Summary of ‘Tantric Śaivism’,

 lectures delivered at the École Pratique des Hautes Études,

Alexis Sanderson

Since the beginning of research in the field of Tantric Śaivism scholars have
directed their attention to one of two bodies of Tantric Śaiva literature concentrated
at opposite ends of the Indian subcontinent. They have worked either on the [Śaiva]
Siddhānta using largely south Indian sources and called this ‘Southern Śaivism’ (le
śivaïsme du sud), or on the works of Abhinavagupta and other Kashmirians of his
tradition and called this ‘Kashmir Śaivism’. This division of labour has been reinforced
by the fact that two bodies of evidence invited very different approaches. ‘Kashmir
Śaivism’ attracted study of its nondualistic doctrines and intuitionist soteriology,
while the literature of ‘southern Śaivism’ was studied primarily for its detailed pre-
scriptions of Śaiva ritual. Nor were researches likely to find much in the living tradition
that would invite them to question this dichotomy. South Indian informants of the
Siddhānta have been Ādiśaiva temple-priests concerned almost exclusively with ritual
and only with the ritual of their own tradition, while the ‘Kashmir Śaivism’ now
accessible in Kashmir and reflected in many modern publication is strongly gnostic
or yogic and neither understands nor wishes to understand the texts on ritual which
were once an important part of its inherited literature, far less the relation between
their ritual and that of the Siddhānta.
This series of lectures parts with this dichotomy in research. It has shown that
‘Southern Śaivism’ and ‘Kashmir Śaivism’ are intimately related in the primary scrip-
tural sources (Tantras) which they claim to interpret and also in the earliest exegesis
of these texts, which was Kashmirian for both schools. The relation is rightly char-
acterized in sources of both trends as that between the common or fundamental
(sāmānyaśāstram) and the specialized (viśesaśāstram).
. The common scriptures at the
root of ‘Southern Śaivism’ are the Siddhāntas or Siddhāntatantras. The specialized
scriptures are those of the Vāma, Daks. ina . and Yāmala cults including or extending
into the Tantras of the Trika and Krama cults which are at the root of ‘Kashmir
Śaivism’. The common Tantras recognize the systems or ritual and observance taught
in the specialized Tantras; but they see themselves as the core of Śiva’s revelation and
the specialized Tantras as secondary teaching which principally serve the desires of
siddhi-seeking sādhakas rather than ordinary (liberation-seeking) initiates (putrakah, .
dīksita
. h). and the officiants (ācāryah,. guruh). appointed from their number. The special-
ized traditions perceived the same relation differently. For them the Siddhāntatantas
were certainly the basic revelation, in the sense that they were the most universally


applicable, their instructions applying everywhere in the canon unless specifically
modified or countermanded; but they were also seen as the more exoteric, as the broad
base beneath a non-Saiddhāntika superstructure consisting of a series of progressively
more esoteric levels.
So the analysis of Śaivism developed in these lectures has treated the Siddhān-
tatantras and the non-Saiddhāntika Tantras as complementary parts of a single greater
system, applying the principle recognised by both aspects of Śaivism that the Siddhān-
tatantras are the basic prescription which the specialised traditions modify or inflect
in various particulars. Consequently we first considered Tantric Śaivism using the
Siddhāntatantras as our primary evidence, and then proceeded to an analysis of the
specialized systems, defining them through their divergences from this norm.
This model of dependence corresponds to the order of historical development
in general; but we saw that there are some aspects of the specialized systems which
probably predate the basic system found in the Siddhāntatantras, namely the doctrine
of the Vāmatantras and the cult of Yoginīs found in the texts of the Yāmala, Trika and
Krama divisions of the canon.
Because we set out to examine Śaivism from its earliest sources, the corpus of Sid-
dhāntatantras used does not include all the texts which Saiddhāntikas have accepted
as canonical. Many works which claim to belong to the Saiddhāntika canon are found
only in south Indian manuscripts; and some, such as Kāmika, the Ajita, the Kārana, .
the Cintya, the Dīpta, and the Suprabheda, are definitely late products of that region.
We have therefore recognized only those works which survive with or in Kashmirian
commentaries, are preserved in early Nepalese manuscripts, or through surviving only
in south Indian manuscripts can be supported by early citations or comparison with
other early materials. The principal among the Siddhāntatantras admitted by these cri-
teria were the Kālottara in various versions, including the Sa
. tsahasra-
. or Brhatkālottara,
.
the Kirana,
. the Mataṅgapārameśvara, the M rgendra,
. the Niśvāsamukhasa mhitā,
. the
Niśvāsasamhitā,
. and the Niśvāsakārikā (including the Dīk sottara),
. the Pauskarapāra-
.
meśvara, the Rauravasūtrasamgraha
. and the Svāyambhuva. These are only a small part
of the titles listed in accounts of the Saiddhāntika canon. However it seems that they
were the core and greater part of the canon known in the tenth century and after,
because there is a close correlation between the range of Siddhāntatantras attested in
Kashmir and that which survived independently in Nepalese manuscripts among the
Newars of the Kathmandu valley.
As introduction to these specific Tantric traditions we considered a number of
matters bearing on the relations between Tantric Śaivism and the non-Tantric tradi-
tions. First we looked at the question of the degree of historical continuity between
Tantric Śaivism and the earlier, non-Tantric Śaivism of the Pāśupatas. We found that
Tantric sources divide that tradition into two distinct groups, the Pāśupatas/Pañcārthas


and the Lākulas. The first are evidently the Pāśupatas about whom something is already
known, the followers of the system of the Pāñcārtha (the Pāśupatasūtra) expounded in
the Pañcārthabhāsya. etc. We saw that though the Vaidikas condemned them as outside
the Vedas (vedabāhyah, . pāsa . n. da
. h),
. the Pañcārtha-Pāśupatas themselves considered
their tradition to be the highest and most esoteric path within Vedic knowledge,
accessible only to a Rudra-inspired élite among regenerate men of the highest caste.
It has already been noted that their mantras, the five brahmamantras, are Vedic.
Our analysis observed other Vedic features. We argued, for example, for the Vedic
origin of their practice of provoking abuse from public in the belief that this will
cause demerit to pass from them to their abusers and merit to be pass from their
abusers to themselves. We connected this with the Vedic rule doctrine that one must
avoid all criticism of one who is following the observance (vratam) consequent on
dīksā
. for a Soma sacrifice, because regardless of whether it is just or injust it will
cause just such a transfer (Kāthakasa
. mhitā
. .; Āpastambaśrautasūtra .. and
the Bhasya
. of Rudradatta). The Pāśupata too is observing a vratam (pāśupatavratam,
atyāśramavratam), though in this case it is a lifelong condition; and he too enters on
his observance through a dīksā. .
Considering the importance of dīksā . or ‘initiation’ in later, Tantric form of
Śaivism, we examined the evidence for the Pāśupata ritual and concluded that it
corresponds closely enough to the first phase of the Tantric to justify the hypothesis
that the latter developed out of the Pāśupata though extension and elaboration. We
saw, however, no evidence of the belief, so crucial in the Tantric texts, that dīksā . is
not only a ritual of qualification (samskāra . h)
. for certain observances but also brings
about liberation at death by destroying the bounds of the soul.
The nature of the practice and doctrine of the Lākulas, the second division of
the Pāśupatas, was clarified with the help of the Niśvāsamukha, an unpublished
Siddhāntatantra surviving in a Nepalese manuscript which may be assigned on palaeo-
graphic grounds to the tenth century. This text is unique in the surviving literature in
that it gives an account of their practice and the hierarchy of cosmic levels which
is the basis of their soteriology. We saw that these Pāśupatas were Kapālavratins
or Mahāpāśupatas, ash-smeared ascetics like the Pañcārtha-Pāśupatas, but set apart
from them by the fact that they included among the insignia of their observance
a begging-bowl fashioned from a human cranium (kapālam) and a staff topped
with a human skull (khatvāṅga. h).
. We saw also that their hierarchy of worlds is
embedded in the Tantric Śaiva hierarchy that appears in the Niśvāsasamhitā . and the
Dīksottara,
. and also, beyond the Siddhāntatantras, in the Svacchanda (Dak s
. . and
ina)
the Tantrasadbhāva (Trika). The embedded versions established the identity of these
ascetics with the followers of the eight Pramānaśāstras. (Pañcārtha, Guhya or Śivaguhya,
Aṅkuśa or Rudrāṅkuśa, Hrdaya,
. Vyūha, Lak sa
. . na, Ākar sa
. and Ādarśa). These scriptures


do not survive. But a passage from the first was found in Ks. emarāja’s commentary
on Svacchanda .–, and it confirmed the identification with the Kapālavratin
Pāśupatas of the Niśvāsamukha. We also established the identity of these ascetics with
the Mahāvratas/Mahāvratins whose views of the universe and the nature of liberation
are mentioned by the Kashmirian Saiddhāntika Bhat.t.a Rāmakan. t.ha. These and other
references showed that it is this same group that is referred to as the Kālamukhas or
Kālavaktras in Kashmirian sources. An ascetic order named thus or Kā.lāmukha (a
variant to be attributed to the influence of Kannada pronunciation) is known to have
flourished in Karnā . t.aka, Mysore and Āndhra from epigraphs dating from the ninth
to the thirteenth centuries. We saw that these epigraphs confirm the conclusion that
this is the same tradition; and we were therefore able to vindicate Yāmunācārya’s brief
account of their observances which Lorenzen had rejected in his study The Kāpālikas
and Kālāmukhas: Two Lost Śaivite Sects () on the grounds that the epigraphs,
which are his sole evidence, show the Kālamukhas to be Pāśupatas, while the use of
the skull-bowl which that account attributes to them is associated with the Kāpālikas.
It is true that the Pañcārtha-Pāśupatas’ observance lacks such mortuary elements, but,
as we can now see, this was not the case with the other division of the tradition.
Continuities between the Lākula/Kālamukha/Pramāna . system and that of the
Śaiva Tantras were observed. The Tantric’s universe (bhuvanādhvā) is an extended ver-
sion of the Lākula’s and inherits its dualistic division between impure levels (aśuddhā-
dhvā: the cosmos of the bound) and pure levels (śuddhādhvā: the cosmos of the
liberated). We also saw that the Niśvāsamukha’s Mahāpāśupatas share the Tantric Śaiva
view that dīksā . is a means of liberation. It was suggested with the support of a verse
quoted from an unknown work by the Kashmirian Somānanda (see Abhinavagupta,
Parātrimśikāvivara
. na
. pp. , l. –, l. ) that the mysterious ‘descent of the word
atha’ (athaśabdanipātah) . which is said by this text to accomplish the dīksā . is the
descent of the power of Śiva (śivaśaktinapātah) . of which the texts speak in the context
of Tantric dīksā..
Continuities between Pāśupatism (which the Niśvāsamukhasamhitā, . referring to
both systems, calls the Outer Way [Atimārga]) and Tantric Śaivism (which that text
calls the Way of Mantras [Mantramārga]) were also seen in the area of observance
(vratacaryā). Both kinds of Paśupata observance were shown to be part of the Tantric
tradition of asceticism. Those who adopted the basic ascetic observance prescribed
by the Siddhāntatantras were barely distinguishable in appearance from Pañcārtha-
Pāśupatas in their first stage, that is to say, the stage in which they assumed sectarian
insignia; and in the rudravratam of the Mataṅgapārameśvara we recognized a more
comprehensive survival of Pāśupata practice which included the courting of contempt.
The observance with the skull-bowl and skull-staff (khatvāṅga . h)
. followed by the
Lākulas was found as a special ascetic observance for the attainment of siddhis in


the Niśvāsasamhitā,
. and very commonly outside the Saiddhāntika part of the canon.
We then considered discontinuities between Pāśupatism and Tantric Śaivism.
The only goal offered by Pāśupatism is liberation through the definitive cessation of
suffering (duhkhānta
. h):
. Tantric Śaivism offers liberation, lesser rewards (bhuktih), . or
both. All Pāśupatas are, at least in theory, permanent ascetics: Tantric Śaivas may be
permanent ascetics (nais.thikavratī),
. temporary ascetics (bhautikavratī) or non-ascetics
(grhastha
. h).
. Only regenerate brahmin men may take up the Pāśupata observances: the
benefits of Tantric dīksā . are accessible to men of all four castes and to women. All
Pāśupatas are required to be actively Pāśupata: Tantric initiates are of two kinds, (i)
those in whom dīksā . initiates a lifelong obligation to perform Tantric rites (sabīja-
dīksitā
. h. and [śivadharmi]sādhakāh) . and (ii) those whose involvement in Tantric ritual
may be limited to passing through the ceremony of their initiation (nirbījadīksitā . h.
and [lokadharmi]sādhakāh). . Pāśupatism is personal religious activity: Tantric Śaivism
extends from this into the public domain.
Since Pāśupatism is a path to liberation for ascetics, while Tantric Śaivism adds
married householders and lesser rewards to its initiates and goals, it might be assumed
that the reward-seekers, termed sādhakas, were the householders. However, although
some texts mention that the sādhaka can be a married man, it seems that it was the
ascetic sādhaka that was the model. It appears, then, that a fundamental aspect of
the relation between the Pāśupata and Tantric systems was the contrast between two
kinds of ascetics: those seeking liberation and those seeking powers.
If Tantric reward-seekers were typically ascetics, were the liberation-seekers, among
whom all Śaiva gurus were numbered, typically householders? The answer is that they
may have been, but not because the Tantric tradition saw any natural connexion
between the householder’s state and a person’s qualification or desire for liberation.
On the contrary, it considered the ascetic liberation-seeker to be purer and superior
type. Tantric Śaivism, then, contains both the types of asceticism whose contrast
we saw as characterizing the difference between this system and Pāśupatism; and it
was suggested that the former is the result of a synthesis of two distinct traditions,
a liberationist tradition prolonging that of the Pāśupatas and a tradition of ascetic
power-seeking represented by the sādhaka.
This model has the strength that it fits the picture of the interaction of the various
classes of initiates that we get from the rules stating their rights and duties. We do not
see a community of initiates of both types fully integrated under the authority of a
guru, but rather a community of liberation-seekers under the control of their guru and
living in his home (gurukulavāsah) . with the addition of one or more sādhakas who,
though nominally under authority of the guru, are in fact independent practitioners
who live apart at a sacred site (ksetram)
. chosen for their mantra-practice or wander
from one site to another.


 The synthesis, however, entails a compromise. The Tantric tradition does not
allow sādhakas to obtain their empowerments from other sādhakas: they must be
initiated and consecrated by gurus, who are liberation-seekers by definition. But it
is conceded that these rituals can only be done by a married guru. Ascetic gurus
can initiate other liberation-seekers: married gurus alone can initiate both liberation-
seekers and reward-seekers. It might well be said that the married guru ought to be
equally unqualified to empower the sādhaka, since he is, after all, a liberation seeker.
The objection is sound and good proof of the awkwardness of the arrangement.
The liberationist tradition could justify its position in the Tantric synthesis only
by applying a principle that should have no place on the supramundane level at
which Tantric activity is supposed to operate, namely the Vedic or exoteric (laukika-)
opposition between the liberation-seeking ascetic and the man in the world seeking
limited and transient benefits from his dharma. Householder gurus are liberation-
seekers, because they are gurus; but they can initiate reward-seekers because they are
householders, who pursue transient rewards rather than salvation. So the price of the
sādhaka’s incorporation and subordination is a problematic domestication.
This domestication is also apparent in Tantric Śaivism’s extensions beyond per-
sonal religious activity. While Pāśupatism is limited to religious activity which benefits
the performer alone, going beyond that only in the dīksā. performed by gurus to qualify
others for such activity, Tantric Śaivism extends the range of rituals for others beyond
those which enable the personal cult to include rituals in which the beneficiary is
deceased, an uninitiated yajamānah. or even a group. Thus Tantric gurus perform
Śaiva versions of the Smārta funerary and postfunerary rites (antyes.ti . h,
. pretakriyā,
śrāddham) for the benefit of a dead individual, ancestor or ancestors at the request of
their relatives. They may be engaged by a uninitiated yajamānah. to consecrate non-
personal idols and temples; and they may be appointed to perform the regular and
incidental rituals in such temples, rituals whose beneficiary is the whole community
within the temple’s range. Their activities, then, extend from the performance of
rituals such as initiation which are consistent with their status as gurus and are shared
by the Pāśupata tradition, to others such as processing the dead and serving in temples
which take them into the less prestigious category of priests.
Another aspect of domestication was brought out in an analysis of the evidence
for the two categories of inactive initiates. It was shown that while the Kashmirian
authors of both schools hold that the nirbījadīksita-
. devotee is not qualified to perform
Tantric rites, the Brhatkālottara,
. Somaśambhu and Aghoraśivācārya hold that initiates
of this kind must take on as much of that discipline as they can. Such initiates are not
excluded from ritual: they are exonerated from the full discipline that binds others.
This meant that Tantric observance was no longer restricted to the presumably small
numbers of individuals who, like Śrautins in the context of the Vedic system, were in


a position to take on an arduous and inflexible routine of rituals. Most significantly,
those whom this view admitted to Tantric worship included women. They might not
become gurus or sādhakas and they might not become ascetics (liñgī); but they were
permitted do Tantric worship, even if only the annual pūjā on Śivarātri believed to
make up for any omissions during the year.
We also saw that the Brhatkālottara
. made Tantric worship even more accessible
to women by admitting them as full initiates, though still not as gurus, to an annexed
cult of Gaurī, Śiva’s subordinate consort.
That Tantric Śaivism was compromised by these various extensions into the
religion of the home and the temple (the laukiko dharmah. of the Niśvāsamukha) was
shown in the case of the Śrāddhas by the fact that such rites contradict the tradition’s
doctrine of salvation. It is equally clear in the case of women’s initiation into the cult
of Gaurī; for we are told by the Brhatkālottara
. that if at any time there should be
a conflict between a woman’s Tantric deity and her husband’s command, the latter
must prevail.
We also saw that Tantric Śaivism was aware that its incorporation of elements
of the exoteric, social religion, and, more generally, its embeddedness in the sys-
tem of castes and life-stages (varnāśramadharma
. h),
. were problematic. After examin-
ing the various orthodox (Vedic) views of the Tantras seen in Kumārila, Aparārka,
Jayantabhat.t.a, the Sūtasamhitā,
. and other works, we considered the way that the
Tantric initiate is required to perceive the Vedic religion or more exactly the relation
between the Vedic and the Tantric in his religious activity. We studied passages in the
Mataṅgapārameśvara, the Sarvajñānottara, and the Jayadrathayāmala. These taught (i)
that Vedic and Tantric prescriptions are so ordered in the totality of revelation that the
Tantric may incorporate the Vedic but never be overriden by it, (ii) that for the same
reason the Vedic elements that coexist with the Tantric are not to be seen as aṅgas
of the Tantric, that is to say, as contributing in any degree to the efficacity of the
Tantric; and (iii) that the true purpose of the incorporation of the Vedic is to preserve
worldly conventions (lokasamv . rti
. h),
. either out of compassion for the unenlightened
or to protect the reputation of the gurus and thereby to protect the faith.
Other subjects in these lectures: (i) an outline of the non-Saiddhāntika Tantric
systems from the Vāma to the Krama with an analysis of the principles behind the per-
ception of these systems as increasingly esoteric: this entailed further study of women’s
role and participation in Tantrism; (ii) a consideration of the question of whether there
was a Kāpālika tradition independent of the Kāpālikism of the more radical of these
systems; (iii) a demonstration that the hierarchical structure of these Śaiva Tantric
systems was mirrored in the development of the Buddhist Tantric canon, and that
in the case of the Śamvaratantras
. of the Yoganiruttara (/Yoginītantra) division, the
most ‘kāpālika-’ stratum of that canon, this dependence extended to incorporations,


often imperfectly disguised, of passages, indeed chapters, of non-Saiddhāntika Śaiva
works surviving in early Nepalese manuscripts; and (iv) a refutation of the doctrine
of Abhinavagupta and his school that dualism is restricted to the Siddhāntatantras.
It was shown that the nondualism which these nondualists claimed for the whole
non-Saiddhāntika canon is found only at the most esoteric periphery of that corpus,
in some texts of the Trika and Krama.