sa'f'(l-sthitaQ, kamarfipi7y,aQ,
subhiiSubhanibandhanf ｾＡ＠ [250] ｳｵｨｲｾｴ｡＠ ｳｵｰｲ｡ｨｾｴ＠ ca ｳｵｨｲｾｴ｡ｑＮ＠ ｳｵｰｲ｡ｨｾｴ＠ ca
__ ,;t
(241] ragas tu raiijakaQ, prokto ,'i, surfipo rfipavardhanal}, surupo rfipavardhanal},
ｶｩｾ｡ｹｮ､ｬｫＷＬｑ＠ • manonmanas samakhyatas
yena sa{'f'(l-sa}rito jantur sumanonmana eva ca manonmano mahavfro
narakiidi§U pacyate [251] mahavfras suvfras ca
[242] vimala'f'(l- ｳ｡ｭｰｲｶｫｾｹｩ＠ vfreso dasamaQ, smrtaQ. vfresaQ, parikfrtitaQ,
jiiiinasargam ataQ, para'f'(l- vamadevas tatha sarvvas
bhavodbhavakaraQ, sarvo tatha caiva bhavodbhavah
vajradehas ｴｲｦｹ｡ｫｾｊＬ＠ [1107] vajradehaprabhus .caiva
(243] prabhur dhata vidhiita ca
kramas ca vikramas tatha dhatara kramavikramaQ,
prabhesaQ, suprabhas caiva suprabhedas ca dasamaQ,
dasaite sankaraQ, smrtaQ,
(244] niyatya'f'(l- sa'f'(l-sthita hy ete niyatya'f'(l- sankaraQ, smrtal},
jiiatavya desikena tu
sankariiS ca samakhyatiiS
1108 yat tad dhrdi sthita'f'(l-
padmam atma tatra vyava-
sthital}, niyatir dalair aha'f'(l-ka-
rah kesarair buddhi karnnike
[1i09] kalatatve siva ｪｦｩｾｹ｡ｬｽＬ＠
sivas caiva nibodha me kathayami samasatal},
 Sources and time History through Textual Criticism 29
28

Samana, Unmana, and Siva in inactive transcendence (Siva!; santa!;) bey-

233a kramas ca conj. : kramasa cod. 1087a ｮ｡ｾｹｶｩ､＠ N 1 N 2 : ｮ｡ｾｩｶ､ｹ＠ K: ond the universe.
240d sumananmana apograph : su + nadinam a G : nanavidya N 6 1087b As the Svacchanda had incorporated and extended the world-hierarchy
nonmana cod. ata(>, param N 1 N 2 : tv ata(>, param K : of the Nisviisa, so the Nisviisa itself had built its system of worlds using
tata(>, param G 1087c candri!'i N 2 : can- materials derived from the yet earlier sources of the pre-Tantric system
drini G : candriki N 1 1088a maulini of the Pasupata Saivas known variously as Lakulas, PramaJ]a[pasupata]s,
N2 : mauninf N 1 : malin> K G Mahavratas, Mahapasupatas, or Kalamukhas. I cannot demonstrate this
1088c ata N 1 N 2 : atha KG • devese by comparison with their scriptures, since their entire canon has disap-
Nl N2 : devesi K 1089b sukhadu(>,kha'f!' peared, with the exception of seven verses of the Paiiciirthapramii'(!a quoted
tathaparam N 1 N 2 : sukhadu(>,khakara'f!' by ｋｾ･ｭ｡ｲｪ＠ ad Svacchanda 1.41-43. But it can be seen from within the
tatha K : sukha'f!' du(>,khan tathaiva ca Nisviisa corpus itself. For the Nisviisamukha comprises accounts of the reli-
G 1106a atardhve N 1 N 2 : athardhve gious systems that it ranks below the Tantric Saivism of the Siddhanta, be-
KG • jiieyas N 1 : ya N 2 : ya K ginning with the mundane Vedic. The highest of these lower systems is the
1106b tasya N 1 N 2 : tasya'f!' K G 1106d Saivism of the Atimarga, which it teaches as having two levels, the first the
bhavadbhava(>, N 1 N 2 : bhavadbhavau KG familiar system of Pasupata observance taught in the Pasupatasiitra, the
1107a dhiitara N 1 N 2 : dhiita ca K : second that in question. It gives a detailed account of its bhuvaniidhvii, and
vidhiita G 1107b kramavikrama(>, N 1 N 2 : by comparing this with that of the Nisviisa itself we can see that the latter
kramavikramamau KG 1107c suprab- is an extension of the former, and that this is a continuity which sets the
hedas N 1 N 2 K : aprabhedas G 1108c Nisviisa corpus apart from all other Saiddhantika Saiva systems. 31 We see,
niyatir dalair N 2 : niyatir dala G : then, a continuous line of development from the Lakulas to the Nisviisa-
tidalairN 1 : tidala'f!' K • aha'f!'kiira(>, Saiddhantikas and then from them to the Svacchandabhairava worship-
kesarair N 1 N 2 : aha'f!'karakesara'f!' KG pers of the non-Saiddhantika Mantrapitha. And this continuity with the
1108d kar!'!'ike N 1 N 2 : kar!'!'ikai pre-Tantric Atimarga, so alien to the Siddhanta of the commentators, is
N4 : kar!'ikam KG 1109a siva jiieya(>, echoed by other features in the practices and doctrines of the Nisviisa and
KG : siva jiieya(>, N 1 : siva jiieya Svacchanda corpora. 32
N 2 lllOc ak§arasa'f!'bhurat N 1 N 2 K The flow of redaction can be followed further, into the literature of
ak§ara sa'f!'bhuna G 1113c suhr§ta(>, K :
31 This post-Atimargic bhuvanadhva is set out in full detail in Nisvasaguhya, ｐ｡ｾｬ＠
suhr§ta N 1 : suhr§ta'f!' N 2 : aprahrta(>. 5 and Nisvasakiirika-D'ik§ottara 19 (prakriyasutram). The demonstration of this con-
G • suprahr§tas K : suprah§ta'f!' N 1 N 2 : tinuity is among the subjects covered in three lectures which I delivered recently in the
prahr§tas G 1114b viresa(>, parikfrtita!J, University of Madras (Ramalinga Memorial Endowment Lectures 1996-97) entitled 'The
KG : vfresa pariMrtita!J, N 2 : viresa(>, Lakulas: New Evidence of a System Intermediate between Paiicarthika Pasupatism and
Agamic Saivism'.
parik<rtita!J, N 1 32 The features of this kind of which I am aware are the following:

Here too we see a degree of rewriting prompted by doctrinal revision. 1. The Nisvasa is unique among the Siddhantas in adding the Pasupata observance
after Vaidika sa'f!'nyasa(>, at the end of its elimination in initiation of the various pre-
But there is also wholesale borrowing from the Nisviisa corpus. Thus Saiva religious rites beginning with conception (garbhadhanam). See Nisvasottara,
the second and third patalas of the Nisvasanaya, except the five intro- Pat ala 3, f. 26r4·-5: garbhapu'f!'savasfmanta'f!' jafnmaj ni§kramaT}a'f!' tatha I prasana'f!'
ductory verses at the beginning of the second have been incorporated by ｣ｵｾ｡ｫｲｭＧｦＡ＠ ca ajina'f!' vratamekhalam I vasasafi capavi:fta'f!' caj da!'rf.a'f!' *sa'f!'dhyri-
the Svacchanda without significant changes as 11.316-12.157. The Svac- -m-uprisanam (em. ｛ａｩｾ｡＠ hiatus-bridger] : sadyii.mupasanam cod.) japa'f!' hama'f!' tatha
tatha vii.nyat tryii.yu§a'f!' cii.bhivadanam I snanam vrataparityii.ga da!'rf.ii.dinli.'f!' ca ho-
chanda's twelfth Patala ends eleven verses after the verses which bring the manam I vivaha'f!' pakayajfias ca haviryajiiii.s ca somakii.(>, I a sahasrii.t tu kartavyii.
third Pat ala of the. Nisvasanaya to an end and with it the topic of the vii.naprasthiintabhaik§ukam I vratam pasupata'f!' (em.: piiSuta'f!' cod.) krtvii. sodhana'f!'
visualization of the reality-levels. As in the case of the bhuvaniidhva the prok§a!'a'f!' tathii. I *nyasa'f!' (em. : jfiii.Sa'f!' cod.) sivahasta'f!' ca; Nisvii.saguhya, Patala
9, f. 77r4: garbhadhanii.dikarmesu yii.vat pii.Supatantikam I satahutyii. visuddhis tu.
Svacchanda has found the Nisvasa's hierarchy insufficiently extended. The 2. It teaches a unique variant. of the Saiva model of deriving the streams of revelation
Nisvasanaya ends with the visualization of Bindu. In its final eleven verses . in pentads from the five faces of Siva in which Tantric Saivism emerges from the superior
the Svacchanda has added intri.lctions for meditation on Sakti, Vyapini, ｉｾ｡ｮ＠ face and the four Pasupata systems from the rest: the Pramil.!fa ( = Lakula) from
30 Sources and time History through Textual Criticism 31

Aghora, the Vaimala from Tatpurueya, the Karuka from Vamadeva and the Paiicartha
the Vidyapitha. For example, the Svacchanda's massive tenth Patala on
( lakulzsottha'Tf') from Sadyojata (12.17c-18): the bhuvanadhva and the first part of its eleventh Patala, dealing with
fSane saivam utpanna'T{' vaimala'T{' puru§a(t) smrtam the topic of the order of creation from the highest to the lowest reality-
18 pramii!ta'T{' hrdayaj ja (taT{' vamade) [va]t tu karukam levels, appear with few divergences as the tenth and eleventh patalas of
sadyac ca lakulzsottha'T{' paiica bhedai], prakzrttitai], the Tantrasadbhiiva, a Vidyapitha scripture of the tradition later known
( ... ) = text now seen in the apograph .alone, due to the subsequent phys- as the Trika. 33 That the direction of redaction was from the Svacchanda
ical deterioration of its ｰ｡ｬｭＭ･ｾＮｦ＠ exemplar. ·
5. The third Patala of the Nisvasaguhya teaches various observances ( vratani) that
18d lakulzsottha'T{' conj. : lakulzsantai], cod. may be adopted by the Siddhi-seeker. Among them are the observances of false
3. The peculiar vocalization 'Hm;n;mN'·(hu¢¢unkarai],), produced by placing the tongue self-condemnation (mithyavratam), the cremation ground (Smasanavratam) and the
against the soft palate, which is required as part of the Pasupatas' daily worship, is Sivaga!fa (ga'(Lavratam):
uniquely preserved in the rituals of the Svacchanda; see Pasupatasutra 1.8: hasitagitanr- 29 gaur mata ca pita trata atithis caiva brahma'(Lai],
ttahu{i{iunkaranamaskarajapyopahiire'(Lopati§fhet; Kam;u;linya ad Joe.: *hu¢¢unkaranama hata me papa[ka]re'(La. caren mithyavratam vratf
( corr. : ¢u'Tf'¢unkaranama ed.) ya e§a jihvagratalusa'T{'yogan ni§padyate pu'(Lyo vr§ana- 30 [+ + + + kapa]lena khapvangfbhasmagu'f}fhitai],
dasadrsai],; Svacchanda 2.182c-3b: hu{i{iunkaranamaskaran k'['tva caiva .tato vrajet / smasane carate ratrau smasanavrata( m) ucyate
agniku'(Ltj,asamzpa'T(' tu arghahastai], subhiivitai],. Rajanaka Keyemaraja ad.loc.: bhakti- 31 nrtyate gayate caiva unmatto hasate bruvan
vaivasyonmi§annadamarsamayo dhvanir mukhavadyaparaparyayo hu¢¢unkarai],; Man- bhasmangz czravasas ca ga'(Lavratam ida'T{' smrtam
odadatta, Kaladzk§apaddhati, f. 39v: hu¢¢u'f}karanamaskaran krtva tadagrasane upavi-
sya. 29a gaur corr. :go cod.
4. The account of the supernatural transmission of the Svacchanda in the Jayad- The first, in which a person accuses himself of the murder of a cow, his mother, his
rathayamala has it taught by Svacchanda to eight Bhairavas (Kankala etc.), and by father or a Brahman guest, is evidently in the tradition of the provoking of unmer-
them to the ｖｩ､ｹ･ｾｶ｡ｲ＠ Ananta. The latter then transmits it from the Pure Universe ited c<;>ndemnation through feigning sin and the like that is required of the PMupatas
to the Impure by teaching it to Srikalftha and ｌ｡ｫｵｬｩｾｶｲＯｰＡｦ＠ The latter (Pasupatasutra 3.1-18). The second, in which one goes about in a cremation ground
teaches it to various sages and among them to his disciple Musa!Indra, who extracted at night smeared with ashes, carrying a skull-bowl and a skull-staff, is, apart from
and taught its essence. $atka 1, f.170r2-7 (36.35-42):
the restriction as to time, the kapalavratam that characterizes the Lakulas and the
35 svacchandabhairava'T{' s'iitra'T(' dviskandham iva agatam Somasiddhantin Kapalikas. The third, in which one smears oneself with ashes, wears
asitangena bhedena svacchandena ca bhairavi rags, dances, sings, laughs and babbles like a madman, is evidently the unmattav-
36 svacchando yai], smrto bhedai], sa vibhedii§faka'T{' gatai], ratam 'the observance of one [who pretends that he] is mad', which according to Ab-
svacchandabhairavaj jiiata'T{' kankaladisitantikaii], hinavagupta's commentary on Bharatanapyasastra 12.85 is the practice of Lakulas in
37 a§fabhir bhairavais tebhyas tato 'nantena veditam the advanced 'Paramayogin' stage of their practice: yadi va paramayogyavasthaya'T{'
suddhadhvatas tatas tasmad yogasaktisamtra'(Lat nakuladarsanapratipannanam unmattavratam apy asti. In the post-Atimargic literat-
38 srrka'(Lfhalakulzsabhyii'T{' dhrta'T{' yogasamzk§aya ure the kapalavratam and the unmattavratam are generally found only in the non-
t sinanirjalatnand!Samahiikalambikadibhii], Saiddhantika systems, particularly in those of the ｖｩ､ｹ｡ｰｾｨＮ＠ The Nisvasa seems to
39 nrlarudramukhodgaravijiiata'T{' tviplu§otmatam be exceptional among Saiddhantika scriptures in not having put them aside.
gautamadimunlndrebhyo vedavidbhyo yathasthitam 33 Three large-scale texts of the ｖｩ､ｹ｡ｰｾｨ＠ survive. In the Yamala division we
40 vistrta'T{' jiianam etad dhi prokta'T{' lakulapa'f}ina have the 12,000-verse Picumata and in the Saktitantra division the 7,000-verse Tan-
tanmadhye 'pi sva.Si§yiiya musallndraya dhrmate trasadbhava and the 24,000-verse Jayadrathayamala. The last, from whose first $atka
41 sa'Tf'hrtya sarabhuta'T{' tu nijagada sa eva tu I have drawn this classification, is permeated in its second, third and fourth $atkas by
yad yena sa'T{'.Sruta'T{' tantra'T{' yac ca yenavadharitam the Krama. The Tantrasadbhava, which survives in eady Nepalese manuscripts (NAK
42 tasya tasya hi namna tat tat tantra'T{' parigfyate 5-445 [A.D. 1097/98], NAK 1-363) and is much cited by the Kashmirian commentat-
ors Abhinavagupta and Keyemaraja, is the only large-scale scripture of that tradition
that has survived intact. We have the Trika's Siddhayogesvarfmata; and we. have the
36a vaj jiiatam em.: vajiiatai], codd. 36b sitantikaii], conj. : sitantikaii],
Malinivijayottara, which claims to be based on it. But we have the first in what is
codd. 38a nandfsa em.: sandfSa codd. 38b mahakala em.: mahakali codd.
evidently a highly abbreviated redaction, most of which has recently been edited by
41a yad yena em. : yad yena em. : yad yeva'T{' codd.
Judit Torzsok (D.Phil thesis, Oxford University, 1999); and the second, though of great
According to vv. 40-41 of this passage there should have should have existed Svac- importance as the base-text .of Abhinavagupta's Tantraloka, is written in a terse karika-
chanda texts in the names both of Lakulisvara and of his pupil Musalindra (/Musu- like style which has left few traces of redactional continuity. A third large-scale Trika
lendra). A Lakulasvacchanda is attested, as we have seen. The second would have scripture, the Trisirobhairava (/Anamaka), survives only in citations. For the Varna di-
had the title * Musulasvacchanda or the like. No such title appears in the literat- vision of the Vidyapitha in the classification of the Jayadrathayamala we have only the
ure. But since Musa!Indra is described here as having extracted the essence of the Vf'l).asikha in about 400 verses ( ed. Goudriaan, 1985): the major Vamatantras, such as
Svacchanda, perhaps the work in question is the Svacchandasara in the list of the the Nayottara, the Sammohana and the Maharaudra have not come to light. However,
Srfka'(Lfhlya (seen. 24). Laku!Isvara is the source of the Lakula tradition, so Musa!Indra there are numerous Varna elements in the first Satka of the Jayadrathayamala (alias
is the source of the related Mausula Pasupata system. See Keyemar;l,ja ad Svacchanda ｓｩｾ｡ｳ｣ｨ･､＠ and Tantraraja), which describes itselr' iO: its account of the canon as a Sakti-
11.7lcd: srrlakule§a.Si§ye'(La musulendre'(La. tantra embodying both the Varna and the Dakeyi!fa streams, as opposed to the Yamalas
 32 Sources and time History through Textual Criticism 33

to the Tantrasadbhava and not vice versa is indicated by the presence of of the work's twenty-nine Patalas. Since then I have had access to a
details in the latter's version that are appropriate only in the system of complete manuscript (NAK 5-445) and now add a further parallel. The
the former. 34 entire fifteenth ｐ｡ｾｬ＠ of the Tantrasadbhava, which_deals with the subject
The Tantrasadbhava was in turn among the Trika works that were of post-initiatory ascetic observance (caryavratam), animating each of its
drawn upon in the redacting of the Kubjikamata, the primary text of the external elements with esoteric meaning extracted through semantic ana-
Kaula Pascimamnaya, a system which in terms of its contents gives every lysis (nirvacanam), appears as vv. 29-171 ｯｦｐ｡ｾｬ＠ 25 ofthe Kubjikamata.
appearance of a being an eclectic version of the Trika accommodating These parallels are of great value to the study of the Kubjikamata since
the cult of a new Mantra goddess and her auxiliaries, including a ver- they enable the critical reader to determine which of the various readings
sion of Svacchanda, the principal deity of the neighbouring ｍ｡ｮｴｲｰｩｾｨＮ＠ transmitted in this highly contaminated recension go back to the time of
I have provided details of the parallels with Trika texts elsewhere. 35 I redaction. It is unfortunate that the editors of the Kubjikamata, though
need add only that at that time I was working from a manuscript of the accepting that their text was the borrower, did not use this means of im-
Tantrasadbhava (NAK 1-363) that breaks off in the course of the tenth proving their edition of the text. 36 Here I shall propose two such improve-
ments to the text where the Kubjikiimata has taken in corrupt readings
and the other Saktitantras (of the Trika etc.) which it defines as purely ｄ｡ｫｾｩｬＩＮ＠ The from its source, in order to illlustrate the point that when dealing with a
Sdok Kak Thoq:t inscription of A.D. 1052 (see n. 5) refers to the Siraicheda as one of the scriptural literature of this kind we must abandon the common assumption
four faces of Tumburu, that is to say, as one of the four principal texts of the Varna of the textual critic that the history of a text begins from an original that
Stream. It is possible that it has the Jayadrathayii.mala in mind; for that frequently
refers to itself as a Vamatantra. But it is more probable, I think, that the surviving
is free of errors and meaningful throughout. Sometimes a new dependent
'Siraicheda' is an transformation of the Varna Siralcheda in a ｄ｡ｫｾｩｬＩ＠ milieu. Note, for scripture came close to this ideal, its redactor approaching the standards
example, that its bhu.vanii.dhvii.locates Tumburu and his pantheon on the level of Isvara. of original authorship in his concern to make perfect sense of his sources,
This presence of Tumburu in the bhu.vanii.dhvii. is unique in the surviving literature, and incorporating unchanged only what seemed to be sound and passing over
it is the clearest of indications of the text's intimate connection with the Varna Stream
'! over which Tumburu presides. But the fact that he is placed on the Isvara level, in the whatever was puzzling in the exemplar or revising it in an honest attempt
Pure Universe but well below its summit, is proof that this is a text which wishes to to restore coherence. But text-production often proceeded at a much lower
incorporate and transcend the Varna branch of the religion. standard of scholarship. In these cases we can see redactors who out of
34 A striking example of this is the reference to the Vidyaraja in Tantrasadbhii.va
incompetence, indifference, or both, used their sources without recognizing
1.159: ekii.litipadii. ye tu vidyii.rii.je vyavasthitii.IJ. / padii. vaf11ii.tmikiis te 'pi *vaf11ii.IJ.
(em.: va111a cod.) *prii"(liitmikiiiJ. (em.: priimii"(lii.tmikii cod.) smrtaiJ.. This is identical or attempting to remedy the manifest defects of the manuscripts through
with Svacchandatantra 4.252, on which ｋｾ･ｭ｡ｲｪ＠ comments: ekasminn api vidyii.- which they had access to them. The result is that the readings established
rii.je naviitmany ekii§itipadii.ni priigvibhaktanltyii yii.ni sthitii.ni tiiny apldiinlm eva ni- by comparing the borrower with its source may be sound only in the sense
f"!lltanltyii Vaf11ii.tmakiini vaf11ii.d ca prii."(laf"Upii.s tadiitmakatvii.t padiiniim. This is the that they are the original readings sanctioned by the redactor. In them-
arrangement for the padiidhvii, the vertical order of the universe as embodied in the
'Mantra-units' (padiini), in the Svacchandatantra. The Vidyaraja, which ｋｾ･ｭ｡ｲｪ＠ selves they may be gibberish. The first of my examples is in the following
rightly identifies as the Navatman, is the last of the eight Bhairavas in the retinue of verse (Kubjikiimata 25.34 [t- Tantrasadbhiiva 15.6]):
Svacchanda. The seed-syllable is given as ＱＡＭｋｾｍｖｌｙｕｉ｜Ｇ＠ in Svacchanda 1.84c-86b (vi-
dyiiriijaiJ. samakhyiito mahiipiitakanManaiJ.) and a Kalpa of the Mantra is taught in
9.49c and following. It is called Navatman, 'the nine-fold', because the seed-syllable yad yad iibhara'T).aT{! tasya yad vii vadati viicayii
(Mjam) comprises nine elements H, R, ｋｾＬ＠ M, L, v, Y, u, 1\'1, though in its application sa carya kathitii tasya mantra§ caiva na saT(!Sayal],
to Adhvanyasa the nine elements are O¥ plus the first eight of these elements; see
ｋｾ･ｭ｡ｲｪ＠ ad 4.102-103c and 5.37c-43b. It is clear that the Tantrasadbhiiva is the
borrower here, because the Tantrasadbhiiva uses the Trika's Parapara Vidya for the SIGLA: T = Tantrasadbhiiva; others as in the editio princeps of the
padiidhvii, and not the Vidyaraja; see Tantrasadbhiiva, NAK 5-1985, f. 34vl: mantrii- Kubjikiimata. 6b yad em.: yam codd., T, ed. • vii vadati T, codd. except
I dhviinaT[' tu. saT['iodhya vidyayii tu. pariiparii / padaiiJ. padiidhvavid ､ｬｫｾｩ＠ kartavyii mama FHJK, ed.: viicayati FHJK • viicayii codd. except G, T, ed.: viiiichayii G
II, vidyayii. Indeed the Vidyaraja is no part of the Mantra system taught by this Tantra.
It owes its presence here to nothing but the inattentiveness of the redactor, who while 6c kathitii tasya conj. [=codd.], ed.: kathito mantre T
incorporating the verse failed to notice the anomaly.
!' 35 See the data in the text of the discussion appended to Teun Goudriaan, 'Kubjika's 36This was no doubt through no fault of Goudriaan's but because the information
Samayamantra and its Manipulation in the Kubjikamata' in Mantras et Diagrammes reached him at a time when their edition was already complete and revision a practical
Ritu.elles dans l'Hindouisme (1986), pp. 163-64. After receiving details of these parallels impossibility. I speak only of Goudriaan here because by that time his collaborator,
I Goudriaan discussed them with due acknowledgement in the introduction to his and
Schoterman 's editio princeps of the Ku.bjikiimata (pp. xii, 15-24), confirming my conclu-
our much-lamented colleague Dr. Jan Schoterman, had passed away. [2001:) On the
value of the parallels with the Tantrasadbhiiva see also my 'Remarks on the Text of the
sion that the direction of the redaction is from the Tantrasadbhiiva to the K ubjikiimata. Kubjikamatatantra' forthcoming in the IIJ (in press).
 34 Sources and time ｈｩｳｾｯｲｹ＠ through Textual Criticism 35

Both ｴｲ｡ｾｳｭｩｯｮ＠ read abhara'(tam ('ornament'). This yields no satis- as an anomalous one-line stanza, numbered 155, and then consider the
factory sense in the context and is, I propose, a corruption of acara'!!am. three-line question an anomalous three-line stanza, numbered 156.
The two ｡ｫｾｲｳ＠ ca and bha are easily confused in early post-Gupta North I turn now to evidence of text-flow between the texts of Saivism and
Indian scripts 37 and the result of the emendation is that sense replaces those of the Paiicaratra. The dating of the sources of the latter is far more
nonsense. We now have the meaning: problematic than that of the Saiva scriptural corpus, because no body of
early commentaries and manuscripts survives. The literature has been
His observance ( carya) and his Mantra are whatever he does transmitted almost ｾｮｴｩｲ･ｬｹ＠ within the ｓｲｩｶ｡ｾｬＧｽ＠ tradition of southern
(yad yad acara'(ta'fJl tasya) and whatever he utters. Of this India and the earliest datable references to many scriptures that have been
there is no doubt. considered to predate that tradition are likewise in the works of South-
Indian ｓｲｩｶ｡ｾｬＧｽｳＮ＠ It is not easy, therefore, to determine which of the
The verse, in keeping with the spirit of the chapter, provides an inner texts considered canonical go back to the early times of the Paiicaratrika
meaning beyond the particularities of ritual for an element of rule-bound tradition and to an origin outside the South. Much reliance has been
observance, in the case of carya through a semantic analysis that reduces placed on the Spandapradfpika of the Kashmirian Bhagavata Utpala, a
carya to the meaning of the verb car 'to do' that underlies it. commentary on a Saiva text but one that draws on Pancaratrika sources. 38
For he has been dated in the tenth century. Certainly he cannot be much
My second example is Kubjikamata 29.41-43b, whose source, I propose, earlier than that since the Spandakiirikii on which he comments is a text of
is Tantrasadbava 15.13-15): the second half of the ninth. But the evidence proposed for the conclusion
that he is not later is merely that Abhinavagupta (ft. c. 975-1025) is not
41 vidyamarge cared yas tu sastradr§tena karma'(tii among the Saiva authorities that he cites. This is hardly conclusive. 39
dhyana'fJl puja japo hamal), samayanii'fJl tu palanam Among the Pancaratrika scriptures that have been considered ancient
42 etad vidyavrata'f!l prokta'fJl nanyatha vfranayike is the Ahirbudhnyasa'f!lhita. Schrader judged it to be "no doubt one of the
vidya jiieya tu yonistha carate dvadasantaga oldest Sarp.hitas extant" and assigned it to the eighth century and Kash-
43 vratasthiine§u sarve§u tena vidyavrata'f!l priye mir.40 I propose that the Ahirbudhnyasa'f!lhita and with it the Lak§mf-
13c ｾｨｹ｡ｮＱｊｴ＠ p'ii.ja EGH, T, ed. : dhyanap'ii.ja ABCDEFJK 14c jiieya
tantra postdate the Pratyabhijiiahrdaya of the Kashmirian Saiva scholar
ｋｾ･ｭ｡ｲｪＬ＠ who flourished c. 1000-1050. My evidence for this proposal is
ABEJHK, T, ed. : jiieyas tu CDF: jiiaya G 14c yonistha codd. : nabhistha
T 14d dvada!antaga conj. [= T]: dvada!antago J, dvada!antage: dvadaianta/ce
that the two scriptures, for all their ｖ｡ｩｾｬＧｽｶ＠ character, are indebted to
E : dvadaiantaro C : dvadaianta reG : dvadaiantagam ABH, ed. : dvadaiantaga
the doctrines and phraseology of that work. The case is especially clear
D : dvada!atuga F 15b priye codd. : smrtam T
with the Lak§mftantra. It draws heavily and transparently on the both the
Sutras and the auto-commentary that make up ｋｾ･ｭ｡ｲｪＧｳ＠ opusculum. 41
It requires no great perspicacity to see that a line has dropped out Nor are these the only instances of its dependence on the Saiva literature
between the first and second lines of this passage. The first line, 41ab, is the of, or current in, Kashmir at that time. The author has also drawn on the
first half of a sentence that is never completed, a relative clause without the 38 0tto Schrader refers to him as ｕｴｰ｡ｬｶｩｾ＠ (Introduction to the Paiicaratra
main clause that must have followed it; and the rest of the passage consists and the Ahirbudhnyasa'T[thita [1916], p. 18). I use the form Bhagavata Utpala on the
of an unbroken sequence of semantically self-contained line-pairs that now authority of the colophon of the Spandapradfpi/ca (ity acarya*bha(corr. bha ed.Jgavato-
tpalaviracita spandapradfpi/ca samapta) and in conformity with many other names of
appear to have their first half in one verse and their second in the next this kind; see, e.g, ｓｵ｢ｨ｡ｾｩｴｶｬｦＴＲＬ＠ 143, 164, 608, 425, 609, 866, 927, 1029, 1242, 1876,
until the end of the passage. The K ubjikiimata's redactor appears to have 2276, 3501, and 3514.
tried to remedy this by giving Kubjika a three-line question at the end of 39 That the Spandapradfpi/ca was composed in the first half of the tenth century was

the passage, so that when Bhairava begins the next topic in answer to her proposed by BUhler (Tour, 1877, p. 79) and followed by Schrader, loc. cit.
40 Op. cit., p. 19.
the text returns to the norm of congruence between sentences and verses. 41 ｌ｡Ｏ｣ｾｭｩｴｮｲ＠ (LT) 6.34-44 +- Pratyabhijii.al].rdaya {PH) 4, 5, 7, and the comment-
The editors do not conjecture the loss of this line and so, it seems, do. not ary on 12 (concerning ｖ｡Ｎｭ･ｾｶｲｩ＠ etc.); LT 6.39 t- PH 7; LT 7.29-30b incorporates PH
see the redactor's remedy. Coming up to the barrier to the numbering 5 while working in PH 1 and the point of PH 4; LT 12.12 +- PH 1; LT 13.21-22b +-
interposed by Kubjika's question (srfkubjikoviica) they treat the extra line PH 1-2; LT 13.23-30b +-PH 4-5, 10-11 and 14; LT 18.18 +-PH 20; LT 43.33-37 +-
PH 11-16. Thus only five of PH's twenty Siitras have not been used, namely 3, 6 and
37 See, e.g., Lore Sanders' Gilgit/Bamyan-Type 2. 17-19.
 36 Sources and time History through Textual Criticism 37

isvarapratyabhijiiakarika of Utpaladeva, 42 the Vijiianabhairava, 43 the Svac-

ndasv.ndaratvat; etc. ｋｾ･ｭ｡ｲｪＧｳ＠ source may be the Mangal<JSloka of Utpaladeva's
chandatantra,44 and the account of creation through the syllabary given by fsvarapratyabhijiiii.viv;ti; see Abhinavagupta, fsvarapratyabhijiiii.viv;tivimarsinf val. 1,
Abhinavagupta in his ParatriT["sikavivara'T}a and Tantraloka, Ahnika 3. 45 p. 3: svecchavabhasitase§alokayatra.tmane nama/: 1 sivaya sasvadacchannasvacchasva-
He has also drawn, in the domain of the Kashmirian Paiicaratra, on the cchandav;ttaye. The expression is also common in the Lak§mftantra; see 8.7cd, 12.4b:
svacchasvacchandacinmayf; 7.2cd, 13.21b: svacchasvacchandanirbharii; 13.24b: sva-
SaT["vitprakasa of Vamanadatta. 46 cchasvacchandacidghanal}; and 14.54d: svacchasvacchandacinmayi.
There are also numerous indications of dependence on the sources of AhSaqt 3.6ab: svii.tantryarii.pii. sii. vi§1}0i} prasphurattii. jaganmayf. Cf. fsvarapratya-
bhijiiiikiirikii. 1.44-45: svii.tantryam etan ... sa sphv.rattii. mahii.sattii. desakii.lii.visesinf.
Kashmirian Saiva nondualism in the AhirbudhnyasaT["hita, though they
Cf. also AhSaqt 49.58d: sphura1_!akhye (voc., in a hymn to ｌ｡ｫｾｭｔＩＮ＠ The term ｭｾｨｩＮﾭ
are generally more attentuated, being perceptible for the most part on the sattii. appears in AhSaqt 9.29cd: nirmalii.nandasar[!bodhamahii.sattii.mayena ca; 14.7ab:
level of distinctive terminology, phraseology and doctrine rather than in yii. sii. saktir mahii.sattii vi§"/}OS taddharmadharmi1_!f, and 16.56a: anii.khyeyii. mahii.sa-
substantial text-parallels. This attenuation may be because the elements ttii.. Cf. 4. 72c. 73b: yii. sa bhagavatal} saktir ahar[!tii sarvabhii.vagii. / ap;thakcii.Ti1_!1 sattii.
mahii.nandamayi: parii..
of Kashmirian Saiva discourse reached it indirectly. 47
The Pratyabhijiiasastra's terminology vimarsal}lfpratyjavamarsal} is reflected in
AhSaqt l.16cd: ｭ｡ｮｩＺｾ＠ jayate pur[!Siir[! paratattvavimarsini and 54.53: nirupii.dhya-
42 LT 18.16 t- !Svarapratyabhijiiiikii.Tikii. (IPK) l.64ab; LT 18.17ab t- IPK l.44a; LT vamarso 'yar[! dehaprii.1_!ii.tyaye sthital} I ahamityii.spadar[! yii.ti sa jfvo nirupaskrtah.
18.17c +- IPK 1.45a. LT 7.18-24 incorporates Utpaladeva's distinctive doctrine of the The Pratyabhijiiasastra's Characteristic equation of the Goddess with · the
four levels of the self as perceiver of differentiation ( miiyiipramiitii), i.e. sii.nyapramiitii, svatantryasaktil} of God is seen in AhSaqt 3.6. The equation of the Goddess with the
prii.1_!apramii.tii., buddhipramii.tii. and dehapramii.tii, for which see IPK 1.53-57, 3.8, 9, 19, God's a_harr>tii (piiT"/}iiharr>tii), which is one of the most obvious imprints of the Kash-
22-26. mirian Saiva tradition on the ｌ｡ｫｾｭｩｴｮｲＬ＠ also appears in AhSaqt: see e.g., 3.43ab: sa-
43 LT 43.79-83 +- Vijiiiinabhairava 69-73. rvabhii.vatmikii. lak§mir ahar[!tii paramatmanal}; 4. 72cd: ya sii. bhagavatah saktir ahamtii.
44 LT 43.30c-31b t- Svacchanda 4.313. sarvabhii.vagii.; 6.1: yo 'sau nii.rii.ya1_!o deval} paramiitmii. sanii.tanal} I aharr>bhavatm:ikii.
45 See LT 19. saktis tasya taddharmadharmi1_!i; 18.6: iyar[! sii paramii. saktir ahar[!teyam ｨ｡ｲ･ｾｽ＠ para;
46 LT 14.9c-10b t- Sarr>vitaprakii.sa (SP) 1-2; LT 14.11-12 t- SP 3-4; LT 14.13-14 t- 3.15cd: parii.ha(n)tii.*sva (carr. : sv. ed.)rii.patvan mantramata praki:rtita.
SP 9-10; LT 14.15 t- SP 39 and 41; LT 14.19-30 t- SP 43-61; LT 14.32ab t- SP 62ab; The distinctive term nirmii1_!asaktil} in AhSaqt 3.42d echoes fsvarapratyabhijiiii.kiirikii.
LT 14.38-39 t- SP 77-78; LT 4.42, 25a,26 t- SP l03-104c, 105. A critical editio prin- 2.8ab: ki1[1 tv. nirmii.1_!asakti!J. sapy evarr>vidu§a iSitul}.
ceps of the Paiicaratrika Prakaral).as of Vamanadatta (Sarr>vitprakii.sa, Atmasaptati, Another distinctively Kashmirian usage is that of the pair unmesah and nimesah as
Vikalpaviplava, Vidyii.viveka, VaT1_!avicii.ra, Paramii.rthaprakii.sa) is being prepared by the expansion and retraction of the powers of God. See ｓｰ｡ｮ､ｫｲｩＮｾ＠ 1.1: ｹ｡ｳｾｮＢＬＮ･＠
Prof. Raffaele Torella. This Vamanadatta tells us that he is the son of Devadatta (son ｾ｡ｮｩｭ･Ｎ｢ｨｹｲ｛Ａ＠ jagatal} pralayodayau. Cf. AhS 3.6: svii.tantryarii.pii sii. vi§1}Di} prasphv.-
of Ratridatta) and RatnadevT (Sarr>vitprakii.Sa 153: rii.tridattir devadatto ratnii.devyii.r[! ratta jaganmayf I v.ditiinuditii.kii.rii nime§onme§arii.pi1}f; AhS 60.6: ﾧ｡ｾ､ｨｶｩｹｴＮ＠
yam ii.tmajam I lebhe vii.manadattii.khyarr>) and therefore should not be confused with ｾ｡ｴｪＮ､ｨｶｩﾧｹｭｫ＠ I ｮｩｭ･ｾｯﾧ｡ｲｰ＠ sii. bhii.vii.bhavalaksanii..
his Saiva namesake, the author of the Bodhavilii.sa and the Svabodhodayamaiijari, Note also the use of the term spandal} / spandanam for the Ｂ｣ｲｾ｡ｴｩｶ･＠ force of conscious-
since that Vamanadatta's father was ｈ｡ｲｾ､ｴ［＠ see Svabodhodayamaiijari 44 (Ben- ness in AhS 3.39: so 'yar[! sv.darsanarr> nii.ma sarr>kalpal} spandanatmakal} 1 vibhajya
ares Hindu University, MS. 3F /32400): ｨ｡ｲｾ､ｴｳｹ＠ sii.nunii. I k;tii. vii.manadattena bahudha rii.par[! bhave bhave Ｇｶ｡ｴｩｾｨ･［＠ 21.14ab: saccidanandasa,r[!doham aspandaspa-
svabodhodayamaiijarl. ndalak§a1_!am.
47 My evidence that the Ahirbv.dhnyasarr>hitii. (AhSaqt) postdates ｋｾ･ｭ｡ｲｪ＠ and The Ahirbudhnyasarr>hitii's account of the origin of the universe as the unfolding of the
other Kashmirian Saiva sources is as follows. letters of the alphabet in Adhyaya 16 and the terminology of that account are derived
AhSaqt 3.8d (riktii. pii.T"/}ii. ca sarvadii., concerning ｖｩｾｬＩＮｵＧｳ＠ Sakti) +- Urmikav.la quoted from Kashmirian Saiva sources (Tantrii.loka, Ahnika 3, 66-17lb; Parii.trimsikiivivarana
by ｋｾ･ｭ｡ｲｪ＠ in Spandanir1_!aya pp. 48-9: ( lelihii.nii. [i.e. riktii.] sadii devi sadii. pii.r1_!ii. and Pariitrirr>sikiilaghuv;tti ad vv. 5-8; Sivasutravimarsini ad 2.7, etc.)· Note that ·in
I ca bhii.sate); ｋｾ･ｭ｡ｲｪＬ＠ Sivasii.travimarsini (SSV) ad 1.6 ( atiriktii.riktatadv.bhayii.tmata- 16.86 AhSaqt refers to the sounds YA, RA, LA and VA as the dhara1_!ii/: and in 16.84 to
yii.py abhidhiyamii.nii.pi anetadrii.pii. anv.ttarii. parii. sviitantryasaktil} kii.py asti yayii. sva- the sounds SA, ｾａＬ＠ SA, SA, HA and ｋｾａ＠ as the Brahmas. This terminology is peculiar to
bhittav. mahyullii.siit prabh;ti parapramii.t;visrii.ntyantarr> ｳｲｩｭ｡ｴｾｹＮ､ｫ｣ｰｨﾭ Parii.tri1[1sika 7: vii.yvagnisalilendrii1_!iir[! dhii.ra1_!iina1[1 catv.§tayam I tadii.rdhve sii.di vi-
ra1_!ii.trmii. ｫｲｦｾ･ｹ｡ｭ＠ iidarsitii.. Cf. also AhSaqt 60.5ab: nil}saktii.saktasanmii.trii. piir1_!ii khyii.tarr> purastad brahmapaiicakam. Other Kashmirian Saiva features in the AhSaqt's
riktii. rtambharii.. account are:
AhSaqt 7.67d: *svaccha(em.: svacchal} ed.)svacchandacinmayal}. The expression sva- (i) its identifying the alphabet-sakti{> (in the form of the first fourteen vowels [a slight
cchasvacchandacid- occurs so frequently in the works of Ksemaraja that one may con- difference here]) with ｫｶＮＢＨＧｾ｡ｬｩｮｦ［＠ cf. Tantriiloka 3.220c-2lb (paraphrasing and develop-
sider it a hallmark. See, e.g., SSV ad 3.5: Ｍｳｶ｡｣ｨｾｮ､ｩｴｭ［＠ ibid. ad ing a passage from the Siddhayogesvarimata, which is given in full in Jayaratha's com-
3.25: paripiin_J.asvacchasvacchandacidanandaghanena sivena; PH p. 3: bhagavatf sva- mentary): siitra ｫｶＮＢＨＧｾ｡ｬｩｮｦ＠ bfja1[1 jfvabhutii. cidii.tmika I tajjar[! dhruvecchonme§ii.khya1[1
cchasvatantrarupa; Netratantroddyota vol.l, p.l: cinmayamahal} svacchasvatantra- trikar[! vaT1_!as tatal} ｰｵｮ｡ｾｽＮ＠
sphurad; ibid. pp. 56: svacchasvatantracidekaghanatvat; ibid. p. 73: svacchasvaccha- (ii) its equating the fifteenth and sixteenth vowels, anusvii.ral} and visarga{>, with
ndamahimnii.; ibid. val. 2, p. 23: svacchasvacchandacidii.nanda-; ibid. p. 186 svacchasva- Sar[!hiira{> and ｳｲｾｴｩｻ＾［＠ cf. Tantraloka 3.110-134b ( anusvii.ral}), 136c-l43 ( visargal}), 201c-
cchandacidghana-; ibid p. 272: svacchasvacchandaprakasatmasiva-; Svacchandoddyota 202b, 5.68ab and commentary.
val. 2, p. 128: svacchasvacchandacidanandaghanai}; ibid. p. 245: svacchasvacchanda- (iii) its assimilation of the alphabet-emanation to self-awareness Ｈａｾ＠ HA ｾ＠ M/¥) at
cidanandaghanam; val. 3. p. 98: svacchasvacchandii.tmakacidghanatavyaktir eva yan 16.64-66. This is the central doctrine of Abhinavagupta's treatment of the alphabet;
ｭｯｫｾ｡ｬｽ［＠ ibid. p. 99: svacchasvacchandadhii.ma-; ibid. p. 139: svacchasvacchandacidana- see Tantraloka3.200c-204b.
 Sources and time History through Textual Criticism 39
38

I
l. It might be thought that this evidence of the influence of Kashmirian is also evidence of movement in the other direction, from Paiicaratrika to
I Saiva authors on the two texts indicates that they were written in Kashmir,
so that to that extent at least Schrader's assessment of the Ahirbudhnya-
Saiva literature, as in the case of the Saiddhantika Saiva Brhatkiilottara
which has acquired its detailed perscriptions governing the. cremation of
samhitii. would have been vindicated. But it is extremely improbable that initiates ( antye§ti/:L) and the Sraddha ceremonies to be offered thereafter
from the Paii.caratrika Jayiikhyasamhita. The 89 verses of its antestimrtod-
either text was Kashmirian in origin and probable, if not certain, that both
were composed in the South, where they are first cited ir{ learned literat- dhii.rapatalal}, and the 64 verses of. the sriiddhapatalal}, that ｦｯｬｾ＠ are it:
ure.48 For both give Paii.caratrika interpretations of Yajurvedic Mantras; a redaction of the 105 verses of the 24th Patala of the Jayakhya and of
and these are Mantras of the Taittiriya rather than the Kathaka recen- 72 verses of the 23rd Patala, from verse 64 to its end. The Saiva re-
sion. No one preaching to the community of Kashmirian Brahmins at this dactor has attempted to eliminate traces of his material's ｖ｡ｩｾＱ｟ｶ＠ origin,
period would be likely to encode the Yajurveda in other than its Kathaka but he has not been sufficiently diligent. In the Brhatkalottara, as in the
recension; and the stronghold of the Taittiriya is, as is equally well known. genuine Kalottara recensions among which he wished his text to be ac-
cepted, the recipient of instruction is Siva's son Kartikeya, whereas in the
the Dravidian South. 49
Jayakhya it is the sage Narada. He has therefore made an effort to replace
But the text-flow is not only from Saivism to the Paii.caratra. 50 There
with suitable metrical equivalents the not infrequent vocatives addressed
AhSam also shows knowledge of the Vijiianabhairava and the Malinivijayottara, both to that sage. For example nii.rada (24.48, 24.54 etc.) is replaced with
of which. are Trika scriptures central to the Kashmirian tradition: §a7Jmukha, dvijottama (24.21d) with sikhidhvaja, and dvija (24.100) with
(i) AhSarp. 54.31c: mukhaTf' saktil]. ｳ｡ｭｵ､ｩｾｴＮ＠ Cf. Vijiianabhairava 20d: saivi mu- vatsa. But one dvija has escaped his net (Brhatkalottara, Sraddhapatala,
kham ihocyate. v. 788ab [=Jayakhya 24.97 cd]):
(ii) AhSaryt 8.29cd: ya sa saktir jagaddhatu!). kathita samavayini. This equals Mali-
nivijayottara 3.5ab. The line is much quoted in the non-dualistic Saiva literature of
Kashmir; see, e.g., Tantralokaviveka 1 (1) 16; Sivasutravimarsinf ad 1.222 and 3.19. iinayeti dvidhii yojyarp, tatsarp,jiiiirp, tad anu dvija
48Schrader (1916:18) claimed that the Kashmirian Bhagavata Utpala ＨＧｕｴｰ｡ｬｶｩｾﾭ
I_lava') knew the AhSaryt since he quotes a verse of that work in his Spandapradipika. SIGLA: K = NAK 1-89; J = NAK 5-778; Jayakhya =consensus of ed. and
But Utpala does not attribute the verse to this source; as Schrader himself reports, it NAK 1-49 (Nepalese palm-leaf ms. of Jayakhya, f. 81 v1-2). 78a ana yeti
is at best a variant of a verse in the AhSaryt; its content does not tie it to this text; and em.[= Jayakhya] : anayanta KJ 78b tatsaTf'jiiaTf' em.[= Jayakhya] :
finally, as Freidhelm Hardy has rightly pointed out ( Virahabhakti. The Early History ta!).sthaTf'jiiaTf' KJ dvija em.[= Jayakhya] : dvija!). KJ
of ｋｲｾｴｷ＠ Devotion in South India [Delhi, 1983], p. 35, n. 102), a single verse cannot
establish a lower limit for the whole work. It is, of course, only too likely that a text as Kartikeya, of course, cannot be addressed in this way.
late as the AhSaryt is incorporating the verse from one of the many early Paricaratrika
sources now lost to us.
The redactor has also edited out certain elements ofthe text which went
49LT 29.18 ff. refers by name to each of the Mantras of Taittirfyara"l)yaka 3.1-10. against Saiva doctrine, sometimes omitting the passage and sometimes
AhSaryt gives them in full in its chapters 58 and 59. rewriting it. Thus Jayiikhya 24.9 says that the bier may be carried to the
50[2001:] The ritual systems taught in the Saiva and Paiicaratrika Sarythitas resemble
cremation ground by initiates or non-initiates and again in 24.74 that those
each other so closely in morphology and syntax that they have the appearance of two
dialects of a single 'Tantric' language. There is no reason to suppose that they have obliged to perform a person's funeral rites may be initiates or non-initiates.
a common source that is neither Saiva nor ｖ｡ｩｾＱ｟ｶＮ＠ It is therefore probable that Both statements offended the redactor's convictions and he has changed
one has fashioned itself in the guise of the other, as, for example, from the tenth the texts accordingly. 24.9b's dfk§itair viipy adfk§itail}, 'by initiates or non-
century onwards the Tibetan Bon religion produced a canon of scripture and a body
initiates' has become dfk§itair na hy adfk§itaiJ;, 'by initiates and not by the
of practice on the model of those of mainstream Tibetan Buddhism (see Per Kvaerne,
The Bon Religion of Tibet [1995], p. 10), and as the Indian Buddhists had produced uninitiated' (Ante?timrtoddharapatala 8d) and 24.74b's karyarp, ciidik§ite§v
by the tenth century a Tantrism that was not only heavily dependent on Saiva models, api 'it may be performed for the uninitiated also' has become karyarp,
for that had been the case from at least the seventh century, but, as we shall see in nadfk§ite kva cit 'it should never be performed for anyone who has not been
the last part of this essay, included scriptures that had incorporated large expanses
of text directly from Saiva sources. However, when I wrote this essay in 1997 I was
initiated' Ｈａｮｴ･ｾｩｭＧ｛Ｂｯ､ｨＮｲ｡ｰｬ＠ 55b). 51 But on the level of content too
not ready to throw light on the question of which was the imitator in the case of the he has occasionally been superficial. Thus, speaking of those who may be
Paiicaratrika and Tantric Saiva systems. Since then I have found evidence, which will
have to await publication elsewhere, that all three of the early Paiicaratrika Sarythitas years after our earliest evidence of Saiddhantika Saivism. Since the Paiicaratra existed
known as the 'three jewels' (see Jayakhya, ed. Krishnamacharya, p. 8, adhikal]. pathal]., in some form throughout the Christian era, I venture the hypothesis that its surviving
vv. 2-3), namely the Jayiikhya, the Sattvata and the ｐ｡ｵｾｫｲＬ＠ were written under the scriptures are the result of a late process of tantricization under Saiva influence. We
influence of Tantric Saiva models; that the first two at least were composed following may think, perhaps,_ in the case of both Buddhist and ｖ｡ｩｾＱ｟ｶ＠ Tantrism, of a process
models that belong to a later stage in the development of Saiddhantika Saiva ritual; of acculturation to Saiva norms.
51 The Nepalese ms. of the Jayakhya, NAK 1-49, confirms the edition's reading of
and that neither of these two is likely to have been composed before c. 850, some 300
.1, .'.','
:'i·.:
,,

i,:
'''
I

40 Sources and time Hist()ry through Textual Criticism 41

ll
ｾＡ＠
iL summoned to take the offerings in Sraddha, Jayiikhya 23.91 adds the option · He sh.ould first thrice contemplate Siva, Mahesvara, as having
priigukta'f{L yatipiirva'f{L vii vai§'[Lava'f{L tv ekam eva hi 'Alternatively, (he may the great ash on his forehead, embodied in the pyre. *There- ,
summon] a single Vail?J?.ava, (one of] the aforesaid beginning with the Yati'. after (?) he should meditate o'n the Millamantra in combinac
The Saiva redactor rewrote this as priigukta'f{L yatipiirva'f[! vii sivabhaktam tion with the Milrtimantra.
athaikakam. 'Alternatively, (he may summon] a single devotee of Siva,
(one of] the aforesaid beginning with the Yati'. A Saiva reading this, and The Saiva redactor faced a number of difficulties. The terms mahiivibhiitil}
no doubt the redactor himself, would have to understand the expression and §iir}gu'[Lyam are Paiicaratrika; He kept the first, probably intending
yatipiirvar[! 'beginning with the Yati' to refer to the well-known sequenceS>' vibhuti- to be taken in the Saiva sense of the 'ash' with which Siva '(and
the celibate student, the married householder, the hermit and the ascetic' Saivas) dust their bodies in general and their foreheads ( bhiilam) in par-
(yatil;), in reverse order. But in fact his source is referring to a series of ticular. But §iir}gur;,yasarzram was intractable. Hence the substitution of
specifically ｖ｡ｩｾｊ＿Ｎｶ＠ practitioners that it had expounded at some length citisar'iram, which may mean 'embodied as consciousness' but more prob-
in the immediately preceding Patala (22.6-64b), a fact indicated by the ably means 'embodied as the pyre', since that is more appropriate to the
words priiguktam 'aforesaid'. We might have been none the wiser had the context. His solution is obviously a makeshift, since it infringes the rule of
redactor not failed to notice the inappropriateness of this detail in its new caesura between quarters. The awkwardness of his sivam for amalam and
location, for that lacks any applicable prior enumeration. mahesvaram for paramesvaram,, so that there are now two names of the
Another example of unconvincing revision is seen in the Ante§timrf;od- one deity in apposition, only adds to the impression of incompetence. If
j dhiirapatala, v. 77, in a passage on the cremating of a' simulacrum when the any doubt remains it should be removed by observing his handling of the
body of the deceased has been destroyed or cannot be found. The source final line. The instruction to meditate on the Mu.Iamantra combin'ed with
l:i,,. the Milrtimantra makes perfect sense in the Jayiikhya; for vv. 61c-69 of
is Jayiikhya 24.95b-96b:
'II:
,, Patala 6 of that work have explained that the Mu.lamantra may be used on
'I its own or, to greater effect, in combination with the Murtimantra. The
Iｾ＠ .......... , .. : .. ｾ＠ . tridhii smrf;vii puriimalam
mahiivibhiitisiifjgu'[Lyasarirar[! parainesvaram
redactor, who would have been well advised to add this section to those too
intractable to be included, has substituted an unprecedented combination
96 tadante miilamantrar[! ca sar[!smaren miirtisar[!yutam
of the Saiva Mu.Iamantra with the five Brahmamantras.
SIGLUM: N = NAK 1-49, f. 81r7-vl. 95cd mahii.vibhii.tisii.rf,gv:rJyasarfra1JI Seeing the artificiality of this last piece of ｲ･､｡ｾｴｩｯｮ＠ one is bound to ask
parame!varam N : mahii.vibhii.te sii.rj,gu'l}yasarfra paramesvara ed. 96a man-
oneself whether this part of the new text ever supported a corresponding
traii ca carr. : mantras tu ed.
practice. It seems to me probable that it did not and that such blanks
may have been a common feature of our scriptural literatures, apredictable
He should first thrice contemplate the Supreme Lord as spot- consequence of the method of redaction itself, in which materials are taken
less, embodying the Great Glory and the Six Qualities. There- over and edited to fit the context of a new revelation. It would have been
after he should meditate on the Mu.Iamantra in combination only too easy to overlook details in one part of the text that had become
with the Murtimantra. incongruous as the result of revision in another, and this would have been
even more so when the new work was expected, as in this case, to cover
This becomes Ante§timrf;oddhiirapatala 75c-76: not only the areas of innovation that motivated its production but all the
main elements of religious practice, and in some detail. Moreover,we, may
.................. tridhii smrtvii purii sivam suspect that the need to remove,such deficiencies pressed less insistentiy on
mahiivibhutibhiila'f[! citisarirar[! mahesvara'f[! the minds of the manufacturers of divine revelation, since they themselves
tadante mulamantrar[! tu sar[!smared brahmasar[!yutam would not be held responsible for their work._ Thus it is that while there
SIGLA: K =
NAK 1-89; J =
NAK 5-778. 76a bhii.laf!l conj.: pii.la1J1 KJ 76c
are some scriptures that are just as coherent as works of good scholarship
tadante tentative conj.: hrdante KJ claimed by authors as their own, there are others that are little better than
careless bricolage.
24.9b. But for 24.74b it has kii.ryaii ｣ｩＮ､ｦｫｾｴｯｳｶ｡ｰ＠ (f. 80v4), which I have emended
to kii.ryaii ｣ｩＮ､ｦｫｾｴ･ｶ＠ api., The edition has kii.ryas ｣ｩＮ､ｦｫｾｴ｡ｲ＠ api. This too is corrupt,
I end with the subject of ｴ･ｸｾ＠ flow from the Saiva Vidyapitha into the
since it voids the passage of its intended collocation of the procedures for the initiated scriptures of the Sarpvara cycle of the Buddhist Yoginitantras. In the
(v. 73a: ､ｦｫｾｩｴＮｮｭＩ＠ and the uninitiated (v. 74). Laghusar[!vara (Tib. Bde mchog nun nu), which the Buddhist tradition
42 Sources and time History through Textual Criticism 43

identifies as the root-text (miilatantram) of this corpus, long passages, Patala 17, 1-11 YS, samay6.ciirace§!6. 12 ＱＰｾ＠
amounting to some 200 verses, nearly a third of the whole, can be seen to ( Sarvayoginf rupa- vidhiinapajalai},
(parivartanacihnamudra-
have been redacted from Saiva originals found in texts of the Vidyapitha pat alai}) 138c-148
division, namely the Picumata, the Siddhayogesvari:mata, the Tantrasa- =AUT 37 [39:2]. 5 et sqq.;
dbhiiva, and the Yogin'tsa'f{!ciiraprakara'T}a of the 3rd ｾ｡ｴｫ＠ of the Jayadra- 3-11= SarppT 4, 2.1-9
thayiimala, or, perhaps, from closely related versions of these text-passages
Pat ala 18 ( Sarvayogin'i- Related to a passage in 19 19
that are no longer accessible. The correspondences I have identified are as var7Jalak§a7Jacihnavidhii],) TS Adhikara 16
tabulated in Table 1. 52 1-4 =AUT 37 [40] (= MMT (yogin;;/ak§a7Jam)
15) (J;Jakinilak§a7Jam). 1-4

TABLE I Patala 19 (Sarvayoginyo- SYM Patala 29 (unnamed). 24 23!

mudranirdesavidhii}). 1-24b 22c-33, 35-40,
(24 vv. in patala) =AUT 43-46b, 48-50, 52. CST
ABBREVIATIONS: AUT = Abhidhiinottaratantra (Patalas as in MS NGMPP, Reel E 38 [41] (L6.m6./ak§a7Jam) 19.2-llc=lacunose NS
695/3; Tenjur numeration in brackets]; LST = LaghusaT{Lvaratantra; MMT = Mahiimu- 6.3-13c
dratilakatantra (from the Hevajra corpus; uncatalogued codex unicus, Staatsbibliothek
Pat ala 20 ( Sarvayoginyai}- No precise 13! 13!
zu Berlin, Hs. or. 8711); V:Q = Vajra¢akatantra; SarppT = SaTfLputodbhavatantra; SUT mudriisaT{Lketavidhii}) parallels,
= SaT{Lvarodayatantra; HT = Hevajratantra; YS = YoginfsaT{Lciiraprakara7Ja of the third =AUT 39 (42] (=VII 8)
$atka of the Jayadrathayamalatantra; PM = Picumata; SYM = Siddhayogesvar'imata;
Patala 21 ( Darsananga- except in the
NS = NisisaT{Lciiratantra; TS = Tantrasadbhiivatantra. mudr6./ak§a7Javidhii}) case of 10 10
•=AUT 40 (43] (=V:Q 7) isolated
verses, for
CST Patalas
LST Saiva sources No. of verses No. of parallel s!2
20-24. But a!2
in LST Pata!a verses
Patala 22 ( Savise§O.nga-
Patala 15 Ｈａｫｾ｡ｲＭ YS, ｓ｡ｭｹ｣ｲ･ｾｴ＠ 8 512
mudrii/ak§a7Javidhii}-)
cchomavidhil)) vidhanapajalaiJ. (verse (verse =AUT 41(44) (=MMT 14)
=AUT'35 (38].2-8; 116-119 and and
Sarpp T 7, 1.13-16 prose) prose) Patala 23 ( f?iikinivicihna- closely
mud ｲ｡ｬｫｾ＠ a7Ja vidhii}) related 7 7
Patala 16 (Saptayogin'i- YS, ｳ｡ｭｹ｣ｲ･ｾｴＬ＠ 17! 17! =AUT 42:1 [45:1] passages are
ｬ｡ｫｾＷｊｰｲｩＶＮｶ､ｨｽＩ＠ vidhiinapa!alai}, found
= AUT 36 [39:1]; 120-138b in TS and
12-18b = SUT 9.7-11 PM

Pat ala 24 ( f?akinlchomma- 16 16

lak§a'(Lavidhil}) (prose) (prose)
52 [2001:]
Since writing this I have encountered other extensive passages that were = AUT 42:2 [45:2]
brought into the text from the same Saiva sources,· covering such topics as initiation
( abhi§ekai}) and the making of offerings into fire to attain supernatural effects (siddhil)),
Pat ala 26 ( ｓｩｾｹ｡ｰｲｫﾧＭ PM Adhikara 85 (samayai}), 9!2 8
amounting with those mentioned in Table 1 to about half the whole. There may well be
saT{LVaravidhii}), 6-12, 4-11 (6-11 = YS,
other sources that I have not yet identified or that may never be identified, since we are
14 (one line), 15 (15 vv. ｓ｡ｭｹ｣ｲ･ｾｴ＠
far from having manuscripts of all the Vidyapitha texts wh?se existence is attested in
those that have survived. As for the rest, much of it follows Saiva models. For example, in patala) = AUT 43:1 vidhii}), 3-10
the substantial sections in which the Mantras are given sound by sound ( mantroddhiirai})
Patala 27 ( Caryavrata- PM Adhikara 85 (samayai}), 24 10
cannot have been lifted directly from a Saiva source, because however dependent the
pujiibalividhil}), 1, 6-14, 23ab 12, 14, 16cd,
text was on Saiva prototypes in other respects, a Buddhist Tantra had to have its own
(24vv. in Patala) 17cd-23b, 25abc,
Mantras. But the convention according to which they are given sound by sound in the
=AUT 43:2 26bcd, 31ab
code of their numerical position in the numbered letter classes ( varga!>), does have its
exact Saiva parallels in this literature; and was no doubt adopted from that source. The
Mantras themselves are in the style of those of the Saiva Vidyapitha.
44 Sources and time History through Textual Criticism 45

Patala 28 (Adhyiitmahoma- PM Adhikara 85 16 12

The commentators had to do what they could with this. Jayabhadra's
vaT!'aikavidhi(>.), 1 (2 pada (samayii(lr), 31cd, 32 cdef, Paiijika has:
verse), 3, 5-7b, 8-11, 13c-16 33-35b, 35c-40b, 40c-43b,
smaranena samayukta'T{! paiicamrtatattvalak{3a"!1ena samanvi-
Pat ala 29 ( Dutflak§a'(!asakty- PM Adhikara 85 10 4! tam iti. dutyadipadanvitam iti anavaratayogabhyasena dutf-
2
avasthiividhi(>.), 1ef-3c, (samayii(>.), 43c-45c,
3d-4a, 4bcd (10 vv. in patala) ＴＶｾ｣Ｌ＠ 47bcd, sarp,yukta!;.. adisabdo 'tra rak{3a"!1fyabhak{3a"!1'iyasa'T{!varasthito
mantrf. tatrastha varal;. pradhanas caturvirp,sati( r) ¢/ikinya}_t
Pat ala 41 ( Caturvi'f[!satiyoginz- TS Adhikara 16 ( Yogini"- 14? 9!
2 bhuktimuktiphalaprada bhavanti,
samuhanyiisavidhi(>.) lak§a'(!am), 60-68
SIGLA: A= NAK 3-365, f.20r1-2; B = NAK 5-212, ff.47v4-48rl. samii-
Patala 49 (Saptajanmasiidhya- TS Adhikara 7, [5], (HT 2 yv.ktam A : samiiyuktii(>. B • tatvalak§a'(!ena B : tvalak§a'(!eiu:l A e sa man-
riipaparivartanavidhi(>.), 1 . 1 xi llcd= [6ab]), vitam A : samanvitiim B • sa'f[!yuktalj, B : sa'f[!yukta'f[! A
3c-4b, 4cd, 5ab 4, 7ab, Sed

Equipped with smara"!1am means provided with that which is

Here more than anywhere in the Tantric literature only religious faith the reality of the five Nectars. As for the compound dutyadipada-
could shield one from the realization that original texts need not to be nvitam it means conjoined with a female consort ( dutt) through
meaningful throughout, since they may incorporate lacunae of meaning ceaseless practice. The word -adi- in the compound refers to
from their source-texts or create them through ignorance or carelessness the Mantra-master (mantr'i) committed to the sa'T{!varii}_t, both
in the process of redaction. I will provide a single example. As Table I those that have to be observed [i.e. the pledges of the initiate]
shows, Patala 85 of the Picumata is the source of Patalas 26 to 29 of the and those that have to be consumed [the 'impure' substances
Laghusa'T{!vara. The subject is the rules that bind those who undertake of the cult]. Taking up their abode in that [which brings these
the postiniatory observance (samayacara}_t). At the end of the passage three together: the Nectars, the consort, and the 'observant
incorporated, in Picumata 85.41-43, the reader is exhorted to maintain Mantra-master] the best, i.e. the principal twenty-four l)akinis
his practice of the smara"!1am. This is a technical term for the basic seed- become bestowers of the rewards of enjoyment and liberation.
syllable of this Tantra, and the passage begins by giving the form 'of the .,
root-Mantra (mulamantral.t) of which it is the first element (NAK 3-370 of Bhavabhata's Cakrasa'T{!varapaiijika has:
A.D. 1052, f. 327v):
smara"!1a'T{! cinta, anena samayukta'T{! samanuyoga}_t. kasya ci-
smara"!1ena samiiyukta'T{! CAt;!QE-tyiidipadiinvitam ntety aha, dutyiidipadanvitam iti, iidisabdiit ｳ｡ｭｹｲｫｾＢＡＱＮ＠
42 KAP ALI NY- antasa'T{!y"likta'T{! SV AHA- ｰ｡､ｶｩ｢ｨｵｾｴｭ＠ pada'T[! mulam. tenanvita'T[! samara"!1( en) a samayuktam. etena-
hrdaya'T{! sarvasaktfna'T{! bhuktimuktiphalaprada}_t itad uktam. samayapare"!1a dut'ibhavana karyeti, dutya eva ki-
yatya ity aha, caturvi'T{!sa para cfakinyo bhuktimuktiphalaprada
The heart of all the Powers, that bestows the rewards of para- iti, praca"!1¢adayas caturvi'T{!Sati( r) cfakinya ity ｵｰ｡ｬｫｾＢＡＱｴｶ＠
dise and liberation, equipped with the smara"!1am, with the dvadasapfha jiieya}_t. kimartha( Tf!) ta bhiivya ity aha, bhukti}_t
word CAt;IQE first, then KAPALINI following it, adorned by SVAHA svargadi}_t. ( mukti}_t) kle5abandhanapagama}_t. te phale ｰｲ｡ｫｾ･ﾭ
"!1a dadatui tas tathii.
Manuscript: Institute for the Advanced Study of World Religions, film-
This gives the nine-syllabled Vidya of Ca1.1\fa Kapalini: HU¥ CAt;IJ?E KAPA- strip no. MBB 1971-33, f.l01v4-102rl.
LINI SVAHA. The redactor of the Laghusa'T{!Vara transformed this into the
versity, Baroda, Ace. No. 13290: Ｇｈ･ｲｵｫ｡ｶｩ､ｨｮｴｦ｛Ａｾ＠ does not contain this verse,
following unmetrical and ungrammatical puzzle in 29.3: since it has a lacuna beginning with 22.3 and extending to the end 'of Patala 29. The
Sanskrit is given here as it is preserVed in the Abhidhiinottara (NGMPP Reel E 695/3,
f. 182r6-v1 ), which has incorporated Pat ala 29 in its Pat ala 43. The latter's fidelity is
smara"!1ena samayukta'T{! dutyadipadanvitam proved by the lemmata in the extant LST commentaries, to be cited below, and by the
caturvi'T{!sa varacfakinyo bhuktimuktiphalaprada}_t 53
Tibetan translation (Peking Rgyud Kha 81a4-5): dran pa dan ni yan dag ldan I pho
na mo sogs go 'phan bcas 1 mkha, 'gro ma mchog ni su bi 1 spyod dan grol ba 'i 'bras
53 The only accessible manuscript of the Sanskrit text (Oriental Institute, M.S. Uni- rab ster.
 46 Sources and time His,tory through Textual Criticism 47

· The word smara'!l-am means awareness and smara'fl-ena samii- interpretations. One can only sympathize, then, with the plight of those
yuktar[! means possession (of it]. Awareness of what? He Tibetans of the tenth and eleventh centuries who were given the task of
explains this with the word dutyiidipadiinvitam [which means translating the Sarp.varatantras into their native language in the conviction
'conjoined with the Diitis etc. and the padam']. The word -iidi- that this task was fully possible. It is not; for the texts are littered with
['etc'.] here indicates [that there should also be] adherence to passages whose meaning is thoroughly obscure, either because they inherit
the samayii/J (pledges/sacramental substances). The padam is the defects of the exemplars in which they drew upon their sources or
the root[-Mantra]. The state of being provided with awareness because their redactors lacked the determination or ability to produce un-
(of these] includes the Root Mantra. What he means by this ambiguous results as they worked to assimilate and disguise their sources.
is that one should perform one's meditation on the Diiti only
if one remains committed to the samayiilJ. He explains how
many these Diitis are by speaking of the twenty-four supreme
:Qakinis who bestow the reward of enjoyment and liberation.
This should be understood as indicating not just these twenty-
four but others of the same kind. So one should understand
that this includes the twelve too. He explains why they should
be the object of one's meditation. Enjoyment means"heaven
and the like. Liberation means the cessation of being bound
by the Defilements. They fully Ｈｰｲ｡ｫｾ･ＧＡｬＭ＠ ＨＫＭｰｲ｡ｾ｝Ｉ＠ l;>estow
these two benefits.

We see these early commentators attempting to make ｾ･ｮｳ＠ of .the in-

coherent by allowing the texts an impossible freedom from the rules of
grammar and syntax. 54 As this example shows, they were bound there-
fore in the free exercise of their imaginations to diverge widely in their
54 Faced with the chaotic Sanskrit produced by the redactors of the Yoginitantras the
commentators were content, for example, to interpret declensional endings in whatever
case or number they, felt would yield appropriate sense. Thus in his commentary on
Laghu5arr>vara 1.10 Bhavabhata reads sidhyate japa mantre'(ta as meaning sidhyati ja-
pena mantra/). 'the Mantra succeeds through repeated recitation', saying that the end-
ing of the instrumental singular hai been elided in the case of the ｳｴ･ｭｾｦｯｲ＠ japa same Tantra Jayabhadra explains that since case-endings are employed ad libitum in
and that mantra/)., a nominative singular, is being used as the instrumental singular the Tantra in order to cause general confusion they should be interpreted as the sense
(cod. cit., f. 24v3-4): japeti trtfyalope. sa ca manaso vyaparah. tena ... mantrah si- requires (If. 38v6-39r1): tantre sarvavyamoharthaT[! ｹ｡ｴｨ･ｾｲ＾＠ vibhaktinirde§ai). krtai)..
dhyati. mantre'(teti prathamtirthe t'[tlya; on 17.1ab ( tato durlabha yoginlnaT[! tu ¢akinl- tadarthavasad unneyai).; and in his ｃ｡ｴｵｾｰｦｨｬｯｫｩＬ＠ NAK 3-360, f.1v3, on the equally
nan tathaiva ca) he reads genitives plural as nominatives plural (f. 72v3-4): yoginlna'f!l chaotic Sanskrit of the Buddhist Catun>thatantra KalyaQavarman says evam ｢ｨ｡ｾｩﾭ
tu ¢akinfnam ityadi. prathamarthe ｾ｡ｴｍ［＠ on 3.16ab ( etair vicarej jagat sarvaT[! ¢aki- tety iirabhya yavad abhyanandann itivacanii( n) ta( '?') vibhaktilingavacanasamasadayai).
nyaii). saha siidhakai). 'The Sadhaka may wander .the whole world. with these J?akinls'), gurulaghuyaticchandadayas carthopadesavasad yathiiyoga'f!l yojanfyai). '[In this whole
in which a feminine noun has a masculine or neuter ending ( etaii). . :. ¢q.kinyail).), he Tantra] from the words· eva'f!l ｢ｨ｡ｾｩｴ＠ at the beginning to abhyanandan at the end, the
observes (f. 38v1-2): etair ljiikinyair ityiidika'f!l va7"!'agamaviniiSaviparyayavikarair ni- suffixes of case, gender and number, [the semantic relations between members of] com-
ruktavihitaii). sadhyam ｩｲｾ｡ｳｹ＠ tatpriidhiinyena sadhyatviit 'Expressions such as etair pounds and [other matters of grammar and syntax], as well as such features as length
ljiikinyaii). should be validated by means of the addition, elision, transposition and and of vowels, caesurae and metre, should be interpreted as each case demands on the basis
transformation of sounds that are prescribed in the semantic analysis (taught in the that they are intended to convey an [apposite] sense.' This emphasis on meaning over
Nirukta), because that is the principal method in the exegesis of the language of scrip- grammaticality or Sanskritic elegance of expression is in accordance with the ancient
. :! ture'. These four with dhatvarthatisayayogai). 'stretching the meaning of a verbal root' principle invoked in all the Buddhist literatures that the Buddhist should concern him-
are the five methods of semantic analysis (paiicavidha'?' niruktam) as stated in a verse self with the meaning alone ( arthapratisara'.'ata ) and not be overly concerned with
cited by Durga ad Nirukta 1.1.1,.ed. H. M. Bhadkamkar (1918), p. 32; see Eivind Kahrs, niceties of language and composition. See in the Tantric literature, e.g., Aryadeva,
'Exploring the Saddanfti', Journal of the Pali Text Society 17 (1992), pp. 22-23, n. Caryamelapakapradfpa (alias Sutramelapaka), Calcutta, ASB, MS. 4837 G, f. 2r): ya-
3, whom I thank for bringing this reference to my attention. On 1.5 Bhavabhata tells thokta'f!l bhagavatii 'arthapratisara'(tena bhavitavyam. sabdas tu yathii tathii, iti, 'As
us that beca.use the text is scripture its metrics, grammar and the like are not open the Lord himself has declared, "You should pay attention to the meaning. The form of
to criticism (f. 18r3-4): neha chandai).sabdadi vivecanfya'f!l ｡ｲｾｴｶｩＮ＠ On 8.4 of the the words is irrelevant."'
PUBLICATIONS DU DEPARTEMENT D'INDOLOGIE- 91

LES SOURCES ET LE TEMPS

SOURCES AND TIME

A colloquium

Pondicherry
11-13 January 1997

edited by
Franc;:ois Grimal

INSTITUT FRANc;AIS DE PONDICHERY

ECOLE FRANc;AISE D'EXTREME-ORIENT
Comite de lecture

Colette Caillat,
Membre de l'Institut de France
Fran<;ois Gros,
Directeur d'etudes a l'Ecole pratique des Hautes Etudes, Paris
Michael Hahn,
Professeur a l'Universite de Marburg
Alexis Sanderson, . .
Spalding Professor of Eastern Religions and Ethics; All Souls' College, Oxford
Raffaele Torella, PREFACE
Professeur a l'Universite de Rome
It takes a great deal of time to arrive at sources, and the retrieving
process takes even longer. It was not expected that any claim would be
made to final answers being received to the questions put to the scholars
participating in the colloquium on Sources and time which took place in
Pondicherry under the joint auspices of the French Institute and the local
branch of Ecole Franc;aise d'Extreme-Orient.
The main preoccupation of the textualists, most of them strangers to
India, was obviously, to display and to examine the manner in which texts
stand the test of time, how they. survive, are preserved and transmitted,
and how philologists struggle to restore the best possible version, going on
to show how, and with which of the available' tools such as comm'entaries,
ancient and modern, oral and written, indigenous or otherwise, that version
itself is better understood and, ultimately, to how it is translated bearing
in mind that the epoch and metalanguage of the translators is but the
latest damaging aspect of time erosion, of course never to be final.
The scholars were however asked which historical and epistemological
language would permit both the definition and preservation of that p11-rt of
cultural patrimony which lies within the traditional transmission of know-
ledge, operating from external approach, as well as permitting exposure of
the inherent metalanguage hidden behind each of the traditional siistra as
within each of our day-to-day self-interrogations.
A special feature of the structure of the French research institutes estab-
lished in Pondicherry is their very close cooperation with Indian scholars
© Institut fran<;ais de Pondichery, 2001 (ISSN 0073-8352) having their own methods and hierarchy. The colloquium was therefore
©Ecole fran<;aise d'Extreme-Orient, 2001 a unique opportunity to evaluate such interactions and to develop the in-
terface between western principles and instruments of textual criticism, as
Ouvrage compose en "Computer Modern" et Velthuis' Devanagari par tested over a long period in classical western philology, and the more in-
Mme T.V. Kamalambal, a l'aide des programmes '!EX, D.'IEX et EDMAC digenous, and very effective, readings of texts by those who have received
(macros pour la realisation d'editions critiques, crees par John Lavagnino et formal training in traditional Indian learning, and who often consider a
Dominik Wujastyk) text as a support to an oral transmission (now fragile and, to an inquis-
itive western mind, sometimes hardly warranted) and tend to see variant
Couverture et impression : Sri Aurobindo Ashram Press, Pondicherry