Most ancient Tantric texts, and some more recent ones too, contain a mixture of sometimes very elaborate theological and metaphysical notions with a mystical and soteriological aim, together with a variety of magical speculations and practices, these two categories of notions and actions forming two very largely overlapping areas rather than two separate ones. A coincidence that should not surprise us since rites are governed by rules, that is, by theoretical notions which give them their meaning. The Tantric initiate for whose benefit such texts were composed, the sādhaka, was normally not a mumukṣu, a seeker of liberation, but a bubhukṣu, a religious virtuoso (to use Weber’s terminology) intent on obtaining supernatural powers or rewards and for whom mokṣa was the Tantric liberation in life, jīvanmukti, with its accompanying siddhis. These texts thus never left aside entirely the quest for liberation, as appears for instance in the Trika system of upāyas expounded mainly in Abhinavagupta’s Tantrāloka (TĀ) on the basis of earlier Tantras such as the Mālinivijayottara (MVT) and the Siddhayogeśvarīmata. Considering the fundamental role of mantras both in the quest for liberation and in the acquisition and practice of powers, it is not surprising that in all these works mantras are described as effective in those two fields. The mantras used for these various purposes are sometimes different; but they may also be the same ones. I would like to consider briefly here a group of mantric practices used when magical effect are aimed at but which can also be used for other reasons – these two different sorts of purposes being admittedly not easy to distinguish.
I shall consider here a passage from the Netratantra (NT),2 a Śaiva text (undatable, but surely not later than the tenth century AD), which was commented upon by Kṣemarāja (fl. c.1000–1050), a disciple of Abhinavagupta, and was interpreted by him in the spirit of the non-dualistic Śaiva system of the Kashmirian exegesis of the Trika-Pratyabhijñā. The NT (often called by Kṣemarāja Mr̥tyujitbhaṭṭāraka) extols the ‘Eye’ (netra) of Śiva (sometimes also called Amr̥teśvara or Amr̥teśa) and especially the mantra of that deity, the netramantra (also named mr̥tyujit or amr̥teśvaramantra), the basic form of which is Oṃ juṃ saḥ. The text describes at length the mystical, yogic and ritual or magical uses of that sacred formula.
Chapter 18 of NT describes a ritual called śrīyāga thanks to which the ‘good sādhaka’ initiated in the netramantra will be able to oppose those who try to counteract the action of his mantra or to make it turn against him. This ‘countermagic’ practice is technically called pratyaṇgirā, pratyaṅgirasa or pratyaṅgiratva3. To make his own mantra irresistible, the sādhaka first submits it to the action of some ritual treatments, of which nine are enumerated.4 One may, however, the tantra adds (śl. 9), give the mantra eleven different forms ‘thanks to which the sādhakas will succeed’, that is, they will obtain what they are seeking. śl. 10b–12a list these forms:
samputaṃ grathitaṃ grastaṃ samastaṃ ca vidarbhitam /
ākrantaṃ ca tathādyantaṃ garbhasthaṃ sarvatovr̥tam /
tathā yuktividarbhaṃ ca vidarbhagrathitam tathā //
ity ekadaśadhā mantrā niyuktāḥ siddhidāḥ smr̥tāḥ /
Thus listed are eleven different ways of interlocking the netramantra with the name (or designation) of the person aimed at, or the action to be effected: the nāma, abhidheya or sādhya.
The grammar of the above list, given here as printed in the KSTS edition of the NT, is not satisfactory. It is clear, however, that it enumerates some things that are to be done to the mantra, and this is indeed how Kṣemarāja explains these terms in his Uddyota (vol. 2, pp. 77–79), confirming his point, in each case, with a quotation from an unspecified text, probably the Ucchuṣmatantra. His definitions coincide with those given for several of these terms in other Tantric works (of which I have checked only a few), the practices listed here being fairly common in magic, especially for the so-called ṣaṭkarmāṇi, the ‘six [magical] actions’ described in a number of Tantric texts whether Hindu or Buddhist.
Here are these eleven terms – and the corresponding practices:
Saṃpuṭa: the word means a bowl or a rounded casket, or the space enclosed inside these objects, or between two covers, etc. The mantra (or any part thereof, or any element with which it is associated) must be symbolically ‘encased’ inside such a space or casket, covered as with a lid or within two covers.5 Kṣemarāja defines it here as the placing of the mantra before and after the other element which is thus encased, as it were, within it: ādyantayor mantranyāsaḥ sa saṃpuṭavaṭ The pattern is thus B A B (where A is the mantra and B the nāma/abhidheya / sādhya). The same practice is prescribed in NT 6.18, which adds (19) that one may thus ‘encase’ any mantra within the amr̥teśvaramantra and, if it is thus recited, this mantra, whether or not intrinsically powerful, will immediately become efficacious (NT, vol. 1, p. 131). The technique of saṃpuṭa (sometimes called puṭa) may also be applied to parts of a mantra, an akṣara of which being thus encased between another one repeated twice (cf. ŚT 7.52), the sequence is then b1 a1 b1 b2 a2 b2, etc.
One may note that in NT 6.18–19 the technique of saṃpuṭa is prescribed for a japa, that is, is used orally; but in the Uddyota on NT 6.18 Kṣemarāja describes it as written: mantram ādau likhet. In the Agnipurāṇa (AgP), chapter 138, where saṃpuṭa is prescribed for the magical action of vaśīkaraṇa and ākarṣaṇa, it is described as the placing of the mantra around the sādhya – above, under and to its right and left: a spatial pattern, not an oral operation. The mantra, in this case, though theoretically a sound pattern, not a written series of letters, is mentally conceived as disposed in that way whilst it is uttered.
grathita this practice is also known as grathana – from the verbal root GRATH (GRANTH), to bind. It consists in binding together, as it were, the mantra and the sādhya, as in a knot, by alternating their syllables, ‘each syllable of the abhidheya being enclosed within those of the mantra’, according to the quotation made by Kṣemarāja: abhidheyārṇam ekekaikaṃ mantravarṇaiḥ saṃpuṭīkr̥tam. These words are also quoted by Jayaratha in his Vivaraṇa on the VM 2.35–6 (or NṢV, 2.36–37, p. 157),6 which prescribes the enclosing of each letter of the sādhya within the bīja of Kāmakalā (ĪṂ), this being done (in writing) on a diagram, around the border of which all the letters of the alphabet are to be inscribed.7 The AgP 138.6, the Tantrarājatantra (TRT)1.72, or the ĪŚGP 1.5 give the same definition of grathana as Kṣemarāja.
grasta the term means swallowed or eclipsed, but also surrounded. Its use, for a practice consisting (according to Kṣemarāja) in putting the name or sādhya in the middle, with the mantra written on all four sides (madhyasthasya nāmno dikcatuṣṭaye mantraniveśaḥ), appears thus wholly justified.8 This practice is in fact what the AgP 138 calls saṃpuṭīkaraṇa.
In its present sense grasta is not found in the AgP nor in the TRT, the ĪŚGP or the Pheṭkariṇītantra (PheṭK). This can be said of the following term also.
samasta the nāman is uttered or written before the mantra, this being done twice: B A B A. Samasta meaning ‘put together’, ‘united’, the term could apply to any other combination as well. Samasta like grasta seems proper to the NT.
vidarbhitam ‘first the name, then the mantra is vidarbha’, says Kṣemarāja, and he quotes: abhidheyaṃ bhavet pūrvaṃ tato mantra sakr̥d bhavet / vidarbhitam. The Pheṭ K 3.148–149 defines it differently as alternating the letters of the mantra and those of the abhidheya, the pattern being not B A, but b1 a2 b2 a3 b3, etc. The AgP 138.7 says that one should alternate two syllables of the mantra with one of the sādhya (mantrākṣaraṃ dvayam. likhyā sādhyākṣaram punaḥ): a1 a2 b1 a3 a4 b2 a5, and so forth.
The VMT 2.33–34 (or NṢA 2.34–35) prescribes this practice for an amulet where the letters of the mantra and sādhya are inscribed in a śrīcakra, a practice described differently by Śivānanda in his R̥juvimarśinī (as: a1 b1 a2 b2 a3 b3 etc.) and by Vidyānanda in the Artharatnāvalī (a1 a2 b1 a3 a4 b2, etc.). The same text (śl. 21–22) prescribes a nyāsa of the mūlamantra on the image of the goddess entwined with the aṇkuśabīja (KROṂ).9 There are thus different forms of vidarbha in japa or in writing.10 Whatever its form, however, vidarbha always consists of associating sādhya and mantra while cutting apart somehow either these two elements, or the whole formula, or their constituting syllables.
But why is this called vidarbha? In this technical mantric sense, the term seems not to be found in Sanskrit dictionaries or encyclopaedias. Edgerton’s Buddhist Hybrid Sanskrit Dictionary, however, defines vidarbhayati (with vidarbhana and pravidarbha) as ‘intertwines (letters), i.e. writes them between letters of a mantra’, and it refers to several passages of the Sādhanamālā. As I have shown elsewhere,11 the term can be explained by reference to darbha, the grass of that name, which has a cutting edge. The Śabdakalpadruma says: ‘darbhaḥ – dr̥ṇati vidārayati, dradalibhyāṃ bhaḥ’, Unādisūtra 3.151, ‘iti bhas’, that is, it refers darbha, in the sense of the verbal root DR̥, ‘to cut’, and since Unā 3.151 says that after DR̥ and DAL one must have bha, this results in darbha. The prefix vi would then add the notion of separation, mantra and sādhya being cut off and separated. darbha can also be linked with the root DR̥BH, to string or tie together, the practice of vidarbha being then understood as first separating (vi), then stringing together mantra and sādhya.
The following six mantric practices – together with grasta and samasta – are, I believe, found in the NT only.
ākrantam madhyasthasya nāmno mantro yadi veṣṭanayā nyāsta ākrāntam, says Kṣemarāja: ‘when the mantra is placed so as to surround the name which is in the centre, [this is called] seized or invaded’. This looks like a variant of what NT calls grasta and AgP saṃpuṭīkaraṇa. (Note that the past participle ākrānta applies to the nāman, not to the mantra.)
ādyanta mantrād anantaraṃ nāma tatas trimantra ity ādyantam, ‘the name after the mantra, then thrice the mantra, this is “before and after’’’. The sequence is thus A B A A A .The term is self-explanatory.
garbhastha madhyasthasya mantrasya caturdikkaṃ sādhyanāmanyāso garbhasthatvam, says Kṣemarāja. This is in fact AgP’s definition of saṃpuṭa, except that the Uddyota on NT mentions the four directions of space instead of the four sides of the written mantra. It is also the reversal of what is defined as grasta. Like the AgP’s saṃpuṭa, this can be done only in writing.
sarvatovr̥ta mantrasya ādyantayoḥ sādhyanāmaniveśaḥ sarvatovr̥tam, ‘the placing of the sādhya or name before and after the mantra is [having it] completely surrounded’. This is the reversal of what is called saṃpuṭa by the NT. The pattern is B A B.
yuktividarbha the text quoted by Kṣemarāja defines this as placing the abhidheya before the mantra and three times after it: B A B B B. It is the inversion of ādyanta. The reason why this practice is so called is not clear.
vidarbhagrathita it is repeating three times the mantra after the nāman: nāmnaḥ paścāt trimantranāsyaḥ: B A A A. The reason why it is so called is also unclear since neither knotting (grath) nor cutting (vidarbha) occur.
This list of eleven mantric practices of the NT is an unusual one both in that it is longer than the usual one of six12 linked each with one of the ṣaṭkarmāṇi (which are not mentioned here), and in that it does not include three of these more usual ones, namely yoga, rodha and pallava, about which a few words must now be said.
yoga is defined in the TRT, PheṭK and ĪŚGP as placing or uttering the sādhya before the mantra: mantrādau nāmasaṃsthānaṃ yoga ityabhidhīyate (PheṭK 3.142). The AgP’s technique is different: adau mantra tataḥ sādhyomadhye sādhyaḥ punar manuḥ. We have thus either B A or A B B A. In both cases there occurs a conjunction or union (yoga) of the mantra and of whatever word or formula is associated with it, such union being deemed to increase the power of the mantra.
rodha consists (according to AgP 138.3–4 and PheṭK 2.144) in placing the mantra before, in the middle of, and after the sādhya – A B (A) B A: nāmnā ādyantamadhyeṣu mantraḥ syād rodhaḥ. rodha, from the verbal root RUDH, is probably understood as the surrounding, investing, or blockading of the sādhya by the mantra. It is mentioned in the ĪŚGP, but not in the TRT.
pallava consists, according to the TRT or PheṭK, in putting first the mantra, then the sādhya (A B). The sādhya appears thus as a shoot or extension (pallava), as it were, of the mantra. The AgP has a different description – it seems to add the mantra both before and after: ādau sādhyaṃ likhet pūrvam. cānte mantrasamanvitam. The pattern would then be A B A.
I have described the foregoing mantric practices not so much because I find them intrinsically interesting, but because I believe they contribute to throw some light on the nature of mantras or of mantric utterance or, more generally, on the Tantric conceptions of the relationship between sound or word and image or writing.
These practices, indeed, concern mantras, that is, oral/aural formulas transmitted by word of mouth from master to disciple and put into practice through utterance or recitation (uccāra, japa) only. Mantras can also be enunciated silently (tūṣṇīm); such silent utterance (where the utterer ‘restrains his voice’, vācaṃ yacchati), and still more the inner and purely mental one (mānasa), being in fact considered as the highest and most meritorious ones: the mental utterance is the highest because the Absolute, the supreme godhead is silent. Silence is higher than the highest plane of the word (vāc) because it is its primal Source. The mantrin, in such case, evokes (or perhaps fuses with) a transcendent plane of vāc which, imperceptible, is present as its animating power in all uttered mantras.
But how is it, then, that this power can also be evoked, made present, when a mantra is written? First, the writing or the placing of a written formula is not mere writing but a ritual performance which can be fruitful insofar only as it uses, as it is permeated by, the power of vāc, of the Word which inhabits the, by nature oral, mantra thus written down. The written formula is thus (deemed to be) infused with the energy of the oral formula. This explains how and why written mantric practices can be prescribed in a work such as the NT whose mūlamantra, the mr̥tyujit OṂ JUṂ SAḤ, is in principle effective only when uttered. It explains too why such practices are endorsed by Kṣemarāja, that is, by one of the main exponents of the non-dualistic Kaśmirian brand of Śaivism, whose concept of mantra is directly issued from that of the Śivasūtra (ŚS 2.1: cittaṃ mantraḥ), the main commentary on which is Kṣemarāja’s Vimarśinī, a work where he states that the supreme Word (parā vāc) or Consciousness (saṃvit) is the true nature of mantras.
Another explanation, and one that fits with the doctrine of the Trika-based pratyabhijñā, is that mantras exist and function on different levels. Abhinavagupta thus said that by nature they ‘consist of discursive thought as well as pure consciousness’ (vikalpasaṃvidmayaḥ).13 In their functioning, too, they may well appear differently. They may be felt and used as pure consciousness (saṃvit) by the liberation seeker (the mumukṣu) – and still more by the liberated in life (jīvanmukta), and be also (even simultaneously) used audibly on the plane of discursive thought and of the gross form of the word (vaikharī); they can even be written by a mantrin, a sādhaka, performing kāmya rites (which are in fact those which are the main concern of the NT).
These uses may differ widely; they are, however, grounded in the same, fundamentally oral, conception of mantras. Just as magic is never entirely separate from religion, of whose concepts it is merely a profane use, so mantras, even in their lowest, magical uses, cannot be separated from their highest, purely oral ones. In fact, such lower uses of mantras are effective only insofar as all mantras are in essence the supreme Word, pure divine energy.14
Thus a text such as the NT may expound some written (and in some respects magical) uses of mantras whilst still upholding the essentially spiritual nature of these ritual formulas. Let us take also, for instance, the case of mantras or bījas placed in ritual diagrams (cakra, yantra or maṇḋala). These may be placed in the diagram by the rite of nyāsa, which is at the same time a bodily and oral and mental action.15 They can also be inscribed, written, there. Some texts assert explicitly the power, value or meaning of these written mantras. A typical instance is that of the kāmakalā as described in the Yoginīhr̥daya (YH), one of the basic texts of the Śrīvidyā tradition.16 A passage of this work (YH 2.21, Dīpikā, pp. 130–131) describes – rather cryptically – the so-called kāmakalācakra as made up of two intersecting triangles with the bīja ĪṂ inscribed in the central part of the cakra.17 This bīja is to be meditated both as the vācaka of the goddess and as the kunḋalinī (coiled like the letter Ī in nāgarī script) in her ascent from mūlādhāra to brahmarandhra: the ritual process involved here is both visual and mental, resting as it does on the meditation of a ‘written sound’. The Ī of the kāmakalā can also be meditated as made up of three dots, that is, as it is written in the Brāhmī script: this use of an ancient script is perhaps a case of deliberate archaism. It can be taken, too, as suggesting an (unlikely!) ancient origin of this practice.
In the ritual of nyāsa, too, the mental visualisation of the written shape of the bīja that is being placed is sometimes prescribed, or a written bīja or mantra is to be assigned somewhere. In the Mantraparibhāṣā (chapter 293) of the AgP, for instance, a lipinyāsa is prescribed, for which the śaktis of Rudra (that is, the fifty akṣaras from A to KṢA) are to be written, then ritually placed on the body of the sādhaka (vilikhet … vinyāset, śl. 147).18 More important, perhaps as regards the role of written letters or mantras (it is well known that any akṣara can be considered as a bījamantra, from AṂ to AḤ and to KṢAṂ) is the conception of Lipidevī, the (written) Alphabet Goddess19 mentioned, for instance, in AgP 293, or that of the ‘Tree of letters’, lipitaru, of the ŚT 7.08–14, under which the Goddess is seated. Among alphabet deities is also the goddess Mālinī of the Trika (and of some Śaiva Kaśmirian systems) which consists in a particular order of the Sanskrit alphabet, going from ṆA to PHA (the ṇādiphāntakrama).20 One finds it written in a triangular ritual diagram, a [mālinī]gahvara shown in the Tantrasadbhāvatantra. But, according to some texts, it can also be written in a gupta script, the letters being correlated with body parts. Since a mantra is always fundamentally an oral principle, when this representation of the Mālinī mantra-deity is ritually used, that is, mentally worshipped, the practice would combine oral, visual and mental elements.21 The written form of the letter A in the devanāgarī script is also sometimes the basis of a meditation, the different parts of the letter being meditated as embodying the movements of the supreme energy, kunḋalinī.22
We must also refer here to the ritual procedures used for the uddhāra, or for the research of the aṃśa of mantras, where letters of the Sanskrit alphabet are, for the mantroddhāra, written (often in a particular secret order) on a diagram, called prastāra or ghavara,23 or, for the research of the aṃśa, on cakras of various shapes, all procedures that are used in all Tantric traditions, whether Śaiva or Vaiṣṇava.24 Incidentally, we may remark that the prescription in tantras of such written procedures shows that these texts, though supposedly revealed, were originally written. We must also not forget the numerous uses of written mantras on amulets (kavaca) which, extremely frequent in modern times and nowadays, are in fact a very ancient use.
Coming back to the oral nature of mantras and to the practices we have seen of saṃpuṭa, etc., one must say that, to a large extent, these notions and practices can be traced back to Vedic origins. The Vedas were composed several centuries before writing began to be used in India. This explains perhaps the bias against the writing of revealed texts that was to appear later on, when writing began to be used (seventh to eighth century BC?). Writing was thus confined to practical, mundane uses and excluded from religious/ritual ones. However, since India was from early times a highly literate society, producing one of, if not the largest, literature in the world, there inevitably occurred interactions between the written and the unwritten traditions, a situation which resulted – to take a particular instance – in the (apparent) discrepancy between the in principle oral–aural nature of mantras and their written uses.25 We may note, too, that such mantric procedures as vidarbha and so forth, where the constituting parts of a mantra are separated, interpolated, duplicated, inverted, etc., have as their forerunners such Vedic usages as viharaṇ a and vikr̥ti (intertwining or transposing). In the same way, the use of ‘meaningless’ syllables or bījas added to, or interpolated in the sequence of a mantra can be traced back to the vyāhr̥tis of the Veda or to the syllables chanted by the Udgāta priest of the Sāmaveda.
Developed in a Sanskrit tradition going back to the Vedas, the Tantric mantraśāstra could not but participate in the main traits of some of the linguistic aspects of that tradition (to which, from a certain time onward, it contributed largely). Seen in this light, that is, as part of a stable and continuous but ever-changing culture, such seemingly out of the norm uses as those we have seen here appear as merely an aspect of a culture where the oral and the written have interacted during more than two millennia, and thus can never be entirely separated.26