If, twenty years ago, you had read most of the literature published before the 1990s about the Śaivasiddhānta, you would probably have received the impression that this was primarily a South Indian movement, whose scriptures, called āgamas, were divided into four sections, or pādas, devoted to ritual (kriyā), doctrine (jñāna), yoga and pious conduct (caryā). The first two of these four sections, the kriyā-pāda and the jñāna-pāda, you would have learnt, were the most important, the kriyāpāda being devoted to describing the rituals practised in the Śaiva temples of the Tamil-speaking area, and the jñānapāda (or vidyāpāda) being devoted to teaching and defending a strictly dualist system that presents an ontological ladder of thirty-six tattvas, but that recognises three irreducible ontological categories: pati, paśu and pāśa. That is to say: the Lord (pati), bound souls (paśu), and the bonds that bind them (pāśa), namely Matter, karman and an innate impurity called mala or āṇava-mala.

Each one of these pieces of received wisdom has been challenged by the discoveries of the last two decades, so that we now know that none of the above propositions actually holds true for the earliest strata of the religion to which surviving primary literature can give us access. A great many of those discoveries are those of Alexis Sanderson and the students to whom for decades he devoted much of his time and energy.

Of course it is wide reading of a very broad corpus of published and unpublished sources that has gradually revealed to us quite a different picture of the early phases of the religion. But if one were to single out any one text for its importance in expanding our knowledge of the early history of the Mantramārga, it would probably be the Niśvāsatattvasaṃhitā.

Ten years ago, hardly any aspect of the text had been explored in print, but, thanks in part to the spotlight of the Franco-German ‘Early Tantra’ project, which between 2008 and 2011 focussed the minds of many people present at the Toronto symposium on the Niśvāsa and on its relation to other early tantric literature, parts of the work have been commented upon in an array of publications. The first major arcticle actually predates the ‘Early Tantra’ project, and is, of course, by Professor Sanderson himself: it is his study of the Lākulas (2006). Apart from the first volume of the collaborative edition and translation, covering the earliest three books of the Niśvāsa—the Mūlasūtra, Uttarasūtra and Nayasūtra—there are now substantial articles on, for instance, the evolution of the system of tattvas that can be traced as it gradually takes shape within the Niśvāsa-corpus (Goodall 2016), and on the lengthy grimoires of magical rites contained in the Guhyasūtra that are similar in style and content to those found in Buddhist kriyātantra works, and most strikingly similar to the those in the Mañjuśriyamūlakalpa (Goodall and Isaacson 2016).

Now that the earliest three books of the Niśvāsa are published at last, and now that the introductory book, the Niśvāsamukhatattvasaṃhitā, has been thoroughly examined in a doctoral thesis defended this year at the University of Leiden by Nirajan Kafle (2015*), what remains is the largest book of them all: the Guhyasūtra. The ninth-century manuscript of the corpus (NGMPP Reel No. A 41/14) comprises 114 folios, and all of the hitherto edited works together cover only the first 40 of those. The remaining 74 folios give us the text of the Guhyasūtra. Here is a very brief outline of the structure of its eighteen chapters:

A. Sādhana

Ch. 1 personality-types of sādhakas and types of liṅgas that may be used for siddhi.

Ch. 2 liṅgapratiṣṭhā.

Ch. 3 preparations for sādhana, prognosticatory rites, vratas, procedures for attaining certain siddhis.

B. Cosmography

Chs. 4–7 a lengthy cosmography (prakriyājñāna).

Ch. 8 a variant form of dīkṣā in the form of worship of a series of maṇḍalas peopled by deities of the different levels of the cosmos (prakriyāyāga).

C. Other Mantra-systems

Finally, the use, primarily for magical powers, of mantra-systems other than those given in the earlier three sūtras, namely

1. Chs. 9–11 The vyomavyāpin.

2. Chs. 12–14 The five brahmamantras.

   Ch. 15 Long forms of their aṅgamantras.

3. Chs. 16–18 A ten-syllable mantra called vidyā.

The Guhyasūtra is somewhat like a series of appendices to the earlier sūtras, containing more detailed accounts of some topics that have already been covered (cosmography), but also entirely new subjects (new mantras) or treatments of subjects that have hitherto only been alluded to, notably the acquisition of siddhis. As I have tried to indicate with the overarching titles (A, B, C) in the brief summary above, I think that it can be said that chapters 1 to 3 have a certain sort of unity because they cover the acquisition of magical powers in much greater detail than we see in earlier layers of the text: the first chapter gives information about sādhakas, then stresses the importance of the liṅga for attaining siddhis, after which, in chapter 2, the installation of liṅgas is covered, and then in the third chapter we return to the preparations for sādhana and finally the procedures to be followed. Chapters 4–7 then give us a very detailed account of the Śaiva cosmos, the higher reaches of which have been further expanded and embroidered upon since the composition of the earlier sūtras of the text.¹ This is undertaken because dīkṣā involves purging the soul of the fruits of karman that would need to be experienced—and thus expended—through every layer of the Śaiva universe. Using the same cosmography, chapter 8 describes an alternative dīkṣā involving the worship of maṇḍalas representing successive layers of the universe, and it then contains a number of add-on discussions that suggest, it seems to me, that the text once drew to a close at that point, as we shall see below. What follow, taking us up to the end of the Guhyasūtra, are three distinct textual layers each devoted to introducing an extra mantra-system, namely 1) that of the 81-word vyomavyāpin, 2) that of the brahmamantras, and 3) that of the ten-syllable vidyā. To each of these is attached a grimoire of magical recipes (kalpa).

Turning to the conclusion of chapter 8, I think that we can see from the summary given below that it reads like a series of codas. Verse 105 gives a clear statement of what we are supposed to have learnt from the preceding chapters, and it is followed immediately by remarks about the persons to whom these teachings may and may not be submitted, a typical closing device. Tagged on to this, from verse 116 onwards, is a treatment of religious suicide, again a theme suitable to the conclusion of a work of scripture. The final section, from verse 125, is introduced by Devī’s question about the status of rival religious traditions. In answer, Śiva explains that He and Devī, as consonants and vowels, are the source of all language, and that they are the source of all the universe in that they are to be identified with the various tattvas from which all else evolves.

Summary of the conclusion of Guhyasūtra 8:²

8.88–89 Devī asks how an initiate may foretell his own death.

8.90–98 Śiva recounts signs of death.

8.99–104 Activities that can be done under particular asterisms that grant release [from death?].

8.105 Summary of teachings from chapter 4 up to this point in chapter 8.

8.106–110 Those to whom one should and should not transmit this knowledge.

8.111–114b The 4 means of liberation: dīkṣā, jñāna, yoga, caryā.

8.114c–115 One should transmit this only to someone worthy.

8.116–117 Devī asks about religious suicide.

8.118–122 Śiva deprecates death in tīrthas for initiates; he teaches instead 5 varieties of a ‘razor’-mantra for suicide by japa.

8.123 Increasing length of life by yogic dhāraṇās.

8.124 The supreme Śaiva knowledge, without which one cannot be liberated, has been taught!

8.125–127 Devī asks about the fate of those who follow rival religions.

8.128–133 Śiva explains that He and Devī, as the consonants and vowels of the alphabet, are the source of all linguistic expression (vāṅmayam) and of all that has evolved (vikārāḥ).

8.134–136 The mūla-mantra is a panacea (mṛtasañjīvanī).

8.137–138 Śiva and Pārvatī are parents of everything in that they are respectively these tattvas: puruṣa and prakṛti; kāla and niyati; īśvara and māyā+vidyā; sadāśiva and kalā.

8.139–140 They are also respectively [supreme] Śiva and His Will (icchā).

8.141 Those who do not know the navātman, who are devoid of dīkṣā and jñāna, who do not know the mūla, do not attain the highest state.

We may remark in passing that a noteworthy feature of this final passage is that the text takes no clearly defined position on the debate between dualism, which was to become a defining characteristic of classical mainstream Saiddhāntika doctrine, and non-dualism: it seems as though this debate had not yet caught the interest and attention of Śaiva thinkers. Here is the passage in question:

ahaṃ pumāṃs tvaṃ prakṛti niyati puna[[r eva ca]]

niya---[2] kālarūpī maheśvaraḥ 8.137

tvam māyā ca tathā vidyā aham punas tatheśvaraḥ

sadāśiva ahan devi tvañ caturdhā kaleśvarī 8.138

īśitvāc ca vaśitvāc ca sarvajñatvāc ca nityaśaḥ

śāntatvān niṣkalatvāc ca samatvāc ca ahaṃ śivaḥ 8.139

mama icchā na hanyā tvaṃ tvaṃ hi śaktibalodayā

tvatsūtañ ca jagat sarvaṃ śivadā sadanugrahe 8.140

I am puruṣatattva and you are prakṛti and also niyati; … Maheśvara is Time; you are Māyā and Vidyā, while I am Īśvara-tattva. I, O goddess, am Sadāśiva [and] you are mistress of the 4 kalās. (137–138)

Because I rule, I control, I am omniscient, because I am permanently at rest, without division and in equilibrium, I am Śiva. (139)

You are my Will, not to be crossed, for you are the one from whom the power of the śaktis arises!

The whole universe has sprung from you; You bestow Śiva-nature, O you of true compassion! (140)

Having proposed an identification of the principal layers of redaction that are detectable in the Guhyasūtra, I should like now to skip back to the verses in the conclusion of chapter 8 that speak about those to whom the teachings may and may not be transmitted, since these verses bear both upon the themes that structure the work and upon the subject of caryā. For, slipped into the middle of that section, beginning in verse 8.111, is a short sequence of verses that make the claim that the teachings of the text comprise four independently salvific parts: dīkṣā, jñāna, yoga and caryā.

etad buddhvā na dātavyaṃ śivadevāmṛtam param 8.110

dīkṣājñānena yogena caryayā ca yathākramam

pratyekaśaḥ śivāvāptis tantre ’smin pārameśvare 8.111

dīkṣayā sukaraṃ mokṣaṃ yad gurus sādhayet sadā

jñānañ ca gurum āsādya labhyate tat[[prasādataḥ]] 8.112

---[f. 72r] te mX yo XX(?) ca gurupādataḥ³

ātmaśaktyā carec caryāṃ sarvasiddhipradāyikām 8.113

etac catuṣṭayam proktaṃ saṃsārabhayanāśanam

parasyaiva na deyan tu yadi ’cchet siddhim ātmanaḥ 8.114

Knowing this, one should not give [lightly] the supreme nectar of Lord Śiva. (110)

According to this scripture of the Lord, one may attain Śiva by each of the following [practised individually] (pratyekaśaḥ): initiation, knowledge, yoga and caryā in due order. (111)

By initiation one attains liberation easily, since it is the guru who invariably accomplishes it.⁴ And knowledge is obtained, once one finds a guru, through his grace. (112)

… yo[ga] … from the feet of the guru; One must practise caryā, which bestows all supernatural powers, using one’s own strength (ātmaśaktyā). (113)

This tetrad has been taught to destroy the dangers of saṃsāra. It should not [lightly] be given to others if one desires supernatural power for oneself. (114)

An innocent might here at first suppose that we find here what may be the earliest allusion to the notion that each Śaiva scripture should be arranged in four text-units called pādas, for it is not difficult to see that the kriyāpāda might easily be referred to by the most significant ritual of all, namely dīkṣā. Now Brunner (1992) and others⁵ have shown that most early scriptures are not in fact divided into four such text-units, and the Niśvāsa certainly is not. Nonetheless, one might reasonably suppose that the four topics to which some later scriptures devote four text-sections called pādas are referred to here. But are they? Plainly the first three, dīkṣā, jñāna and yoga, may be found treated at length in the Niśvāsa; but is there anything that we might recognise as caryā? This is a word we are rather used to seeing translated as “conduct” or “comportment,” as for instance in the title of Brunner’s 1985 translation of the kriyāpāda and caryāpāda of the Mṛgendratantra: Mṛgendrāgama[.] Section des rites et section du comportement. When she characterises the content of the caryāpāda there, she observes (p. xxxvii):

La presque totalité de l’ exposé (śl 1–105) est consacrée à un sujet unique: le comportement normal des différents groupes d’ initiés.⁶

This is a topic that we really do not find addressed in the Niśvāsatattvasaṃhitā, and it is therefore clear that caryā probably does not mean “comportement normal” in this text. When speaking in Toronto, I was unaware that Christian Wedemeyer, faced with similar difficulties of interpretation resulting from assuming such a meaning, had already devoted a chapter of his work on Making Sense of Tantric Buddhism (2013, 133–169) to discussing how caryā and related terms should be understood in early Buddhist and Śaiva tantric works.⁷ An exploration of this theme therefore now seems in some respects less pressing to me than it once did. But there are still some issues that can usefully be commented upon, and there are several early Śaiva attestations of the nexus of caryā-related terms of which Wedemeyer was not aware and which serve to adjust, I think, some of what he has said about this semantic field, and that go some way to explaining a significant juncture in the semantic voyage of the term caryā that led to its being commonly assumed in modern scholarship to mean something like “comportement normal,” even in passages in which such a meaning does not fit. Wedemeyer’s account does clarify a number of confusions, and he is to be commended for taking into account several Śaiva passages, but a combination of a desire to show that it is the Śaivas who have borrowed from the Buddhists rather more than the Buddhists have from the Śaivas here (2013, 136–137, 154) and of not having had access to the earliest known Śaiva material (which I should like to have made widely available long ago, but editing concurrently the Niśvāsa, the Kiraṇa and the Sarvajñānottara is proving a very time-consuming project) have led him to some problematic assertions and assumptions, some of which I hope to correct below.⁸ In what follows, I will be expanding upon and shoring up what was advanced rather too tentatively in a lengthy note on Mūlasūtra 4.17c–18 (Goodall, Sanderson, Isaacson et al. 2015, 284–287).

In fact, the basic difficulty with the central term caryā had arguably already been resolved, in nuce, by Alexis Sanderson in his 2006 article on the Lākulas, but in a somewhat laconic fashion. What he writes, just before presenting the vratas in the ninth chapter of the caryāpāda of the Mataṅgapārameśvara, is the following:

The Śaivas have conventionally divided the means of liberation taught in the Āgamas, that is to say their subject matter, into the four categories, ritual (kriyā), doctrine or gnosis (jñānam, vidyā), meditation (yogaḥ), and ascetic observance and other rules governing the conduct of the various classes and kinds of initiate (caryā). Continuities between the Lākulas and the Śaivas have now been shown in the areas of the ritual of initiation and in the doctrine of the path to liberation, […] Insufficient evidence exists to permit much of a comparison in the domain of meditation. […] This leaves only the domain of ascetic observance (vratacaryā).

It is clear, in other words, that caryā, in early Śaiva works, may refer specifically to ascetic observance, presumably indeed because it is a contraction of the collocation vratacaryā/vratacaraṇa, “the performance (caryā/caraṇa) of timed religious observances (vrata).” The verb car, “to move,” but also “to be engaged in,” has indeed long been the natural idiomatic verb of choice for use with vrata, and this accounts for the frequency of such bahuvrīhi expressions as cīrṇavrata (“who has observed his observances”), both in non-Mantramārga works (e.g. Bodhāyanagṛhyasūtra 4.12.2 on p. 118, Yājñavalkyasmṛti 3.298c, Mahābhārata 3.81.135c) and in works of the Mantramārga (e.g. Mālinīvijayottara 10.17c and 10.34c, Mohacūḍottara 1.14a, etc), as well as for the distinctively tantric bahuvrīhi expression cīrṇavidyāvrata (e.g. Siddhayogeśvarīmata 13.1a),⁹ to which we shall return below.

If we were concerned only with the meanings of the word caryā, then it might seem that we could almost end our essay here: Wedemeyer has pointed out that caryā conventionally refers to virtuous behaviour and conduct in accordance with religious precepts in a number of early non-tantric Buddhist texts (2013, 135), where he characterises it as “by far the most common generic term for the spiritual undertakings of buddhas and bodhisattvas,” just as it does in much later Śaiva works of the Mantramārga, such as the Mṛgendra; Sanderson has alluded (in the passage just quoted) to the observed fact that it may refer in Śaiva sources both to the prescribed “conduct of the various classes and kinds of initiate” as well as to “ascetic observance,” and he has pregnantly suggested that this second meaning is connected with the notion of vratacaryā; finally, Wedemeyer has observed that caryā in Buddhist tantric sources, and in some Śaiva ones, refers not to life-long virtuous conduct, but rather to timed antinomian practices, in troubling places such as cremation grounds and involving transgressive sexual and mortuary elements.

But, as my title indicates, there is in fact a nexus of terms to be examined here. Wedemeyer indeed points out that there are several other related terms that seem to be used in places where caryā might have served instead, caryāvrata and vratacaryā being apparently “used with identical meaning” (2013, 136), to which he adds instances of these words “in compound with qualifiers related to ideas of secrecy or madness,” such as guhyavrata, guhyacaryā, prachannavrata, unmattavrata, and “a cluster of interrelated terms that appear in the same contexts, and which seem to be largely synonymous,” which he tabulates on p. 137. Among these, he singles out vidyāvrata, for which he suggests the translations “knowledge observance, spell observance, and/or consort observance” (2013, 136) as being “treated as essentially equivalent to caryāvrata/vratacaryā in Buddhist and Śaiva sources.”

Now it may indeed be the case that several of these terms appear to be used interchangeably, but a slightly broader and chronologically deeper slice of Śaiva samples reveals, it seems to me, both how the terms in fact differ from each other and also why it is that they may in some contexts appear to be interchangeable, while at other times they are not. This may seem hair-splittingly tedious, but, as Wedemeyer points out, if we do not understand the words, then we cannot understand what it is that they serve to express.¹⁰ The Viennese endeavour that has so far produced three out of five volumes of the Tāntrikābhidhānakośa is a step towards a better understanding of technical terms and of common terms used with technical senses in the literature of the Mantramārga, even if it does not, alas, cover Tantric Buddhist literature.¹¹

Turning to the Viennese dictionary for an understanding of caryā is, however, not yet particularly useful, for the entry under this word consists only of a cross-reference to the term caryāpāda. But the account of that term does contain what will one day be a useful cross-reference to a future article on the term vrata, and it includes one useful pointer to a moment in the history of the term caryā:

Note that vrata is substituted as a synonym for caryā in Kir[aṇa] 6.6c; indeed it is conceivable that the term is an abbreviation of vratacaryā (Kir[aṇa] 49.4).

The volume of the Tāntrikābhidhānakośa in question appeared back in 2004; by the time the fifth volume appears, including the terms vrata and vidyāvrata, this dictionary will be a still more useful resource for tracing out the shifting semantics of this and many another nexus of tantric terms.

Let us follow up this reference to the Kiraṇatantra. Since its chapter, 49, on vratacaryā is short, we may quote much of it below, omitting from the middle a detailed treatment of the ideal kamaṇḍalu, and giving just the readings of the Devakottai edition (E_D) and the Nepalese manuscript of 924 ad (N, f. 70r):

garuḍa uvāca—

samayī putrakaś cāpi deśikaś ca maheśvara

eṣāṃ vṛttiḥ samākhyātā sādhakasya bravīhi me 49.1

bhagavān uvāca—

sādhakaḥ sātviko dhīraḥ sahiṣṇur mantradhīr varaḥ

apradhṛṣyo mahāprājñaḥ samaloṣṭāśmakāñcanaḥ 49.2

udyukto homaniṣṭhaś ca japadhyānarataḥ sadā

vighnaprotsāraṇe kalyo vrataniṣṭhaḥ samaḥ śuciḥ 49.3

sasakhāyo vanaṃ gatvā vratacaryāṃ samārabhet

asahāyo tadā tasmiṃ svasakhāyaḥ kamaṇḍaluḥ 49.4

1ab samayī putrakaś cāpi deśikaś ca] N; samayīsūtayoś cāpi deśikasya E_D ●1b maheśvara] E_D; maheśvaraḥ N ●1c eṣāṃ vṛttiḥ] E_D; eṣā vṛtti N ●2a sādhakaḥ sātviko dhīraḥ] E_D; sādhaka sātviko dhīra N ●2b varaḥ] E_D; vara N ●2c apradhṛṣyo] E_D; apradṛṣyo N ●3c °protsāraṇe] E_D; °procchāraṇe N ●3d śuciḥ] E_D; śucit N ●4a sasakhāyo] N; sasahāyo E_D ●4b vratacaryāṃ samārabhet] E_D; vratācaraṇam ārabhet N ●4cd tadā tasmiṃ svasakhāyaḥ] em.; tadā tasmiṃ svasakhāya° N; yadā tasmin susahāyaḥ E_D

Garuḍa spoke:

You have taught me, O great Lord, the activities of the neophyte, the putraka and the ācārya. Tell me those of the sādhaka. (1)

The Lord spoke:

The excellent (varaḥ) sādhaka [should be] full of sattva, firm, capable of endurance, his mind fixed on [his] mantra, unassailable, of great wisdom, looking impartially on mud, stones and gold, (2)

engaged, regular in [the performance of] oblations, always devoted to recitation and meditation, dexterous in the dispelling of obstacles, firm in [the practice of his] religious observance, calm, pure. (3)

Accompanied by his ritual assistant, he should go to the forest and begin the practice of his religious observance (vratacaryāṃ). [If he is] without a ritual assistant, then his spouted water-pot is his ritual assistant in that [practice].¹² (4)

[Description of spouted water-pot omitted.]

saśalyas tumbako vā syād evaṃ kṛtvā vratañ caret 49.13

jaṭāmakuṭasāṭopaṃ śūlakhaṭvāṅgalāñchitam

śuddhamuṇḍārdhasaṃyuktaṃ tṛlocanakṛtādaram 49.14

vyāghracarmāmbaraṃ śāntaṃ raudraṃ vratam idaṃ śubham

suniṣṭhasya bhavet ṣaḍbhir mmāsaiḥ siddhir ihottamā 49.15

madhyā māsaiś caturbhiś ca kṣudrā māsais tribhir bhavet

vratānām pravaraṃ raudram tatsiddhau sakalo bhavet 49.16

kāryaṃ mantravrataṃ siddhyai sādhakenānurūpakam

13c saśalyas tumbako] em.; saśalyatumbako N; saśalāstambhako E_D ●14c śuddha°] N; śuddhaṃ E_D ●14d tṛlocanakṛtādaram] N; trilocanakṛtodaram E_D ●15b raudraṃ] N; raudra° E_D ●15cd suniṣṭhasya bhavet ṣaḍbhir mmāsaiḥ siddhir ihottamā] N; kaniṣṭhasya bhavec chuddhir māsaiḥ ṣaḍbhir ihottamā E_D ●16ab caturbhiś ca kṣudrā māsais] E_D; caturbhi syāt kṣudra māsai N ●16c vratānām pravaraṃ] N; vratamapravaraṃ E_D ●16d sakalo bhavet] N; sakaḷaṃ punaḥ E_D ●17a kāryaṃ mantravrataṃ] E_D; kāryamantravrata N ●17b sādhakenānurūpakam] N; sādhakair nānurūpataḥ E_D

Alternatively, [instead of a kamaṇḍalu,] it may be a gourd with a shaft. Having made this [ready], he should practise his observance. (13cd)

This is the auspicious Raudra-vrata: imposing with a chignon of matted locks, marked by a trident and khaṭvāṅga, equipped with a clean half skull, awe-inspiring with a third eye, clothed in the skin of a tiger, peaceful. (14–15b)

For one firm [in this observance], the highest siddhi will arise in six months; middling [powers] in four months; the lowest [powers] will arise in three months. (15c–16b)

The highest of the observances is the Raudra[-vrata]. On accomplishing that, one becomes [equal to] the Sakala [form of Śiva]. (16cd)

For attaining siddhi, the sādhaka should perform a mantra-observance that is appropriate [to the mantra in question].¹³ (17ab)

What we see here, it seems to me, is a reflection of the old notion that caryā refers to vratacaryā, “the performance of a vrata,” where vrata is a timed religious observance that typically involves adopting an unusual diet (not mentioned here), an unusual style of dress (often with accoutrements of the cremation ground, in this case the khaṭvāṅga), and unusual behaviour (sexual transgressions, mortuary obsessions, or, as here, ascetic detachment). This observance is furthermore a preparation for the attainment of magical powers through the use of a mantra.

This is, I think, in essence, the same as what is meant by the term vrata in all early tantric literature. It explains therefore how vratacaryā, “the performance of such a vrata,” and therefore sometimes also caryā have come to be used interchangeably with vrata.¹⁴ But the term caryā evidently began to expand and then slip in meaning as the Mantramārga expanded to include not just sādhakas (who seem to be the only audience of the Niśvāsatattvasaṃhitā¹⁵), but also other categories of initiates. We can see that this slippage has in fact already taken place by the time of the Kiraṇa, for that work begins its thirty-first chapter with an announcement that the next topic to be taught will be caryā, and yet, as we have just seen above, does not deal with the vratacaryā of the Sādhaka until chapter 49, which follows eighteen chapters later.

Here is the beginning of chapter 31, in which the topic called caryā is first introduced in such a way as to suggest that the primary meaning has now become something like regular enjoined “comportment.”

garuḍa uvāca—

samayisutayor deva kā vṛttis tu dine dine

evaṃ mayi samācakṣva caryā me noditā purā 31.1

1a samayisutayor] N (unmetrical); samayīputrayor E_D ●1b vṛttis tu] E_D; vṛtyatra N ●1c mayi] conj.; mayā N; sarvaṃ E_D

Garuḍa spoke:

What are the day-to-day activities of the neophyte and the putraka? Tell me this. You have not taught me the [regular rules of] behaviour (caryā) before. (1)

No such clear evidence can be found of this broadening of the meaning of the word caryā in another post-Niśvāsa but pre-tenth-century Saiddhāntika scripture for which we have an early Nepalese palm-leaf witness, this time apparently of the ninth-century, namely the Sarvajñānottaratantra.¹⁶ That work gives us an account of another sort of vrata that will be useful to us in the discussion below, in this case somewhat more detailed, but involving no transgression of brahmanical rules about purity and sexual behaviour.

ataḥ paraṃ pravakṣyāmi vratānāṃ vratam uttamam¹⁷

śivavrateti vikhyātaṃ sadevāsurapūjitam¹⁸ 18.1

viśuddhaṃ pāṇḍaraṃ bhasma¹⁹ śuklavāsonulepanam

śuklayajñopavītī²⁰ ca jaṭāmakuṭamaṇḍitaḥ²¹ 18.2

sarvābharaṇasampannaḥ²² śuklamālyavibhūṣitaḥ²³

carubhug brahmacaryasthaḥ²⁴ śivāgnigurupūjakaḥ²⁵ 18.3

mantramūrtiḥ²⁶ śivasyaiva yathā rūpaṃ²⁷ prakīrtitam

tathā vai sādhakendrāṇāṃ²⁸ vrataṃ jñeyaṃ tadātmakam²⁹ 18.4

śuklakaupīnavāso vā uṣṇīṣākṣakamaṇḍaluḥ³⁰

śivālaye vasen nityaṃ³¹ bhikṣābhakṣo³² jitendriyaḥ 18.5

japadhyānarato³³ maunī śivāgnigurupūjakaḥ

saṃvatsare vyatikrānte śivatulyo bhaved asau³⁴ 18.6³⁵

ataḥ paraṃ pravakṣyāmi lakṣaṇaṃ³⁶ tu śivālaye

sthāpanaṃ caiva liṅgasya yasmin sarvaṃ pratiṣṭhitam³⁷ 19.1

brahmādyā devatāḥ sarvā liṅgam āśritya saṃsthitāḥ³⁸

tasmād vai sthāpayel liṅgaṃ³⁹ śāstradṛṣṭena karmaṇā 19.2

cīrṇavidyāvrato⁴⁰ yogī gurudevāgnipūjakaḥ⁴¹

Next, I shall teach the best observance among observances, which is known as the Śiva-vrata and which is revered by asuras and gods alike. (18.1)

Pure pale ash [should be used, and] white dress and unguents; he should wear a white sacred thread and be adorned by a chignon of matted locks. (18.2)

He should be equipped with all [suitable] ornaments, [and] adorned with white garlands; he should consume [only the pure ritual gruel-offering known as] caru; he should observe the chaste conduct of a student; he should venerate Śiva, the fire and his guru. (18.3)

He should be mantra-bodied;⁴² the appearance (rūpam) of excellent sādhakas [who follow this observance] is to be the same as that of Śiva: the observance must be understood as consisting in this. (18.4)

Alternatively, he may wear [just] a white loin-cloth, [and bear] a turban, rosary and spouted water-pot. (18.5ab)

He should dwell constantly in a temple of Śiva, eating alms, controlling his senses, devoted to recitation and meditation, maintaining silence, venerating Śiva, the fire and his guru. When a year has passed, he will become equal to Śiva. (18.5c–6)

Next, I shall teach the characteristics of a temple of Śiva, as well as [how to perform] the installation of the liṅga, in which the universe is [itself] ‘installed.’ (19.1)

All the gods, beginning with Brahmā, reside in the liṅga; therefore a yogin who venerates his guru, God and the fire and who has performed his vidyāvrata should install the liṅga, following the procedure taught in scripture. (19.2–3b)

We shall return below to the use of the term vidyāvrata, which I have not translated here. First we may observe that these passages of the Kiraṇa and Sarvajñānottara might appear to confirm Wedemeyer’s observation that the vratas in early Śaiva works were observances in which the sādhaka imitated God (2013, 165).

The early Śaiva Tantric paradigm for the transgressive vrata, then, was one of imitatio dei—mimicking the activity of the god in the interest of eliding the (presumably mistaken) sense of a gulf between him and the devotee. In none of these rites is there mention of transcendence of conceptuality or attainment of any epistemic nonduality—the concern seems entirely to be one of nonduality in the sense of union with the god Śiva.

Imitation of forms of god, whether pure or transgressive, seems indeed to be typical of vratas in early Śaiva sources, but it is not the invariable rule, as the Niśvāsa demonstrates.⁴³ Several vratas are described in the course of the work, but there is one passage in which a set of nine is concisely described together, namely in chapter 3 of the Guhyasūtra. A brief summary of the contents of that chapter will help to contextualise that description, showing that it is part of a chapter devoted to 1) preparations for magical pursuits, and 2) magical procedures:

Summary of Guhyasūtra 3

Preparations for magical pursuits

3.1–2 Having set up the God of gods in a suitable place, one may employ a ritual assistant (uttarasādhaka) for attaining the highest siddhi.

3.3–6 Qualifications of the uttarasādhaka.

3.7–11 Construction of a special dwelling for the pursuit of siddhi.

3.12c–16 Alternative: a suitable cave or empty temple. One should live from vegetables or begging or from roots, and perform fasts (cāndrāyaṇa, etc.)

3.17–22 Prognostication of success in siddhi by consulting Svapnamāṇavaka by calling him to appear in one’s sleep.

3.22–23 Prognostication by consulting Amoghamantrarāja.

3.24–27 Catoptromantic prognostication (prasīnā) using virginal children and the mantra of Caṇḍī.

3.28–29 japa using the akṣamālā [in order to prepare it].

3.31–43b vratas.

Magical procedures

3.43c–112 sādhanas for attaining various siddhis.

Having underlined the context, we may now turn to the vratas given in this chapter, followed, by way of example, by one magical procedure. We may note that all of these observances transgress social norms to some degree, but that none unambiguously involves imitatio dei:

siddhi-m-aiśvaryayogyas⁴⁴ tu na ca hiṃsanti hinsakāḥ 3.29

siddhavidyāvratastho hi jape⁴⁵ ca vratam ārabhet

go mātā ca pitā bhrātā atithir mitra brāhmaṇaḥ 3.30

hato me pāpa[[(cāre)]]ṇa⁴⁶ caren (1) Mithyāvratam vratī

<<karasthena kapā>>lena⁴⁷ khaṭvāṅgī bhasmaguṇṭhitaḥ 3.31

śmaśāne carate rātrau (2) Śmaśānavrata ucyate

nṛtyate gāyate caiva unmatto hasate bruvan⁴⁸ 3.32

bhasmāṅgī cīravāsaś ca (3) Gaṇavratam idaṃ smṛtam

japayukto bhaikṣabhujo loṣṭuśāyī jitendriyaḥ 3.33

dhyānasaṃyamayuktaś ca (4) Loṣṭukavratam ācaret

rikṣavyāghrasamā[[kīrṇe]]⁴⁹ <<vane siṃ>>hasamākule 3.34

jitanidrāśano⁵⁰ jāpī (5) Kāṣṭhavratam idañ caret

gāyate nṛtyate jāpī strīrūpī valayabhūṣitaḥ⁵¹ 3.35

śūrppakandukaveṇībhiś (6) Citravratam idañ caret

śastrapāṇir dayāyukta-m-aṭe trāteva⁵² †jatavān† 3.36

japadhyānārccanirato (7) Vīravratam idañ caret

varṣaśītātapair ddehan tāpayed dhi su--- 3.37⁵³

japadhyānarataś caiva (8) Mahāvratas sa ucyate

ratisaṃbhogakuśalāṃ rūpayauvanaśālinīm 3.38

īdṛśīṃ striyam āsādya niruddhendriyagocaraḥ

cumbanāliṅganaṅ kuryāl liṅgaṃ sthāpya bhagopari 3.39

japadhyānaparo bhūtvā (9) Asidhāravratañ caret

yadi kāmavaśaṃ gacchet patate⁵⁴ narake dhruvam 3.40

navātmakañ japel lakṣaṃ ((tasya)) ---ddhaye⁵⁵

abdaṃ ṣaṇmāsamātraṃ vā yaś cared vratam uttamam 3.41

tasya siddhiḥ prajāyeta adhamā madhyamottamā⁵⁶

vratasthaḥ⁵⁷ pañcalakṣāṇi punar japtvā tu siddhyate 3.42

sarve mantrāś ca siddhyante īpsitañ ca phalaṃ bhavet

[A spell for travelling great distances:]

oṃ namo vāyupathacāriṇe amitagatiparākramāya vimale kulu 2⁵⁸ ṭhaṭha 3.43

śilā suvarṇadhātuñ ca ((ka)) --- tam

varāhavaśasaṃpiṣṭaṃ⁵⁹ sahasraparimantritam 3.44

navātmakam punar japtvā pādau caiva pralepaye[[t]]

gacchate so ’pariśrānto yojanānāṃ śatadvayam 3.45

[Once the rosary has been thus prepared, he becomes] ready for siddhis and power. (29cd)

Dangerous creatures do not harm one who has [first] accomplished an observance [that qualifies one] for [using] Spells: he should begin an observance by means of recitation (jape[= japena]).⁶⁰ (30ab)

The one engaged in observance should practise the False Observance (mithyāvrata) [by wandering about proclaiming]: “I have committed bad deeds: I have killed a cow, mother, father, brother, a guest, friend, brahmin!” (30c–31b)

[If] one wanders in the cremation-ground at night, with a skull in one’s hand and a khaṭvāṅga, covered in ashes, that is called the cremation-ground observance (śmaśānavrata). (31c–32b)

If one dances, sings, laughs and talks madly, with the body smeared in ashes and wearing rags, this is called the Gaṇavrata. (32c–33b)

One performs the Clod-of-Earth Observance (loṣṭukavratam) by being engaged in recitation, feeding on alms, sleeping on the earth, with senses controlled, engaged in meditation and restraint. (33c–34b)

One may perform the Block-of-Wood Observance (kāṣṭhavratam) in a forest full of bears, tigers and lions, conquering the urges to sleep and eat, [constantly] reciting. (34c–35b)

If one takes on the appearance of a woman and sings and dances, adorned with bracelets, with a winnowing fan, ball and plait, one observes the Colourful Observance (citravratam). (35c–36b)

With a weapon in hand, full of compassion, if one wanders like a saviour of creatures (?)⁶¹ focussed upon recitation, meditation and worship, one performs the Warrior Observance (vīravratam). (36c–37b)

If one torments the body with rain, cold and heat, …, devoted to recitation and meditation, this is called the Great Observance (mahāvrataḥ). (37c–38b)

A woman skilled in the pleasures of love-making, endowed with beauty and youth; such a woman one should procure, holding one’s senses back from the objects of the senses, and one should kiss and embrace [her], placing the penis upon her sex while remaining focussed upon recitation and meditation—one performs [thus] the Sword-Blade Observance (asidhārāvratam). If one should succumb to the control of desire, then one certainly falls into hell. (38c–40)

One should recite the navātman one lakh times … for [si]ddhi: one who [thus] observes such an excellent observance for a year or just six months attains lowest, middling or best siddhi. But if, while observing such a vrata, someone recites five lakh times, then [that mantra] succeeds [for him] (siddhyate), and all mantras succeed for him and he attains the fruits he desires. (41–43b)

[Using the mantra] oṃ namo vāyupathacāriṇe amitagatiparākramāya vimale kulu kulu svāhā, [and taking] arsenic, gold [and?] a mineral, …, ground up with pig fat/marrow, over which one has recited [the navātman] 1000 times, he should smear [the mixture] on his feet/legs, while once again reciting the navātman: he will travel 200 yojanās unwearied! (43c–45)

Right at the beginning of the above-quoted passage, we find a further attestation of the term vidyāvrata, and this time, rather than prevaricating further, I have proposed translating it as “an observance [that qualifies one] for [using] Spells.” There are other passages that can and will be adduced in support of this, but I think that it should already be becoming clear from this passage of the Guhyasūtra and from the passage quoted just before from the Sarvajñānottara that the different particular vratas that are performed serve to prepare the performer for some subsequent religious activity that involves the use of the mantra or vidyā. In the case of the Guhyasūtra, it is the pursuit of siddhi for which the sādhaka is prepared; in the case of the Sarvajñānottara, the individual is prepared for the performance of the installation (pratiṣṭhā) of a liṅga. From the beginning of chapter 10 of the kriyāpāda, we learn that it is also an essential to the consecration of an ācārya in the Mataṅgapārameśvara:

catuṣpādārthakuśalaṃ mahotsāhaṃ hy aninditam 10.2

ṣaṭpadārthapraṇītārthaṃ sarvabhūtahite ratam

gurus tam abhiṣiñcet tu cīrṇavidyāvrataṃ naram 10.3

Rāmakaṇṭha: atha kiṃ tad vidyāvrataṃ yat tena cīrṇam ity ucyate:

vidyāśaktir ihopāttā japtavyā prāk chivālaye

saṃniyamyendriyagrāmam abdam ekaṃ śuciṣmatā 10.4

nityaṃ carubhujā bhūmyāṃ kuśaprastaraśāyinā

pūjāgnibhavane yuktacetasā bhāvitātmanā 10.5

Rāmakaṇṭha: [vidyāśaktiḥ] vyomavyāpilakṣaṇā.⁶²

The guru should consecrate [as an ācārya] a man who is skilled in what is taught in all four pādas, who has great energy, who is beyond reproach, who expounds the meaning of the teachings [encapsulated] in the six topics [of this scripture], who is devoted to the welfare of all beings, who has performed the observance for [the propitiation of his] mantra.

Rāmakaṇṭha: Now if you ask what this vidyāvrata is which he must have observed, this is what the text teaches:

The power of the vidyā that is mentioned here [in this compound vidyāvrata] is first to be recited for a year in a temple to Śiva, while exercising control of the senses, maintaining purity, eating daily [only the sacrificial gruel known as] caru, sleeping on the ground in the room reserved for pūjā and fire[-sacrifice] on a spread of kuśa-grass, with his mind engaged [in meditation], focussed.

Rāmakaṇṭha: It [viz. the power of the vidyā] is the vyomavyāpin.

Once this preparation, taking the form of the observance of one among a variety of possible vratas (but ideally one suitable to the mantra to be put to use, as seems already to be implied in Kiraṇa 49.17ab above, and as we shall see confirmed below), is complete, the observer can be called cīrṇavidyāvrataḥ (as here and in Sarvajñānottara 19.3a) or siddhavidyāvratasthaḥ (as in Guhyasūtra 3.30a).

If we make such an assumption, then we can see how the expressions vratacaryā and vidyāvrata might be regarded as interchangeable in some contexts,⁶³ even though they are not actually synonymous. It also becomes clear how it is that vidyāvrata can be characterised as preparatory mantra-propitiation and therefore equivalent to what may also be called puraścaraṇa. Sanderson characterises it in such a way when referring to the account of the Bhairavācārya in Bāṇa’s Harṣacarita:

With this account of a pūrvasevā, also called puraścaryā or vidyāvratam, i.e. the initial period of ascetic japaḥ etc. to be undertaken after one has received a Mantra, whereby one becomes able to accomplish feats (karmāṇi) with that Mantra …⁶⁴

So if vrata and vratacaryā and some other terms may seem to be used interchangeably in some works with vidyāvrata, it is because the principal purpose of the vratas taught in the early Mantramārga seems to be to propitiate mantras prior to further religious activities involving those mantras, rendering those who complete such observances describable by such terms as cīrṇavidyāvrata, siddhavidyāvratastha, vidyāvratasnāta and so forth.⁶⁵ Various observances can, in other words, be observed in order to become one “who has completed the observance [required for the propitiation] of a vidyā.”

This discussion might seem to suggest that in finding the original meaning of the expression vidyāvrata we believe that we have found its immutable semantic core, but that is not really what I intend to say. Of course the term vidyāvrata may have gone on to evolve and be used in contexts that suggest that one might elsewhere also or instead render it as “knowledge observance” or “consort observance” (Wedemeyer 2013, 136) or, as we shall see below, “observance relating to a vidyāṅgamantra.”⁶⁶ Furthermore, one might argue that we have in any case not started from its point of origin, for the term has presumably been drawn from or at least coloured by the brahmanical expression vidyāvratasnāta that is common from the Gṛhyasūtras onwards and that we find, for instance, in Manusmṛti 4.31:

vedavidyāvratasnātān śrotriyān gṛhamedhinaḥ

pūjayed dhavyakavyena viparītāṃś ca varjayet

Olivelle (2005, 125–126) translates:

At rites for gods and ancestors, he should honor individuals who have bathed after completing the Vedas, vedic learning, or vedic vows, who are vedic scholars, or who are householders, but avoid individuals different from these.⁶⁷

It seems to me very likely that the use of vidyāvrata in the Mantramārga—and a fortiori of vidyāvratasnāta (Siddhayogeśvarīmata 10.20 and Svāyambhuvasūtrasaṅgraha 21.35)—should have been influenced by earlier brahmanical usage such as we see in the Gṛhyasūtras and in the Manusmṛti. Nonetheless, in the Niśvāsa certainly, and probably throughout the early Mantramārga, the use of vidyāvrata to mean “observance for [the propitiation of] a mantra” seems to be the norm. As Sanderson has observed in the note of his that we have just quoted, vidyāvrata seems indeed to be used in the same way as pūrvasevā in the Niśvāsa. Many short paragraphs of prose in the grimoires (kalpa) that we find in the Guhyasūtra sketch out the essential features of particular observances, and these paragraphs are very often concluded with a succinct statement of the magical powers that can be won by following them (the power to fly, for example, or to disappear); but sometimes we find instead the assertion that the observance fulfils the requirements of pūrvasevā (10.27, 10.99, 14.26) or puraścaraṇa (14.24) or, as here in Guhyasūtra 10.91, the requirements of the vidyāvrata:

devaṃ pūjyāgnau juhuyād audumbarasamidhānāṃ tryaktānāṃ sahasraṃ tṛsandhyaṃ. kṣīrāśī saptā dināni juhuyāt. cīrṇṇavidyāvrato bhavati.

Having worshipped the Lord, he should oblate into the fire at the three junctures of the day a thousand pieces of Udumbara-wood smeared with the three [sweet substances]. Consuming [only] milk, he should make oblations [in this manner] for seven days. He will become one who has accomplished the vidyāvrata.

Before I wrap up the discussion, it should be mentioned that a different hypothesis as to the meaning of vidyāvrata was advanced some years ago, when many of the above-cited passages had not yet come to light, by Judit Törzsök when translating chapter 10 of the Siddhayogeśvarīmata. We quote here the beginning of the chapter (without the apparatus) from Törzsök’s 1999 edition:

devy uvāca

mayā deva purā pṛṣṭaṃ vratayāgavivarjitam

siddhayogeśvarīṇāṃ tu mataṃ mantraprasādhakam 1

kiṃ tu deva pratijñātaṃ siddhir vidyāṅgasaṃsthitā

tasmāt teṣu samāsena vratacaryāṃ bravīhi me 2

bhairava uvāca

ādau tu sarvasiddhyarthaṃ sarvavighnavināśanam

sarvapāpāpanodārthaṃ vidyāvrataṃ samārabhet 3

sādhakaḥ sādhakī vātha mantratadgatacetasaḥ

yāgaṃ kṛtvā vidhānena vratacaryāṃ samācaret 4

bhasmalepitasarvāṅgo maunī śuklāmbaraḥ sudhīḥ

sitayajñopavītaś ca akāmo niyame sthitaḥ 5

Here is Törzsök’s translation (1999*, 143):

The Goddess spoke

I have previously asked you about the Doctrine of the Yoginīs (Siddhayogeśvarīmata), O God, which helps to make mantras effective (mantraprasādhakam) without any observances or worship. (1)

However, you have asserted, O God, that success depends on the ancillary mantras; therefore, tell me briefly about how to practise the observance[s] associated with them (teṣu). (2)

Bhairava spoke

First [before any other practice to attain a specific supernatural power], for all kinds of supernatural powers, [and] for expiatory purposes, one has to start the observance of the [ancillary] mantras, which destroys all obstacles. (3)

The male or female practitioner, with his/her mind focused on the mantra, should perform worship according to prescriptions and then undertake the vow (vratacaryāṃ). (4)

[In the first of these] all his limbs covered with ashes, the practitioner is to observe silence and should wear a white garment; he should be of good understanding. He must have a white sacred thread, he should be free from desire and established in self-restraint. (5)

Now the reason that Törzsök translates vidyāvrata with “the observance of the [ancillary] mantras” is that each of the vratas in the chapter is specific to the cultivation of a particular auxiliary mantra—the first one, given above in verse 5, must, by elimination, be an observance for the hṛdaya-mantra—and those auxiliary mantras belong to a set known in this work and in others as the vidyāṅgamantras.⁶⁸ It was therefore reasonable for her to assume that vidyāvrata was short for vidyāṅgavrata, for she had no evidence to suggest otherwise, and she had parallel evidence that seemed to reinforce this hypothesis, namely the testimony of Svāyambhuvasūtrasaṅgraha 21 (from which Törzsök quotes, citing Alexis Sanderson’s collation, in her notes on p. 78).

That passage again gives a series of vratas, which are, by the way, again not instances of imitatio dei, and which are again specific to the vidyāṅgamantras; so it is wholly understandable that this seemed to Törzsök in 1999 to confirm the notion that the element vidyā in the collocation vidyāvrata must refer to the vidyāṅgamantras. I think, however, that it will now be clear that chapter 21 of the Svāyambhuvasūtrasaṅgraha might in fact be interpreted equally well internally if we assume that vidyāvrata is used instead to mean “observance for [propitiation of] a mantra” and that, given the other attestations of the term that we now know about in, for instance, the Guhyasūtra, the Mataṅga, and the Sarvajñānottara, it actually makes better sense to assume this broader interpretation in this passage too.

There is somewhat better evidence for pinpointing the place of the Svāyambhuvasūtrasaṅgraha in a relative chronology of Saiddhāntika writings than there is for most other pre-tenth-century Siddhāntatantras, for in terms of both doctrinal and social developments, it seems later than the sūtras of the Niśvāsatattvasaṃhitā (see Goodall, Sanderson, Isaacson et al. 2015, 41–44, 47–50, 58), and yet it cannot be later than Sadyojyotiḥ, who has written a commentary upon it and who, Sanderson argues (2006b, in particular p. 76), lived between c. 675 and 725 ad. The edition is not widely accessible, which may be why Wedemeyer did not refer to this passage, and its text almost invariably needs to be corrected against manuscripts, but this particular chapter has just been published anew, in the form in which it appears when quoted by Hṛdayaśiva in his Prāyaścittasamuccaya (see Sathyanarayanan 2015, Appendix chapter 10).⁶⁹

japtvā vrataiś ca vidhivat snātaḥ siddhyai japet manum 10.27

na vilambitam aspaṣṭan na cāsvīkṛtam adrutam

nāsaṃkhyaṃ na manobhrāntaṃ japaṃ kuryād vicakṣaṇaḥ 10.28

sitavāsāḥ sitoṣṇīṣī sitayajñopavīty api

sitānulepanasragvī cared vidyādhipavratam 10.29

raktāmbaradharo mantrī raktamālyānulepanaḥ

māsam ekam puroktañ ca cared brahmaśirovratam 10.30

pītavāsāś caren māsaṃ pītamālyānulepanaḥ

pītayajñopavītī ca rudrāṇyā vratam uttamam 10.31

vrataṃ puruṣṭutasyāpi māsam ekaṃ cared budhaḥ

sarvakṛṣṇopacāreṇa śivārcanarataḥ sadā 10.32

citrāmbaradharo dhīraś citramālyānulepanaḥ

sarvasiddhiparo mantrī caret pāśupataṃ vratam 10.33

vratavratasamāptau ca kalaśena śivāmbhasā

svamantraparijaptena svātmānam abhiṣecayet 10.34

evaṃ vidyāvratasnātaḥ sarvatrādhikṛto ’naghaḥ

japen mantram anudvignaḥ svakalpavidhinā tataḥ 10.35

27c vrataiś] H, Ed.; vrajaiś N ●27d manum] H, Ed.; matam N ●28a na vilambitam] H, Ed.; avilambitam N ●28c manobhrāntaṃ] H; mobhrāntaṃ N (unmetrical); manobhrānta° Ed. ●29ab sitavāsāḥ sitoṣṇīṣī sita°] em.; sitavāsā śitośnīṣī sita° N; śitavāsā śitoṣṇīṣī śita° H; sitavāsāḥ sitoṣṇīṣaḥ sita° Ed. ●29c sitānulepanasragvī] conj.; sitānulepanaḥ sragvī N, Ed.; śitānulepanaḥ sragmī H ●29d vidyādhipa°] Ed.; vidyādhipaṃ N, H ●30a raktā°] H, Ed.; rattā° N ●31a pītavāsāś] N, Ed.; pītavāsā H ●31c rudrāṇyā] N, Hac, Ed.; rudrāṇī° Hpc ●32a puruṣṭutasyāpi] N, H; puruhūtasyāpi Ed.pc; purutasyāpi Ed.ac ●33c sarvasiddhi°] N, H; sarvasiddha° Ed. ●33d pāśupataṃ] N, H; pāśupata° Ed. ●34a vratavratasamāptau ca] Ed.; vrataṃ vratasamāptau ca N; vratavrataṃ samāptau tu H ●35a vidyāvrata°] N, H; vidyāvrataḥ Ed. ●35b sarvatrādhikṛto] N; sarvatrāvikṛto H; sarvato dhikṛto Ed.

Having recited [a particular mantra] along with [the practice of one of the] observances in accordance with the rules, and having bathed [at the end of the observance], one may recite that mantra for attaining supernatural powers. (27cd)

The skilled practitioner should do his recitation not too slowly, not indistinctly, not without taking [the meaning of what he recites] in, not too fast, not without counting, and not with his thoughts in confusion. (28)

Dressed in white, with a white turban and a white sacred thread and white unguents and garland, he should perform the observance for the vidyādhipa-mantra. (29)

Dressed in red garments and red garlands and unguents the Mantrin should first perform for one month the stated observance for the brahmaśiraḥ. (30)

Wearing yellow garments and yellow garlands and unguents and a yellow sacred thread he should perform the excellent observance of rudrāṇī for a month. (31)

The competent ritualist (budhaḥ), constantly devoted to the worship of Śiva, should perform the observance for puruṣṭuta for one month with all accoutrements being black. (32)

The mantrin, intent on attaining all manner of special powers, should perform the observance for the pāśupatāstra resolutely (dhīraḥ) dressed in multi-coloured garments and with multicoloured garlands and unguents. (33)

And upon the completion of one or another of these observances (vratavratasamāptau),⁷⁰ he should pour upon himself Śiva-water that has been consecrated by recitation of his mantra over it from a pot. (34)

Being thus bathed after the observance [in propitiation] of [his] mantra, invested in the right to [pursue] all [manner of special powers], faultless, he should then recite [his chosen] mantra according to the rules of his hand-book,⁷¹ without being afraid. (35)

In the above lines, 27d and 35 seem to make particularly plain that these vratas are performed by way of pūrvasevā, also known as vidyāvrata, as a preliminary to the pursuit of siddhi.

Csaba Kiss, following Judit Törzsök’s lead, has also alluded to the link between vidyāṅgamantras and the name vidyāvrata, but I think it will be clear from what he says below that the evidence that his new edition of parts of the Brahmayāmala has recently brought to the discussion again supports rather the broader interpretation, in which the element vidyā alludes to any mantra, not just a vidyāṅgamantra. Below are his remarks (2015, 211) on Brahmayāmala 21.4–5b, which he constitutes and translates as follows:

ete nava vratā proktā vidyābhede vyavasthitā

eteṣāṃ tu yathānyāyaṃ yathā caryā bhave tv iha

kathayāmi mahādevi tan me nigadataḥ śṛṇu

These are the nine ascetic observances (vrata), corresponding to [the syllables of] the Vidyā[, Caṇḍā Kāpālinī’s nine-syllable mantra]. I shall now teach you how to perform them correctly, O Mahādevī. Listen to me [while I] teach you.

… these observances are in fact called the vidyāvratas in 21.10c, 42d, 47a, 51b, 53b, 75d and 102d; the nine types of observances obviously correspond to the nine syllables of the Navākṣaravidyā (oṃ caṇḍe kāpālini svāhā), taught in BraYā 2; vidyāvratas may also serve, as seen in many tantric texts, as preliminary propitiation, by the use of vidyāṅgamantras, of a mantra to be applied later, or simply for the purpose of gaining mastery over the Vidyā, similarly to the way it is taught in the Yoginīsaṃcāra; as Sanderson (2009:134 n. 311) remarks: …

More closely parallel to these nine observances, as we can now see, are the nine taught in chapter 3 of the Guhyasūtra, which are probably each for one of the nine elements of the navātman. So perhaps the association with vidyāṅga-mantras is simply a red herring.

Kiss’s reference there to Sanderson 2009, 134, note 311 proves to be another passage in which Sanderson reveals that, although he did not spell out every detail of his assumptions and the evidence upon which he based them, he had in fact already assumed the interpretation for which we have been somewhat long-windedly arguing here, both of vidyāvrata and of its relation to individual named vratas:

The Yoginīsaṃcāra requires anyone who has gone through its initiation ceremony and then received consecration (abhiṣekaḥ) to adopt one of three forms of ascetic observance in order to gain mastery over the Vidyā (vidyāvratam): the Bhairavavrata, the Cāmuṇḍāvrata, or the Triṣaṣṭikulavrata, the observance of the sixty-three families [of the Mothers], which it also calls the Kāpālavrata, i.e. the Kāpālika.

So let me reiterate my conclusion: we should probably assume that, even if the expression vidyāvrata was originally drawn from the common brahmanical expression vidyāvratasnāta that we saw in Manusmṛti 4.31, and even if it may have been subsequently coloured in some contexts by other associations of the word vidyā (vidyāṅgamantra, “knowledge,” “consort”), the expression vidyāvrata appears throughout the early Mantramārga to be used with the understanding that it refers primarily to an “observance for [the preliminary propitiation of a] mantra.” While some works (such as the Saiddhāntika Kiraṇatantra, and the Sarvajñānottaratantra) appear to mention only one way of fulfilling the requirements of the vidyāvrata, many others (Niśvāsa, Yoginīsaṃcāra, Brahmayāmala, Svāyambhuvasūtrasaṅgraha, Siddhayogeśvarīmata, Tantrasadbhāva, Kubjikāmata) teach several vratas, not all of which are equally transgressive and not all of which involve imitatio dei, as alternative ways of realising the vidyāvrata.

Finally, let me show how this assumption seems to me to throw light even on passages where none of the terms that we have been discussing actually occur. The first few times I read the beginning of chapter 9 of the Guhyasūtra, it seemed to begin with a curious non sequitur: the goddess asks a question about how the alphabet, treated as mantra, can be used to bring about supernatural power and liberation; Śiva’s reply, however, first explains at some length how someone should dress half his body as the goddess and the other half as Śiva. How could this, I asked myself, be an answer to the goddess’ question? Was the apparent incoherence of the text here an indication that it had become corrupt?

devy uvāca—

mātṛkāyā bhavet siddhir mmokṣañ caiva maheśvara

mātṛkāsiddhim ākhyāhi mokṣañ caiva yathā bhavet 9.1

tatsamutthāś ca ye mantrāḥ kimarthaṅ kathitās tvayā

etat praśnavaraṃ brūhi ((bha))---phalaṃ hi me 9.2

īśvara u—

arddhastrīveśadhārī tu arddhena puruṣas tathā

arddhena alakaṃ kuryād arddhenaiva jaṭādharaḥ 9.3

tilakārddhena netrārddhe vālikā hy ekakarṇṇake

kuṇḍalaṃ hy ekakarṇṇe tu śūlan dakṣiṇahastataḥ 9.4

vāmapārśve⁷² stanaṅ kuryād vāmārdhe caiva mekhalām

valayaṃ vāmahaste tu vāmapāde tu nūpuraṃ 9.5

rucakaṃ dakṣiṇe pāde muñjamālāṃ tathā kaṭau⁷³

kaupīnan dakṣiṇe kuryād vāme strīvastradhāritā 9.6

śūrppaṃ vāmakare gṛhṇed ardhanārīśvaravrate⁷⁴

etad vrataṅ gṛhītvā tu bhikṣāśī tu jitendriyaḥ 9.7

japahomarato nityam pratigrahavivarjjitaḥ

triṣkālam⁷⁵ arccayed devaṃ triṣkālaṃ snānam ācaret 9.8

śākayāvakabhikṣāśī skandamūlaphalāśinaḥ⁷⁶

māsam e[[ka]]---samanvitaḥ 9.9

mucyate ’sau⁷⁷ mahāpāpāt kṣudrasiddhiñ ca vindate

dvimāsān madhyamā siddhir abdārddhād uttamā bhavet 9.10

saṃvatsareṇa siddhis tu vidyāsiddhim avāpnuyāt

aṇimādyās tu jāyante siddhaiś ca saha modate 9.11

īpsitāṃ labhate kāmān akāmo mokṣam āpnuyāt

The Goddess spoke:

From the mātṛkā supernatural power and liberation can come about, O Lord. Tell me [how to attain] supernatural power and liberation through the mātṛkā. (1)

Why did you teach the mantras that arise from it? Tell me [the answer to] this excellent question. … fruit to me. (2)

The Lord spoke:

Wearing half the dress of a woman and half [that of] a man, on one half, he should place [feminine] tresses, on one half, he should wear matted locks. (3)

On one half, there should be a forehead mark; on one half a [forehead] eye. A ring (vālikā) [should be] in one ear; a [pendant] ear-ornament (kuṇḍalam) in one ear. He should put a trident (śūlam) in his right hand and a breast on his left side, a girdle (mekhalām) on the left half, a bangle on the left arm, a woman’s anklet on the left leg, a man’s anklet on the right leg and a muñja-grass belt. At the hips, he should put a loin-cloth on the right and wear a woman’s garment on the left. (4–6)

In the left hand, he should hold a winnowing fan in the observance of Ardhanārīśvara. Adopting this observance he should eat alms, keep his senses under control, be devoted to regular obligatory recitation and oblation, rejecting the receipt of gifts. (7–8b)

He should venerate God three times [a day] and perform ablutions three times [a day]. Eating vegetables and barley-gruel, eating bulbs (skanda° [= kanda°]), roots and fruits, for one month … (9)

He will be released from [the retributive force of] major transgressions; and he will attain low siddhis after two months, middling siddhis after half a year and high siddhis after a year; he will attain power over the spell (vidyāsiddhim). (10–11b)

The ability to make himself atomic, along with the others [of the yogic powers], will arise. He will take pleasure in the company of siddhas. He will attain the wishes he desires; if he is without desires, he will attain liberation. (11c–12b)

Although the words vrata, caryā and vidyāvrata are none of them to be found, it is now clear to me that this passage makes implicit allusion to the structure now familiar to us from numerous other passages: the sādhaka propitiates a given mantra, here the mātṛkā, by performing a timed religious observance involving unusual dress and diet, the rules of which are ideally held to be in some way appropriate to the mantra in question, and then becomes eligible for the pursuit of particular siddhis. In the case of the mātṛkā, adopting the appearance of Ardhanārīśvara is particularly appropriate because the mātṛkā is made up of feminine vowels and masculine consonants, which, as is explained elsewhere in the Niśvāsa-corpus, are to be applied respectively to the left and right halves of the sādhaka’s body before worship in a preliminary rite that prefigures what came to be called sakalīkaraṇa.⁷⁸ The imitatio dei that is such a prominent feature of some observances, such as this one, now seems as if it should be more precisely characterised as identification with the mantra-deity being propitiated.