As we have seen, Tantric mantras are the phonic form of deities – deities as mantras or mantras as deities. They are in this respect divine entities existing eternally as aspects of the supreme Word (vāc), or at least existing on a pure, perfect level of speech, free of all imperfections. But they are also conceived as limited entities, either as ‘souls’ (aṇu), of different degrees of perfection, or as spiritual powers existing on different levels of the cosmos. Finally, in ritual practice, they are empirically existing forms of phonic utterance: they are quoted in ritual manuals, ritually, often audibly, uttered and heard, sometimes even written, partaking thus of the imperfection of the created world.2 This explains why they can suffer from flaws or defects, doṣas, of which they must be freed so as to recover their pristine perfection and be able to be perfectly effective (or harmless for their user). Hence the existence of such ritual practices as mantraśuddhi and mantrasaṃskāra.
We may note in passing that, whether pristine or retrieved, the divine, transcendent nature of mantras is always present in practice on the empirical, human plane, since mantras are to be used in this world as empirically uttered formulas of a divine nature. They would be powerless if they were not divine: the immaterial absolute essence and the physically uttered sound are thus brought together in action: the nature of mantras is contradictory, paradoxical, even.3
A number of Tantric works deal with mantradoṣas. I shall refer here to a few of them only, mostly from the Śaiva traditions. The longest list of doṣas I know of is that of the Kulārṇavatantra (KT), one of the main Kaula tantras (dated c. eleventh to fourteenth century). The KT 15.65–69 quotes sixty doṣas. The eleventh-century Śāradātilaka (ŚT), of Lakṣmaṇadeśika, one of the most important digests of mantraśāstra, enumerates (2.64–110) then describes fifty such defects, its descriptions being expatiated upon in Rāghavabhaṭṭa’s commentary, where other works are quoted – notably the Piṅgalāmata, an important ancient Yāmalatantra, and the Mantramuktāvali, a voluminous, more recent, work.4 The Tantrarājatantra (TRT), a Śrīvidyā work (1.75–48 or 1.73–815), mentions twenty-five doṣas, the Kaulāvalīnirṇaya (7.10–12) a dozen. The earliest (eighth century) of these texts, the Netratantra (NT), in a passage on the efficacy of the netramantra (8.59–63), quotes nine doṣas as instances of such defects as can be cured by a particular use of that mantra. Many other works mention mantradoṣas without expatiating on the subject: the Gandharvatantra (GT), chapter 9, for instance, the Somaśambhupaddhati (SP), the Īśanaśivaguudevapaddhati (ĪŚGP), Appayadīkṣita’s Śivārcanacandrikā, and so forth.
The lists of mantradoṣas I have seen do not appear to enumerate them according to any logical order, nor do these listings follow the same order: they look like random collections. If one compares the two longest lists, those of the KT (sixty doṣas) and of the ŚT (fifty), only twenty doṣas are quoted in both texts and ten only are among the twenty-five ones of the TRT. This results in a total of nearly a hundred: a rather large number. The doṣas enumerated are not always clearly described. The ŚT (2.72–110) describes all the doṣas (with Rāghavabhaṭṭa’s added comments). Kṣemarāja’s Uddyota on NT 8.59–63) gives some explanations and so does, too, briefly, Prāṇamañjarī on TRT. If the names of some doṣas are self-explanatory, others are not, having sometimes apparently little or nothing to do with their actual nature.
The doṣas enumerated in the ŚT 2.64–110 are all described as being of a purely linguistic nature, that is, as resulting either from the presence of particular letters or bījas or from the fact that these letters or bījas occur in certain portions of the mantra, or else because the mantra has a particular, inauspicious, number of syllables. Thus, if we look at this list, we find that a mantra of one syllable (akṣara) is niraṃśa;6 one of two letters is sattvahīna, devoid of essence or reality; a mantra of four varṇas is kekara (squint-eyed); one with two bhūbījas (including twice the bīja LAṂ of the earth, that is) is ruddha, ‘obstructed’, bringing neither bhukti nor mukti. A mantra of three letters and without haṃsa (HAṂ) is said to be in deep sleep (suṣupta) and therefore ineffective;7 and so forth. One may well ask why defects considered as resulting from purely phonetic traits should often be called by names evoking human bodily or mental traits (squint-eyed, sluggish, frightened, and so forth). This may be nothing else than a picturesque way of expressing the alleged inability of such mantras to effect properly what they are supposed to do. The verbal identification of mantras with human beings, however, is perhaps not entirely whimsical: it may reflect somehow the conception of mantras as particular types of beings, as individual souls (aṇu), which is indeed how they are conceived of in the Shaivasiddhānta, which is also consonant with the Śaiva system of the seven categories of experiencers (pramātr̥), one of which is that of mantras.
The fact that negative effects on the efficacy of mantras should be attached to certain traits of their phonetic pattern should not surprise us, for mantras are utterances whose power is grounded in their phonetic substance, in their form, which is therefore essential. As is well known, the phonetic content and pattern of mantras is far more important that their meaning – supposing they have a meaning, which is often not the case. This has been mentioned before.8 But mantras are also never spontaneous, improvised utterances. They are conventional rule-governed utterances transmitted by tradition. They are the essential, phonic, form of deities (expounded, for the adept in dhyānaślokas), or words of power transmitted by the ritual process of uddhāra, their phonetic content being always carefully described in ritual texts so as to ensure their faultless transmission, their original perfection. How then can such imperfections as those listed in the ŚT or other such works happen? Also, why should some syllable or word order be deemed faulty when such a large number of very different mantras exist? To the outside observer, this looks very arbitrary – which it probably largely is. A possible explanation could be that patterns listed as faulty are phonetic patterns that should not result from the rules of mantroddhāra laid down by a particular tradition and which would appear if these rules were not followed. It is also possible that these lists are in fact meant to exclude phonic patterns proper to other traditions. The lists of doṣas may also prove useful in such mantric practices as vidarbha, pallava, grathana,9 etc. where the transpositions, interpolations, etc. of the syllables or words of a mantra can result in one of the forbidden patterns. The question, however, remains as to why a number of patterns that do not look prima facie abnormal should be set apart, being considered as faulty or inefficacious.
There are, however, other doṣas which (more reasonably, we would be tempted to say) result from the way in which a mantra is transmitted, uttered, or made use of. Such is the case of several of the twenty-five doṣas listed in the TRT (1.75–80). Thus a mantra which has been heard by a third person is ‘burnt’ (dagdha), it is trasta (frightened) when not recited the prescribed number of times, garvita (haughty) when not transmitted according to the rules, chinna or khaṇḋita (cut or torn) when incomplete, hīnavīrya (without force) when given by a master who does not possess the adhikāra, the authority to transmit it, and so forth. The TRT also mentions the case of mantras not used at the proper time, not recited as they should be (they are rugna, destroyed) when not uttered clearly, kliṣṭa (in bad condition) when recited too slowly, avamanita (despised) when recited without paying attention or without faith, etc.). The imperfections in those cases lie more with the user than with the mantra, but they are nevertheless classified as mantradoṣas, defects to be carefully avoided since mantras are rule-governed utterances and are therefore effective only insofar as these rules are carefully respected.
Mantradoṣas can also result from the fact that they are uttered in cases when they are not to be mentally formulated. The mantra then issues from the mouth of the mantrin, hence, for some texts, the existence of doṣas due to the contact of the mantra with the organs of speech. One finds this sort of doṣa in various texts, in the Śivārcaṇacandrikā10 for instance, or in a passage of Aghoraśiva’s commentary on ŚP 1.3,43 (vol. 1, p. 145) where he quotes an unidentified text which runs: dantādhāroṣṭṭasaṃcare mantrasyāśaucanirhanāt/nādāntoccāreṇaivamantraśuddhir udāhr̥tā, mentioning thus both cause of impurity and its cure. There is also the TRT 1.80 which says that mantras suffer from a doṣa when the teeth of the mantrin are not clean …
Among the doṣas quoted in various texts there is also the case where a mantra is said to be asleep (supta, susupta, prasupta, svāpaga). According to the ŚT (2.84), for instance, a mantra is susupta when it has three varṇas and no haṃsa, this being apparently a defective phonetic pattern. For other texts, a supta or svāpaga mantra is one used when it is ‘asleep’ or not ‘awakened’ at the proper time. This we find, for instance, in the TRT 1.77. Other texts, however, do not consider the supta (or svāpaga, etc.) state of the mantra as a doṣa, but as a state in which a nondefective mantra happens to be during certain periods of time, as are human beings, or because of a particular movement of prāṇa. Supta in this case, is taken as opposed to prabuddha:11 it is not a case of doṣa.
The KT 15.57–63, as mentioned previously, enumerates sixty dosas. It begins, however, by mentioning some preliminary conditions for a mantra to be effective. The first is that it should not suffer from the impurity of birth at its beginning or of death at its end, which would occur, it seems, if it were not to begin and end with OṂ. The ignorance of its aim (artha) or of its ‘consciousness’ (caitanya) – by which is meant, we may assume, the lack of consciousness of the mantrin – would rob the mantra of all efficiency; as would also the ignorance of the yonimudrā. Some other preliminary conditions are also mentioned. The same defect, with the same sort of śuddhi, is also to be found in the GT. The doṣas listed there (65–69) are only partly the same as those of the ŚT, and include such traits as ripu usually considered not as a mantradoṣa but as a bad aṃśa.
We may note, finally, on the problem of the faultless state of mantras, that though they must be pure to be effective, they must also, according to some, not avoid, but on the contrary bear necessarily a particular form of impurity, the aiśvaryamala, the impurity connected with power, which exists even in the purest beings insofar as they act. In the system of the seven pramātr̥, this impurity as well as the ‘impurity of authority’ (adhikāramala), permitting an activity, is considered to be present in mantras since they are considered as a category of conscious subjects (pramātr̥), and this gives them the capacity to act in this impure world.
The general rule is, however, that mantras must be without defects or impurity so as to be effective. We shall therefore see now how this necessary purity is given or restored to them.
A mantra is to be made free of all defects so as not only to be more effective but also to protect its user from the dire effects which could result from a flawed formula. This is done through a ritual practice usually called mantraśuddhi or mantraśodhana (purification of the mantra), which is mentioned or described in several texts. Of these the older and more authoritative ones (among those I have access to) are the NT (with Kṣemarāja’s commentary), which enumerates nine saṃskāras, and the ŚT, listing ten. Kṣemarāja quotes the Ucchuṣmatantra12 (recording seven saṃskāras only); whilst Rāghavabhaṭṭa’s commentary on the ŚT cites abundantly the ancient yāmalatantra Piṅgalamata. Mantrasaṃskāras are also enumerated in more recent works such as Kr̥ṣṇānanda’s Tantrasāra (sixteenth century), which is quoted in the TBhS (seventeenth century). The practice has not disappeared: the texts quoted here are, I believe, still made use of in India by practitioners of mantras.
We will limit ourselves to quoting and comparing the mantrasaṃskārās of the NT and the ŚT which, interestingly, are practically the same in name and aim whilst having to be performed in completely different ways. The nine NT (18.6–8) saṃskāras are: dīpana, bodhana, tāḋana, abhiṣecana, vimalīkaraṇa, indhananiveśana, saṃtarpaṇa, guptibhāva and āpyāyana, that is illuminating, awakening, thrumming, sprinkling, purifying, using as fuel, satiating, hiding and strengthening. ‘Thanks to these nine saṃskāras’, says the NT (18.8), ‘one who masters the science of the mantras (mantravāda) will be master of the efficiency of the mantras.’ According to Ksemarāja’s commentary (and to the Ucchuṣmatantra), most of these saṃskāras are simply linguistic devices, consisting as they do in adding before the mantra and/or at its end such elements as OṂ, or as one of the jātis (the ritual exclamations at the end of mantras), or in interpolating them in the mantra. Dīpana is thus done by adding two OṂ, bodhana by adding namaḥ, saṃtarpana is done with two LAṂ (the bija of water). Tādana, however, as its name implies, consists in thrumming13 the letters of the mantra (which is thus to be written) whilst uttering twice the jāti phaṭ. Guptibhāva is more complex since it consists in ‘enclosing’ (saṃpuṭīkaraṇa14) the mantra between two Netramantras and then reciting it ten thousand times. saṃpuṭīkaraṇa is also used in indhananiveśana, ‘using as fuel’, the mantra being able in that case to burn all fetters or poisons afflicting a person.
The ŚT 2.112–123 quotes, and then describes, ten mantrasaṃskāras which include eight of those of the NT list excepting indhananiveśana, to which are added at the outset janana (birth) and jīvana (vivifying), which are not in the NT. These two first saṃkāras are, in fact, preliminary rites since janana is simply the extraction of the letters, the akṣaras, of the mantra from among the fifty phonemes, the mātr̥kās,15 of the Sanskrit alphabet ritually written on a lotus-shaped maṇḋala, which is nothing else than a variant of the rite of mantroddhāra described in chapter 2 of this book. It is the preliminary rite necessary for all mantras to exist as words of power. It is here quite logically followed by jīvana, since once ritually ‘born’, the mantra, to be effective, is to be animated by this rite, which, according to the ŚT, is done by reciting (japa) the mantra a hundred times while interpolating the praṇava OṂ between each of its letters: an oral rite, therefore. The eight other saṃskāras, however, though being the same in name as those prescribed in the NT, differ from these in that they are not purely linguistic, oral, performances, but rites performed on a written mantra. Thus bodhana is done by hitting the written mantra with laurel flowers while uttering the fire-bīja RAṂ. Abhiṣeka is done by touching the mantra written on birch-bark with aśvattha leaves (this to be done 108 times according to the Piṅgalamata, or as many times as there are letters of the mantra according to ‘another tantra’, both quoted in the commentary), and so forth for other saṃskāras. The purification, vimalīkaraṇa, is to be done, according to ṢT 2.119, by ‘burning’ the mantra with a fiery mantra (jyotirmantra), but it may also be done, according to the commentary, by a yogic practice, the raising of the mantrin’s kuṇḋalinī. For dīpana and guptībhava, mantras and japa are prescribed. We may be tempted to consider this variety of procedures – purely linguistic or ritual, written or oral, or yogic, as illogical, or even whimsical, which they perhaps are. The reason why a particular procedure is to be used for a particular saṃskār̥a surely does not appear logical. However, whether logical or whimsical, this variety can (ought?) be seen as reflecting the complex nature of mantras which are speech acts, thus oral, but are put into action by ritual; they are endowed with a divine power of which kuṇḋalinī is the bodily aspect; and in actual practice they can be written. The KT 15.71–72 enumerates the same ten saṃskāras: ‘As a sword rubbed on a whetstone gets sharpened, concludes the tantra, in the same way the mantras bloom and radiate thanks to the ten saṃskāras.’